[SIESTA-L] Convergence problem with slabs
Dear siesta users, I'm dealing with a convergence problem on a slab system (gamma-maghemite). Everything works fine for the bulk, I could optimized the atomic positions as well as the cell parameters. Also, I computed the DOS giving the expected one. But when I add a vacuum region I have a divergence on the scf. I checked several times the symmetry on my system and I could affirm that is correct. The scf started with a negative Energy but after some steps it becomes positive. I'm using a slab dipole correction. Also, I tried to increase de temperature (for the bulk I was working at 5 K) until 298 K. Moreover, I have to notice that I was using DM.MixingWeight 0.01, using the spinpolarized .true. and LDA+U. This problem could be related to the symmetry of my system? If not, what could be the problem? Could someone enlighten me? Thanks in advance Best Regards Rubén -- Aunque mi madre no sabía nada de ciencia, también ejerció sobre mí una gran influencia. Yo, un universo de átomos, un átomo en el universo. -- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)
[SIESTA-L] Convergence problem
Dear Siesta users I am doing transport calculation using transiesta package implemented in siesta 4.0 version. Generally I see that as we apply higher bias voltage between the electrodes, then it becomes difficult to converge. Can anyone please tell me some general methods how to converge the transiesta run( calculation) at higher bias voltage. Thanks and regards Rupesh Kumar Tiwari C/o: Prof. Gopalan Rajaraman Junior Research Fellow (CSIR) Department of Chemistry IIT Bombay Mumbai- 400076 -- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)
[SIESTA-L] Convergence failure
Hi all I was running a calculation for a molecular junction device in transiesta. In siesta calculation, it got converged but in transiesta calculation, it's not converging. Can anyone please help me asap. For your convenience, I am attaching the input file here. Should I change any parameter in the input? Please have a look in the input file also. Rupesh Kumar Tiwari C/o: Prof. Gopalan Rajaraman Junior Research Fellow (CSIR) Department of Chemistry IIT Bombay Mumbai- 400076 INPUT.fdf Description: application/vnd.fdf -- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)
Re: [SIESTA-L] Convergence problem
Hello Rupesh This is a trial and error situation So I’m other words if 0.5 V did not work try 0.501 0.503 V because I had similar issues and it worked when I changed a bit the voltage. Hope that works and Good luck ! El abed Sent from my iPhone On 6 Nov 2019, at 8:02 am, RUPESH TIWARI mailto:rupesh199...@gmail.com>> wrote: Hi Everyone, I am doing a transport calculation of a molecular junction device using the siesta software 4.0 version. There was no convergence problem until 0.4 V applied bias. But at 0.5 it's showing a convergence problem. To run the calculation at 0.5 V, I had copied the TSHS and TSDE file generated from transiesta run at 0.4 V. Can anyone please suggest me to solve the convergence problem. I am attaching the input file and some major output file here. -- Rupesh Kumar Tiwari C/o: Prof. Gopalan Rajaraman Junior Research Fellow (CSIR) Department of Chemistry IIT Bombay Mumbai- 400076 -- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/) -- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)
[SIESTA-L] Convergence with number of basis functions
Dear SIESTA Team, Recently I have performed computations for polycyclic aromatic hydrocarbons. I have tried different default basis sets. As a result, I have pretty good results fro SZP set - it turns out this basis set works much better than TZTP. That seems strange for me since I expected a convergence with the number of basis functions. Can you, please, give me a hint to understand why this may happen. Thank you in advance Best regards, Mike
[SIESTA-L] convergence with respect to k mesh
Dear Siesta users, I have been trying to optimize a 32 atom 2D supercell of Germanium Carbide. For that I am trying to test the convergence of my system with respect to the k-mesh value. Following according to the siesta hands on tutorial, I have run optimization of the system with k-mesh value starting from 4x4x1 up to 30x30x1. Now even though the lattice parameter has obtained a constant value from 20x20x1 k-mesh, the free energy values are fluctuating very much, even near 30x30x1 k-mesh. I am attaching a data file listing the k-mesh used and the corresponding free energies, lattice parameters and pressures. #kfree_energy(eV) a(Ang)press(kBar) #--- 4-4404.68656913.0719191.58775270 5-4404.69200913.0719261.57574542 6-4404.68364413.0719201.58159257 7-4404.69038313.0719181.58333081 8-4404.69046813.0719201.58158769 9-4404.65695013.0719181.58728229 10-4404.68947913.0719201.58758950 11-4404.65124213.0719231.55663793 12-4404.65243013.0719241.55242216 13-4404.65278513.0719241.55406592 14-4404.69045313.0719171.58822447 15-4404.69032613.0719201.58736858 16-4404.64894613.0719241.58553524 17-4404.69040913.0719191.58581311 18-4404.65669513.0719241.51914825 19-4404.64795313.0719231.56108155 20-4404.65682313.0719241.51265983 21-4404.66229813.0719241.56920158 22-4404.66009913.0719251.57130732 23-4404.65439413.0719241.52465516 24-4404.66623513.0719241.56528804 25-4404.65527313.0719241.51864296 26-4404.65496913.0719241.57698239 27-4404.65402613.0719241.51899044 28-4404.66068613.0719241.57076477 29-4404.64796513.0719241.58495969 30-4404.65091813.0719241.58155660 I have been optimizing the above structure with spin-polarization turned ON. Is it the reason why the free energy values are not reaching a converged value? Should I carry out the convergence tests with spin-polarization option turned off and turn it on only while calculating bands, DOS and other properties? Please help me in this regard. Thank YOu in advance. Sunetra Das
[SIESTA-L] Convergence criteria of mesh cut off 0.01 eV
Dear all During mesh cutoff convergence study for graphene I am using 0.01 eV as a convergence criteria. Is that ok? please guide me. Thank u Regards Riya
Re: [SIESTA-L] Convergence tests & transmission
Dear Rajan, thanks for this information. I will in future work with this procedure. Some more things that confuse me are the following: 1. When I re-run the same setup, I get deviations up to 15% in the transmission. What could be the reason for this? 2. I did calculations of the same setup with transiesta 3.2-pl-5 and 4.0b-485 and also got quite different characteristics in the transmission spectrum. Same question: what could be the reason for this? I didn't notice any changed defaults. Apart from this, are there reliable ways to tell if my compilation is stable? Thanks and best regards, JH Rajan Singh schrieb am 2016-03-14: > Hii > You should do following steps: > 1. Decide your structure > 2. Run structure with different k pts and plot energy vs k pt > See after some k pt energy values nearly same > 3. Do same with mesh cuttoff > 4. Relax structure with CG steps > 5. Then electrode calculation > And finally tranmission calculations > Regards > Rajan Singh > Sent from my Sony Xperia™ smartphone > j_hek...@uni-muenster.de wrote: > >Dear users, > > > >I am pretty new to Siesta/TranSiesta and I'm currently struggling with the > >following topic: I want to compute equilibrium transmission of > >Poly(3-Methylthiophene) with different chain lengths. My current (ideally > >periodic) _test_ system consists of 100 atoms (a 10mer) of which 2x40 atoms > >build the electrodes. At the moment I'm doing lots of tests with different > >parameters like k vectors, box sizes, MeshCutoff and others. The finally > >important quantity is the transmission. > >My problem is now that I'm not really seeing a "pattern": I find it hard to > >understand which physical quantity indicates a good or bad convergence > >w.r.t. the transmission. I'm observing energies and of course the > >transmission function. What confuses me most is that the transmission > >function changes shape very drastically for changes like e.g. increasing or > >decreasing the kx,ky vectors by 1. Sometimes transmission values de- or > >increase by up to 6 magnitues, in combination with a totally different > >characteristic. > >Also, the same calculation yields deviations of up to 15% at for example > >T(E=Ef) when repeated. Could this be connected to a numerical instability > >due to a compilation problem or a "defective" math library? > >Literature indicates a transmission T(E=Ef) ~ 0-1 G0. Most of my results > >range between 0.17 and 0.01 G0 (again: with very different transmission > >function shapes). > >I'm afraid of having an understanding problem somewhere or missing out > >something important. > > > >I hope that I can get some advice. > > > >Thanks and best regards, > >JH
RE: [SIESTA-L] Convergence tests & transmission
Hii You should do following steps: 1. Decide your structure 2. Run structure with different k pts and plot energy vs k pt See after some k pt energy values nearly same 3. Do same with mesh cuttoff 4. Relax structure with CG steps 5. Then electrode calculation And finally tranmission calculations Regards Rajan Singh Sent from my Sony Xperia™ smartphone j_hek...@uni-muenster.de wrote: >Dear users, > >I am pretty new to Siesta/TranSiesta and I'm currently struggling with the >following topic: I want to compute equilibrium transmission of >Poly(3-Methylthiophene) with different chain lengths. My current (ideally >periodic) _test_ system consists of 100 atoms (a 10mer) of which 2x40 atoms >build the electrodes. At the moment I'm doing lots of tests with different >parameters like k vectors, box sizes, MeshCutoff and others. The finally >important quantity is the transmission. >My problem is now that I'm not really seeing a "pattern": I find it hard to >understand which physical quantity indicates a good or bad convergence w.r.t. >the transmission. I'm observing energies and of course the transmission >function. What confuses me most is that the transmission function changes >shape very drastically for changes like e.g. increasing or decreasing the >kx,ky vectors by 1. Sometimes transmission values de- or increase by up to 6 >magnitues, in combination with a totally different characteristic. >Also, the same calculation yields deviations of up to 15% at for example >T(E=Ef) when repeated. Could this be connected to a numerical instability due >to a compilation problem or a "defective" math library? >Literature indicates a transmission T(E=Ef) ~ 0-1 G0. Most of my results range >between 0.17 and 0.01 G0 (again: with very different transmission function >shapes). >I'm afraid of having an understanding problem somewhere or missing out >something important. > >I hope that I can get some advice. > >Thanks and best regards, >JH
[SIESTA-L] Convergence tests & transmission
Dear users, I am pretty new to Siesta/TranSiesta and I'm currently struggling with the following topic: I want to compute equilibrium transmission of Poly(3-Methylthiophene) with different chain lengths. My current (ideally periodic) _test_ system consists of 100 atoms (a 10mer) of which 2x40 atoms build the electrodes. At the moment I'm doing lots of tests with different parameters like k vectors, box sizes, MeshCutoff and others. The finally important quantity is the transmission. My problem is now that I'm not really seeing a "pattern": I find it hard to understand which physical quantity indicates a good or bad convergence w.r.t. the transmission. I'm observing energies and of course the transmission function. What confuses me most is that the transmission function changes shape very drastically for changes like e.g. increasing or decreasing the kx,ky vectors by 1. Sometimes transmission values de- or increase by up to 6 magnitues, in combination with a totally different characteristic. Also, the same calculation yields deviations of up to 15% at for example T(E=Ef) when repeated. Could this be connected to a numerical instability due to a compilation problem or a "defective" math library? Literature indicates a transmission T(E=Ef) ~ 0-1 G0. Most of my results range between 0.17 and 0.01 G0 (again: with very different transmission function shapes). I'm afraid of having an understanding problem somewhere or missing out something important. I hope that I can get some advice. Thanks and best regards, JH
Re: [SIESTA-L] convergence test in the level broadening and number of grid points to obtain the non-equilibrium part of density matrix
Dear Dr. Papior, Thanks so much for your reply. Must I do the energy convergence test? Thanks in advance, Best regards. Nadia Salami On Fri, Jul 3, 2015 at 11:58 AM, Nick Papior nickpap...@gmail.com wrote: As you said, convergence tests is the way to go. 2015-07-03 9:06 GMT+02:00 Nadia Salami nadiasala...@gmail.com: Dear Dr. Papior, Thank you very much for your reply. I have applied these options, TS.BiasContour.Eta 0.0001 Ry TS.BiasContour.NumPoints 10 But I don't know how to obtain the optimized value. Your Guidance will be highly appreciated. Thanks a lot. Best, Nadia Salami On Fri, Jul 3, 2015 at 10:55 AM, Nick Papior nickpap...@gmail.com wrote: Please read the documentation for transiesta. In the currently released versions these keywords should peak your interest: TS.BiasContour.Eta TS.BiasContour.NumPoints 2015-07-03 5:33 GMT+02:00 Nadia Salami nadiasala...@gmail.com: Dear Transiesta users, According to DOI: 10.1103/PhysRevB.65.165401, I know that the non-equilibrium part of density matrix must be evaluated for a finite level broadening and on a fine grid. I need to do convergence test in the level broadening and number of grid points. But I don't know how to do the convergence test? Any comments and guidance will be appreciated. Best, Nadia Salami -- Kind regards Nick -- Kind regards Nick
Re: [SIESTA-L] convergence test in the level broadening and number of grid points to obtain the non-equilibrium part of density matrix
As you said, convergence tests is the way to go. 2015-07-03 9:06 GMT+02:00 Nadia Salami nadiasala...@gmail.com: Dear Dr. Papior, Thank you very much for your reply. I have applied these options, TS.BiasContour.Eta 0.0001 Ry TS.BiasContour.NumPoints 10 But I don't know how to obtain the optimized value. Your Guidance will be highly appreciated. Thanks a lot. Best, Nadia Salami On Fri, Jul 3, 2015 at 10:55 AM, Nick Papior nickpap...@gmail.com wrote: Please read the documentation for transiesta. In the currently released versions these keywords should peak your interest: TS.BiasContour.Eta TS.BiasContour.NumPoints 2015-07-03 5:33 GMT+02:00 Nadia Salami nadiasala...@gmail.com: Dear Transiesta users, According to DOI: 10.1103/PhysRevB.65.165401, I know that the non-equilibrium part of density matrix must be evaluated for a finite level broadening and on a fine grid. I need to do convergence test in the level broadening and number of grid points. But I don't know how to do the convergence test? Any comments and guidance will be appreciated. Best, Nadia Salami -- Kind regards Nick -- Kind regards Nick
Re: [SIESTA-L] convergence test in the level broadening and number of grid points to obtain the non-equilibrium part of density matrix
Dear Dr. Papior, Thank you very much for your reply. I have applied these options, TS.BiasContour.Eta 0.0001 Ry TS.BiasContour.NumPoints 10 But I don't know how to obtain the optimized value. Your Guidance will be highly appreciated. Thanks a lot. Best, Nadia Salami On Fri, Jul 3, 2015 at 10:55 AM, Nick Papior nickpap...@gmail.com wrote: Please read the documentation for transiesta. In the currently released versions these keywords should peak your interest: TS.BiasContour.Eta TS.BiasContour.NumPoints 2015-07-03 5:33 GMT+02:00 Nadia Salami nadiasala...@gmail.com: Dear Transiesta users, According to DOI: 10.1103/PhysRevB.65.165401, I know that the non-equilibrium part of density matrix must be evaluated for a finite level broadening and on a fine grid. I need to do convergence test in the level broadening and number of grid points. But I don't know how to do the convergence test? Any comments and guidance will be appreciated. Best, Nadia Salami -- Kind regards Nick
Re: [SIESTA-L] convergence test in the level broadening and number of grid points to obtain the non-equilibrium part of density matrix
Please read the documentation for transiesta. In the currently released versions these keywords should peak your interest: TS.BiasContour.Eta TS.BiasContour.NumPoints 2015-07-03 5:33 GMT+02:00 Nadia Salami nadiasala...@gmail.com: Dear Transiesta users, According to DOI: 10.1103/PhysRevB.65.165401, I know that the non-equilibrium part of density matrix must be evaluated for a finite level broadening and on a fine grid. I need to do convergence test in the level broadening and number of grid points. But I don't know how to do the convergence test? Any comments and guidance will be appreciated. Best, Nadia Salami -- Kind regards Nick
[SIESTA-L] convergence test in the level broadening and number of grid points to obtain the non-equilibrium part of density matrix
Dear Transiesta users, According to DOI: 10.1103/PhysRevB.65.165401, I know that the non-equilibrium part of density matrix must be evaluated for a finite level broadening and on a fine grid. I need to do convergence test in the level broadening and number of grid points. But I don't know how to do the convergence test? Any comments and guidance will be appreciated. Best, Nadia Salami
Re: [SIESTA-L] convergence test
Dear Maxim: I'm so sorry to disturb you again,but I've completed the kpoint grid convergence with a reletive high value of Meshcutoff just follow your words.Then I've got many FORCE_STRESS files related to different kpoint grid.In these files the first four lines is just as follows： -1470.4169519132 -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 I want to know what does the first line -1470.4169519132 mean?And you've told me the parameter used to convergence test is Force.Is Force you mean the value -1470.4169519132 or the next 3*3 matrix -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 ? Thans very very much . Dexi Shao 2014-07-25 20:39 GMT+08:00 邵德喜 dxshao...@gmail.com: Thanks very very much for all your replay,saluted all of you ! Dexi Shao 2014-07-25 19:22 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Yes, you right. To clarify 1st keep fixed high meshcutoff and vary kpoint - select kpoint and keep it fixed but vary meshcutoff Maybe this can be vice versa, or not,but possibly it is not so important (final value of meshcutoff about 350-400 Ry, sometimes for late heavy transition metal elements, with oxygen maybe it can be more) 2014-07-25 5:53 GMT+04:00 邵德喜 dxshao...@gmail.com: I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao -- Sincerely yours, Kanhao Xue Laboratoire de Réactivité et Chimie des Solides (LRCS) UMR
Re: [SIESTA-L] convergence test
it is *.FA file contains forces (eV/Ang) Best wishe, Maxim. 2014-07-26 17:00 GMT+04:00 邵德喜 dxshao...@gmail.com: Dear Maxim: I'm so sorry to disturb you again,but I've completed the kpoint grid convergence with a reletive high value of Meshcutoff just follow your words.Then I've got many FORCE_STRESS files related to different kpoint grid.In these files the first four lines is just as follows： -1470.4169519132 -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 I want to know what does the first line -1470.4169519132 mean?And you've told me the parameter used to convergence test is Force.Is Force you mean the value -1470.4169519132 or the next 3*3 matrix -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 ? Thans very very much . Dexi Shao 2014-07-25 20:39 GMT+08:00 邵德喜 dxshao...@gmail.com: Thanks very very much for all your replay,saluted all of you ! Dexi Shao 2014-07-25 19:22 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Yes, you right. To clarify 1st keep fixed high meshcutoff and vary kpoint - select kpoint and keep it fixed but vary meshcutoff Maybe this can be vice versa, or not,but possibly it is not so important (final value of meshcutoff about 350-400 Ry, sometimes for late heavy transition metal elements, with oxygen maybe it can be more) 2014-07-25 5:53 GMT+04:00 邵德喜 dxshao...@gmail.com: I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy
Re: [SIESTA-L] convergence test
Thanks very much.Is it Spanish? I can't understant it ,but I 'll try my best to catch it . Dexi Shao 2014-07-26 21:24 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: more easily to take only max force for plot from output. See pages 79-80 of file attached - it can help you to choose these parameters 2014-07-26 17:18 GMT+04:00 Максим Арсентьев ars21031...@gmail.com: it is *.FA file contains forces (eV/Ang) Best wishe, Maxim. 2014-07-26 17:00 GMT+04:00 邵德喜 dxshao...@gmail.com: Dear Maxim: I'm so sorry to disturb you again,but I've completed the kpoint grid convergence with a reletive high value of Meshcutoff just follow your words.Then I've got many FORCE_STRESS files related to different kpoint grid.In these files the first four lines is just as follows： -1470.4169519132 -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 I want to know what does the first line -1470.4169519132 mean?And you've told me the parameter used to convergence test is Force.Is Force you mean the value -1470.4169519132 or the next 3*3 matrix -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 ? Thans very very much . Dexi Shao 2014-07-25 20:39 GMT+08:00 邵德喜 dxshao...@gmail.com: Thanks very very much for all your replay,saluted all of you ! Dexi Shao 2014-07-25 19:22 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Yes, you right. To clarify 1st keep fixed high meshcutoff and vary kpoint - select kpoint and keep it fixed but vary meshcutoff Maybe this can be vice versa, or not,but possibly it is not so important (final value of meshcutoff about 350-400 Ry, sometimes for late heavy transition metal elements, with oxygen maybe it can be more) 2014-07-25 5:53 GMT+04:00 邵德喜 dxshao...@gmail.com: I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for
Re: [SIESTA-L] convergence test
It's in French. Best wishes, Maxim. 2014-07-26 17:36 GMT+04:00 邵德喜 dxshao...@gmail.com: Thanks very much.Is it Spanish? I can't understant it ,but I 'll try my best to catch it . Dexi Shao 2014-07-26 21:24 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: more easily to take only max force for plot from output. See pages 79-80 of file attached - it can help you to choose these parameters 2014-07-26 17:18 GMT+04:00 Максим Арсентьев ars21031...@gmail.com: it is *.FA file contains forces (eV/Ang) Best wishe, Maxim. 2014-07-26 17:00 GMT+04:00 邵德喜 dxshao...@gmail.com: Dear Maxim: I'm so sorry to disturb you again,but I've completed the kpoint grid convergence with a reletive high value of Meshcutoff just follow your words.Then I've got many FORCE_STRESS files related to different kpoint grid.In these files the first four lines is just as follows： -1470.4169519132 -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 I want to know what does the first line -1470.4169519132 mean?And you've told me the parameter used to convergence test is Force.Is Force you mean the value -1470.4169519132 or the next 3*3 matrix -0.01407 -0.00013 0.4 -0.00013 0.00460 0.4 0.4 0.4 0.50126 ? Thans very very much . Dexi Shao 2014-07-25 20:39 GMT+08:00 邵德喜 dxshao...@gmail.com: Thanks very very much for all your replay,saluted all of you ! Dexi Shao 2014-07-25 19:22 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Yes, you right. To clarify 1st keep fixed high meshcutoff and vary kpoint - select kpoint and keep it fixed but vary meshcutoff Maybe this can be vice versa, or not,but possibly it is not so important (final value of meshcutoff about 350-400 Ry, sometimes for late heavy transition metal elements, with oxygen maybe it can be more) 2014-07-25 5:53 GMT+04:00 邵德喜 dxshao...@gmail.com: I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead o pportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to
Re: [SIESTA-L] convergence test
Yes, you right. To clarify 1st keep fixed high meshcutoff and vary kpoint - select kpoint and keep it fixed but vary meshcutoff Maybe this can be vice versa, or not,but possibly it is not so important (final value of meshcutoff about 350-400 Ry, sometimes for late heavy transition metal elements, with oxygen maybe it can be more) 2014-07-25 5:53 GMT+04:00 邵德喜 dxshao...@gmail.com: I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao -- Sincerely yours, Kanhao Xue Laboratoire de Réactivité et Chimie des Solides (LRCS) UMR CNRS 7314 Université de Picardie Jules Verne 33 Rue Saint Leu, 80039 Amiens Cedex, France -- Best wishes, Maxim Arsent'ev, D.Sc.(Chemistry) Laboratory of research of nanostructures Institute of Silicate Chemistry of RAS -- Best wishes, Maxim Arsent'ev, D.Sc.(Chemistry) Laboratory of research of nanostructures Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] convergence test
Thanks very very much for all your replay,saluted all of you ! Dexi Shao 2014-07-25 19:22 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Yes, you right. To clarify 1st keep fixed high meshcutoff and vary kpoint - select kpoint and keep it fixed but vary meshcutoff Maybe this can be vice versa, or not,but possibly it is not so important (final value of meshcutoff about 350-400 Ry, sometimes for late heavy transition metal elements, with oxygen maybe it can be more) 2014-07-25 5:53 GMT+04:00 邵德喜 dxshao...@gmail.com: I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao -- Sincerely yours, Kanhao Xue Laboratoire de Réactivité et Chimie des Solides (LRCS) UMR CNRS 7314 Université de Picardie Jules Verne 33 Rue Saint Leu, 80039 Amiens Cedex, France -- Best wishes, Maxim Arsent'ev, D.Sc.(Chemistry) Laboratory of research of nanostructures Institute of Silicate Chemistry of RAS -- Best wishes, Maxim Arsent'ev, D.Sc.(Chemistry) Laboratory of research of nanostructures Institute of Silicate Chemistry of RAS
[SIESTA-L] convergence test
Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao converg-1.xls Description: MS-Excel spreadsheet
Re: [SIESTA-L] convergence test
Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao -- Sincerely yours, Kanhao Xue Laboratoire de Réactivité et Chimie des Solides (LRCS) UMR CNRS 7314 Université de Picardie Jules Verne 33 Rue Saint Leu, 80039 Amiens Cedex, France
Re: [SIESTA-L] convergence test
Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao -- Sincerely yours, Kanhao Xue Laboratoire de Réactivité et Chimie des Solides (LRCS) UMR CNRS 7314 Université de Picardie Jules Verne 33 Rue Saint Leu, 80039 Amiens Cedex, France
Re: [SIESTA-L] convergence test
I see.You mean that I should set CG steps as usual as we do structual relaxation(foer example :500) during all the convergence tests . And consider kpoint grid convergence first with a reletive high value of Meshcutoff.When it is done ,plot how it depends on FORCES from the output file in the final convergency step .Then consider meshcutoff convergence test with the convergency kpoint grid. When the convergence tests is completed ,structual relaxation is also completed,so I do not need to do structual relaxation additionally. Is it all right? Thanks for all your replay! ( PS: I‘m so sorry but the service is just out of work.I'll upload my input fdf file as soon as it recovers.) Dexi Shao 2014-07-25 1:14 GMT+08:00 Максим Арсентьев ars21031...@gmail.com: Dear Dexi! You should not set nember of CG steps to zero - it is not necessary. You right that you should perform convergence of your property of interest, but first structural relaxation is important as I suppose - maybe not in your case, but often it is so. So select high meshcutoff about 400-500 and vary density of kpoint grid and plot how it depends on FORCES. From this plot you will select kpoints, then take this kpoints fixed and vary meshcutoff to plot dependence on forces again. It is forces, not total energy are relevant for structural relaxation. Enter in google graphene transiesta and 1st link will help I believe. четверг, 24 июля 2014 г. пользователь 邵德喜 написал: Thanks very much.The unit is Ry. I was told that when I get the original structure ,I should do convergence tests first,structual relaxation comes second. In my opinion ,do convergence test without structual relaxation I should set CG step to zero. problem1:I check the convergency (in terms of gradually increase Meshcutoff) of the total energy may be useless because I was told we should check the convergence not of the total energy but of those properties which are relevant for our study in question . But I just want to study the current-voltage relation ,it's expensive for me to check the convergence of its transport property. So I check the convergence of total energy instead opportunisticly. The input.fdf file is as follows. Thans for your replay. Dexi Shao 2014-07-24 18:22 GMT+08:00 Kanhao Xue xuekan...@gmail.com: Dear Dexi, I wonder what is the unit of your Mesh cutoff? Is it in Ry? Siesta is using atomic orbital as the basis, which is quite different from plane-wave based codes like VASP. In plane-wave code, the plane wave cutoff energy directly determines your basis, and thus influences the result a lot. In Siesta we have the mesh cutoff concept. A large cutoff is required for accurate result. Usually for LDA, 250 Ry is sufficient and for GGA 350 Ry is sufficient. In my opinion you do not need to go up to 600 Ry to check. However, indeed your energy changes too much with Mesh cutoff. Can you upload your input fdf file for more information? Kanhao On Thu, Jul 24, 2014 at 11:43 AM, 邵德喜 dxshao...@gmail.com wrote: Dear everyone: I have done the convergence test when I check the convergency (in terms of gradually increase Meshcutoff) of the total energy, I found that the Total energy rise up when I increase the parameter of Meshcutoff (please refer to the accesory) .Is it reasonable? Ahy replay will be appreciated. Dexi Shao -- Sincerely yours, Kanhao Xue Laboratoire de Réactivité et Chimie des Solides (LRCS) UMR CNRS 7314 Université de Picardie Jules Verne 33 Rue Saint Leu, 80039 Amiens Cedex, France -- Best wishes, Maxim Arsent'ev, D.Sc.(Chemistry) Laboratory of research of nanostructures Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence
On 12/19/13, Suman Chowdhury sumanchowdhur...@gmail.com wrote: Dear all siesta user, How do I know if my molecular dynamics simulation is converging or not? Hi dear Suman when you face with outcoor: Relaxedatomic coordinates (Ang).in your output file.
[SIESTA-L] Convergence
Dear all siesta user, How do I know if my molecular dynamics simulation is converging or not?
Re: [SIESTA-L] Convergence problem on heterostructure.
Sorry for my sluggishness - the picture above vise versa - it seems that higher MD.MaxCGDispl gives higher forces, but maybe they will decrease faster... Also my MeshCutoff is already 800 because of Ce Best wishes, Maxim. 2013/11/25 Максим Арсентьев ars21031...@gmail.com Dear Frank! It seems for normal_main that you never reach 0.04 eV/Angstrom. How to fix that? And here is mt result (seems that MD.MaxCGDispl 0.15 Ang and DM.Tolerance 0.1 behave better). Best wishes, Maxim. 2013/11/22 fma...@icp.uni-stuttgart.de A fluctuation in max. forces is normal. Attached is a max. force graph of a few CG geometry optimizations I did of isolated molecules. Further details are in the following PDF: (keyword EggBox Effect) http://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdf So the lower the max. force is, the less CG steps are required in my experience and only then can you achieve low MD.MaxForceTol values. To reduce the max. forces you have to increase the finess of the real space grid, by either increasing the MeshCutOff value or making use of GridCellSampling: http://departments.icmab.es/leem/siesta/Documentation/Manuals/siesta-3.1-manual/node47.html Best wishes, Frank I'm doing geometry optimisation if I correctly set it (as I understand it's by default) Best Wishes, Maxim. 2013/11/22 Herbert Fruchtl herbert.fruc...@st-andrews.ac.uk Just to clarify: You are doing MD, not a geometry optimisation, and you are surprised that the forces don't go to zero? What are you trying to find? For a minimum energy structure, you need a geometry optimisation. In an MD simulation the total (potential+kinetic) energy should remain constant (depending on your ensemble...), but the forces will vary. Herbert On 22/11/2013 08:58, Максим Арсентьев wrote: Dear Sebastian! My calculations convverge at each MD cycle (about 300-400 SCF cycles), but max force oscillate (about 0.1-0.04 eV/anstrom) and I can not finish MD, and am already at 63 cycle and the calculation cycle is still going, although the system is quite large and contains 47 atoms. My counter question to all: does this oscillation of forces normal? I would like to strict DM.Tolerance to 0.1 (it is VERY important to accurate calculation of forces) and to limit MD.MaxCGDispl . Best Wishes, Maxim 2013/11/21 Sebastian Caicedo Davila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure
Re: [SIESTA-L] Convergence problem on heterostructure.
of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com mailto:ars21031...@gmail.com mailto:ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es mailto:siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/ http://www.nature.com/nnano/journal/v7/n9/extref/ nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammourim...@nmsu.edu mailto:m...@nmsu.edu mailto:m...@nmsu.edu mailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com mailto:ars21031...@gmail.com mailto:ars21031...@gmail.com mailto:ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano. 2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D.sebastian.caic...@correounivalle.edu.co
Re: [SIESTA-L] Convergence problem on heterostructure.
number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/ nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammourim...@nmsu.edu mailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com mailto:ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano. 2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D.sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence problem on heterostructure.
Dear Frank! It seems for normal_main that you never reach 0.04 eV/Angstrom. How to fix that? And here is mt result (seems that MD.MaxCGDispl 0.15 Ang and DM.Tolerance 0.1 behave better). Best wishes, Maxim. 2013/11/22 fma...@icp.uni-stuttgart.de A fluctuation in max. forces is normal. Attached is a max. force graph of a few CG geometry optimizations I did of isolated molecules. Further details are in the following PDF: (keyword EggBox Effect) http://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdf So the lower the max. force is, the less CG steps are required in my experience and only then can you achieve low MD.MaxForceTol values. To reduce the max. forces you have to increase the finess of the real space grid, by either increasing the MeshCutOff value or making use of GridCellSampling: http://departments.icmab.es/leem/siesta/Documentation/Manuals/siesta-3.1-manual/node47.html Best wishes, Frank I'm doing geometry optimisation if I correctly set it (as I understand it's by default) Best Wishes, Maxim. 2013/11/22 Herbert Fruchtl herbert.fruc...@st-andrews.ac.uk Just to clarify: You are doing MD, not a geometry optimisation, and you are surprised that the forces don't go to zero? What are you trying to find? For a minimum energy structure, you need a geometry optimisation. In an MD simulation the total (potential+kinetic) energy should remain constant (depending on your ensemble...), but the forces will vary. Herbert On 22/11/2013 08:58, Максим Арсентьев wrote: Dear Sebastian! My calculations convverge at each MD cycle (about 300-400 SCF cycles), but max force oscillate (about 0.1-0.04 eV/anstrom) and I can not finish MD, and am already at 63 cycle and the calculation cycle is still going, although the system is quite large and contains 47 atoms. My counter question to all: does this oscillation of forces normal? I would like to strict DM.Tolerance to 0.1 (it is VERY important to accurate calculation of forces) and to limit MD.MaxCGDispl . Best Wishes, Maxim 2013/11/21 Sebastian Caicedo Davila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/ nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis
Re: [SIESTA-L] Convergence problem on heterostructure.
Dear Sebastian! My calculations convverge at each MD cycle (about 300-400 SCF cycles), but max force oscillate (about 0.1-0.04 eV/anstrom) and I can not finish MD, and am already at 63 cycle and the calculation cycle is still going, although the system is quite large and contains 47 atoms. My counter question to all: does this oscillation of forces normal? I would like to strict DM.Tolerance to 0.1 (it is VERY important to accurate calculation of forces) and to limit MD.MaxCGDispl . Best Wishes, Maxim 2013/11/21 Sebastian Caicedo Davila sebastian.caic...@correounivalle.edu.co Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland -- *From:* siesta-l-requ...@uam.es [siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com] *Sent:* Wednesday, November 20, 2013 7:29 AM *To:* siesta-l@uam.es *Subject:* Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as here http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammouri m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased
Re: [SIESTA-L] Convergence problem on heterostructure.
Just to clarify: You are doing MD, not a geometry optimisation, and you are surprised that the forces don't go to zero? What are you trying to find? For a minimum energy structure, you need a geometry optimisation. In an MD simulation the total (potential+kinetic) energy should remain constant (depending on your ensemble...), but the forces will vary. Herbert On 22/11/2013 08:58, Максим Арсентьев wrote: Dear Sebastian! My calculations convverge at each MD cycle (about 300-400 SCF cycles), but max force oscillate (about 0.1-0.04 eV/anstrom) and I can not finish MD, and am already at 63 cycle and the calculation cycle is still going, although the system is quite large and contains 47 atoms. My counter question to all: does this oscillation of forces normal? I would like to strict DM.Tolerance to 0.1 (it is VERY important to accurate calculation of forces) and to limit MD.MaxCGDispl . Best Wishes, Maxim 2013/11/21 Sebastian Caicedo Davila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammourim...@nmsu.edu mailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com mailto:ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D.sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been
Re: [SIESTA-L] Convergence problem on heterostructure.
I'm doing geometry optimisation if I correctly set it (as I understand it's by default) Best Wishes, Maxim. 2013/11/22 Herbert Fruchtl herbert.fruc...@st-andrews.ac.uk Just to clarify: You are doing MD, not a geometry optimisation, and you are surprised that the forces don't go to zero? What are you trying to find? For a minimum energy structure, you need a geometry optimisation. In an MD simulation the total (potential+kinetic) energy should remain constant (depending on your ensemble...), but the forces will vary. Herbert On 22/11/2013 08:58, Максим Арсентьев wrote: Dear Sebastian! My calculations convverge at each MD cycle (about 300-400 SCF cycles), but max force oscillate (about 0.1-0.04 eV/anstrom) and I can not finish MD, and am already at 63 cycle and the calculation cycle is still going, although the system is quite large and contains 47 atoms. My counter question to all: does this oscillation of forces normal? I would like to strict DM.Tolerance to 0.1 (it is VERY important to accurate calculation of forces) and to limit MD.MaxCGDispl . Best Wishes, Maxim 2013/11/21 Sebastian Caicedo Davila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/ nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammourim...@nmsu.edu mailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com mailto:ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano
Re: [SIESTA-L] Convergence problem on heterostructure.
mailto:siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/__nnano/journal/v7/n9/extref/__nnano.2012.127-s2.txt http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammourim...@nmsu.edu mailto:m...@nmsu.edu mailto:m...@nmsu.edu mailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caicedo@__correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co mailto:sebastian.caicedo@__correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com mailto:ars21031...@gmail.com mailto:ars21031...@gmail.com mailto:ars21031...@gmail.com__ wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/__journal/v7/n9/extref/nnano.__2012.127-s2.txt http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D.sebastian.caicedo@__correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co mailto:sebastian.caicedo@__correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual
Re: [SIESTA-L] Convergence problem on heterostructure.
A fluctuation in max. forces is normal. Attached is a max. force graph of a few CG geometry optimizations I did of isolated molecules. Further details are in the following PDF: (keyword EggBox Effect) http://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdf So the lower the max. force is, the less CG steps are required in my experience and only then can you achieve low MD.MaxForceTol values. To reduce the max. forces you have to increase the finess of the real space grid, by either increasing the MeshCutOff value or making use of GridCellSampling: http://departments.icmab.es/leem/siesta/Documentation/Manuals/siesta-3.1-manual/node47.html Best wishes, Frank I'm doing geometry optimisation if I correctly set it (as I understand it's by default) Best Wishes, Maxim. 2013/11/22 Herbert Fruchtl herbert.fruc...@st-andrews.ac.uk Just to clarify: You are doing MD, not a geometry optimisation, and you are surprised that the forces don't go to zero? What are you trying to find? For a minimum energy structure, you need a geometry optimisation. In an MD simulation the total (potential+kinetic) energy should remain constant (depending on your ensemble...), but the forces will vary. Herbert On 22/11/2013 08:58, ÐакÑим ÐÑÑенÑÑев wrote: Dear Sebastian! My calculations convverge at each MD cycle (about 300-400 SCF cycles), but max force oscillate (about 0.1-0.04 eV/anstrom) and I can not finish MD, and am already at 63 cycle and the calculation cycle is still going, although the system is quite large and contains 47 atoms. My counter question to all: does this oscillation of forces normal? I would like to strict DM.Tolerance to 0.1 (it is VERY important to accurate calculation of forces) and to limit MD.MaxCGDispl . Best Wishes, Maxim 2013/11/21 Sebastian Caicedo Davila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic...@correounivalle.edu.co Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl d.otalv...@utwente.nl mailto:d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland *From:*siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es [siesta-l-requ...@uam.es mailto:siesta-l-requ...@uam.es] on behalf of ÐакÑим ÐÑÑенÑÑев [ars21031...@gmail.com mailto:ars21031...@gmail.com] *Sent:*Wednesday, November 20, 2013 7:29 AM *To:*siesta-l@uam.es mailto:siesta-l@uam.es *Subject:*Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/ nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammourim...@nmsu.edu mailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co mailto:sebastian.caic
RE: [SIESTA-L] Convergence problem on heterostructure.
Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland From: siesta-l-requ...@uam.es [siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com] Sent: Wednesday, November 20, 2013 7:29 AM To: siesta-l@uam.es Subject: Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as here http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammouri m...@nmsu.edumailto:m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.comailto:sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.commailto:ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.comailto:sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence problem on heterostructure.
Dear Maxim, I'd like to ask you something about your simulation. It turns out that when I do a geometry optimisation, I don't get convergence for the values of energy in the self consistent cycle. When you performed your simulation, did the values of energy converged for every MD step, or did they eventually converged for the last MD steps? Thank you very much for your help! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 18, 2013, at 11:50 AM, Максим Арсентьев ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence problem on heterostructure.
Hello Diana!! Thanks a lot for your suggestion! I'll give it a try and maybe I'll be bothering you again, if it is not an imposition, of course. Gracias de nuevo, y saludos desde tu tierra! Sebastian Caicedo Davila Bionanoelectronics Research Group School of electrical and electronics engineering Universidad del Valle Cali - Colombia. On Nov 21, 2013, at 11:02 AM, d.otalv...@utwente.nl d.otalv...@utwente.nl wrote: Hello Sebastian, I have had similar problems with convergence of the scf cycle. They were solved when I changed from Pulay mixing to Broyden mixing for the charge density. Usually if something doesn't converge in 250 cycles, it probably never will. Also you can play with the DM.Tolerance. For very difficult systems I set it to 0.0008 and then restart from that DM file with a stringer convergence criterion. Here is the settings I am using for my calculations: #Convergence of SCF--- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations250 # Maximum number of SCF iter DM.NumberBroyden3 DM.MixingWeight 0.01 # New DM amount for next SCF cycle DM.Tolerance0.0001 # Tolerance in maximum difference DM.UseSaveDM.false. # to use continuation files DM.NumberKick 100 DM.KickMixingWeight0.01 #-- Suerte! PS: yo tambien soy caleña! Diana M. Otálvaro PhD Candidate Computational Material Science MESA+ Institute of Nanotechnology University of Twente. Enschede, Nederland From: siesta-l-requ...@uam.es [siesta-l-requ...@uam.es] on behalf of Максим Арсентьев [ars21031...@gmail.com] Sent: Wednesday, November 20, 2013 7:29 AM To: siesta-l@uam.es Subject: Re: [SIESTA-L] Convergence problem on heterostructure. I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as herehttp://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammouri m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence problem on heterostructure.
I forgot to tell you - besides DM.MixingWeight 0.05 I also set PAO.EnergyShift to 65 meV as here http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt . Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS 2013/11/18 Mahmoud Hammouri m...@nmsu.edu You could also try to increase the non-equilibrium contour points. Mahmoud On Nov 18, 2013, at 11:22, Sebastián Cauce do Dávila sebastian.caic...@correounivalle.edu.co wrote: Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence problem on heterostructure.
I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence problem on heterostructure.
Thank you very much for your suggestion. I will give it a try! Sebastián Caicedo Dávila Bionanoelectronics research group School of electrical and electronics engineering Universidad del Valle Cali - Colombia On 18/11/2013, at 11:50, Максим Арсентьев ars21031...@gmail.com wrote: I also try some siesta only six months, but for a more heterogeneous structure I noticed that the reduction DM.MixingWeight reduces oscillations in convergence, so try to play with it (decrease DM.MixingWeight). See http://www.nature.com/nnano/journal/v7/n9/extref/nnano.2012.127-s2.txt Maybe someone will tell something else. 2013/11/18 Sebastian Caicedo D. sebastian.caic...@correounivalle.edu.co Dear Siesta/Transiesta users, I've been using the software for a couple months and made some calculations of InAs and InP unit cell quite successfully. Now I intend to work on transiesta calculations of a heterostructure of both these materials (InAs bulk with InP barriers). As a first approach I want to simulate a couple of InAs unit cells and the barriers shall be unit cells of InP, just to get a qualitative idea. The electrodes .TSHS files were obtained from the previous calculations of the InAs unit cell. I wanted first to run a siesta calculation of the whole heterostructure, so I can take a look at the eigenvalues and define a correct value for the parameter TS.ComplexContour.Emin, as suggested in the manual. The structure has 120 atoms and I'm using LDA Pseudopotentials, obtained from the siesta webpage. (In Pseudo includes semicore electrons, so we have enough electrons to build a zinc-blende structure). The issue is that after 1000 SCF steps, the energy doesn't converge at all. I'm doing a Gamma-point calculation, but even when I make the k-grid 1x1x15 it doesn't work either. I used the same MeshCutoff as for the individual materials unit cell an 200Ry and I increased the Energy Shift, so the calculation would be a bit faster though less precise. I attach the Input (INPUT.fdf, STRUCTURE.fdf) and output files. I would really appreciate any suggestions. sincerely Sebastian Caicedo Davila Bionanoelectronics Group School of Electrical and Electronics Engineering Universidad del Valle Cali-Colombia -- Best wishes, Dr. Maxim Arsent'ev Laboratory of inorganic synthesis Institute of Silicate Chemistry of RAS
Re: [SIESTA-L] Convergence for charged system...
First of all thank you very much for the suggestions, now I can reach convegence quite easily. The results should be correct: I have compared them with other calculations and for example the, charged, relaxed configurations are the same as those obtained with other meathods that I know are right. But I am still puzzled by the electronic structure: In Si5H12 with no extra charge the homo is the 16th level and I expect that if I add an electron the homo will be the 17th energy level and that it will be half filled. Neverthelss the Fermi level I obtain is still in between the 16th and 17th energy level so the homo is still the level 16! What makes me think that the result is anyway correct is that the Fermi energy is very near to the level 17 (but below it) so since I am adding some smearing (300K of electronic temperature), maybe the 17th energy level is actually occupied. Is my reasoning sound, or the Fermi energy should be abouve the 17th energy level? May I ask a brief explanation on how the Fermi energy is calculated and how the extra charge is added (and remooved)? Thanks again for past and eventual future help, Luigi. 2013/6/3 apost...@uni-osnabrueck.de Dear Luigi, Si5H12 has an even number of electrons, therefore when adding or removing one, in principle, you ought to use spin polarized = true. (This will become less and less crucial as you increase the system so it becomes more solid-like and/or metallic. But at the beginning this extra electron is likely to be carried by a isolated level in the gap, I guess). A bad convergence is probably due to swapping between competing states, where this extra electron is differently allocated in space. This might be a manifestation of a genuine physical problem in a sense that, given the symmetry of your system, the level in question may want to be degenerate (jus a guess...) whereas you populate it with a single electron. If this is indeed the case, a small distortion may help to fix the issue... However in real world you'd need to allow spin-orbit interaction to get it right. By setting spin polarization off you cheat the system and tell it to search only among doubly degenerate states. The result may converge better, but place the energy level at a wrong place. It could be useful to have a look at where the levels are (where your extra charge is sitting) and try to make sense out of it. If not caring about all above, but just technically aiming at convergence, I'd suggest you to i) reduce the mixing factor considerably, ii) first set ElectronicTemperature higher (say to 600 K) to facilitate convergence, and then gradually decrease it. Moreover the MeshCutoff of 150 Ry might be not high enough to get a good physical result, and might also affect the convergence. Good luck Andrei Postnikov Hello everybody, I am trying to perform a calculation of a charged system composed of a molecule of Si5H12 with an extra electron (and in the future also one less electron) so I set: NetCharge -1.00 SpinPolarized .true. FixSpin .true. TotalSpin 1.0 But the calculation even for this small system does not converge. If I put SpinPolarization .false. the calculation converges rapidly but I am not sure that it's correct. recently I added also: SimulateDoping .true. It helps only slightly but convergence is still far from beying achieved. These are the other parameters of my input file: -- SystemName SiH SystemLabel SiH #(1) SPECIES AND ATOMS NumberOfAtoms 17 NumberOfSpecies 2 %block ChemicalSpeciesLabel 1 14 Si # Species index, atomic number, species label 2 1 H # Species index, atomic number, species label %endblock ChemicalSpeciesLabel #(2) FUNCTIONAL xc.functional GGA xc.authors PBE SpinPolarized .true. FixSpin .true. TotalSpin 1.0 MeshCutoff150 Ry NetCharge -1.00 SimulateDoping .true. #(3) BASIS PAO.SplitTailNorm .true. #(4) SCF OPTIONS MaxSCFIterations 4000 DM.MixingWeight0.01 DM.Tolerance 1.d-4 DM.NumberPulay 3 SolutionMethod Diagon ElectronicTemperature 300 K #(5) MD OPTIONS MD.TypeOfRun CG MD.NumCGsteps 1000 #(6) OUTPUT OPTIONS #(7) CRYSTAL STRUCTURE LatticeConstant 30.0 Ang %block LatticeVectors 1.000 0.000 0.000 0.000 1.000 0.000 0.000 0.000 1.000 %endblock LatticeVectors #(8) ATOMIC COORDINATES AtomicCoordinatesFormat NotScaledCartesianAng AtomicCoordinatesAndAtomicSpecies Si-H.siesta -- I hope someone has a hint. Thanck you very much in advance. L.
Re: [SIESTA-L] Convergence for charged system...
Dear Luigi, Si5H12 has an even number of electrons, therefore when adding or removing one, in principle, you ought to use spin polarized = true. (This will become less and less crucial as you increase the system so it becomes more solid-like and/or metallic. But at the beginning this extra electron is likely to be carried by a isolated level in the gap, I guess). A bad convergence is probably due to swapping between competing states, where this extra electron is differently allocated in space. This might be a manifestation of a genuine physical problem in a sense that, given the symmetry of your system, the level in question may want to be degenerate (jus a guess...) whereas you populate it with a single electron. If this is indeed the case, a small distortion may help to fix the issue... However in real world you'd need to allow spin-orbit interaction to get it right. By setting spin polarization off you cheat the system and tell it to search only among doubly degenerate states. The result may converge better, but place the energy level at a wrong place. It could be useful to have a look at where the levels are (where your extra charge is sitting) and try to make sense out of it. If not caring about all above, but just technically aiming at convergence, I'd suggest you to i) reduce the mixing factor considerably, ii) first set ElectronicTemperature higher (say to 600 K) to facilitate convergence, and then gradually decrease it. Moreover the MeshCutoff of 150 Ry might be not high enough to get a good physical result, and might also affect the convergence. Good luck Andrei Postnikov Hello everybody, I am trying to perform a calculation of a charged system composed of a molecule of Si5H12 with an extra electron (and in the future also one less electron) so I set: NetCharge -1.00 SpinPolarized .true. FixSpin .true. TotalSpin 1.0 But the calculation even for this small system does not converge. If I put SpinPolarization .false. the calculation converges rapidly but I am not sure that it's correct. recently I added also: SimulateDoping .true. It helps only slightly but convergence is still far from beying achieved. These are the other parameters of my input file: -- SystemName SiH SystemLabel SiH #(1) SPECIES AND ATOMS NumberOfAtoms 17 NumberOfSpecies 2 %block ChemicalSpeciesLabel 1 14 Si # Species index, atomic number, species label 2 1 H # Species index, atomic number, species label %endblock ChemicalSpeciesLabel #(2) FUNCTIONAL xc.functional GGA xc.authors PBE SpinPolarized .true. FixSpin .true. TotalSpin 1.0 MeshCutoff150 Ry NetCharge -1.00 SimulateDoping .true. #(3) BASIS PAO.SplitTailNorm .true. #(4) SCF OPTIONS MaxSCFIterations 4000 DM.MixingWeight0.01 DM.Tolerance 1.d-4 DM.NumberPulay 3 SolutionMethod Diagon ElectronicTemperature 300 K #(5) MD OPTIONS MD.TypeOfRun CG MD.NumCGsteps 1000 #(6) OUTPUT OPTIONS #(7) CRYSTAL STRUCTURE LatticeConstant 30.0 Ang %block LatticeVectors 1.000 0.000 0.000 0.000 1.000 0.000 0.000 0.000 1.000 %endblock LatticeVectors #(8) ATOMIC COORDINATES AtomicCoordinatesFormat NotScaledCartesianAng AtomicCoordinatesAndAtomicSpecies Si-H.siesta -- I hope someone has a hint. Thanck you very much in advance. L.
Re: [SIESTA-L] Convergence for charged system...
Dear Luigi, Si5H12 has an even number of electrons, therefore when adding or removing one, in principle, you ought to use spin polarized = true. (This will become less and less crucial as you increase the system so it becomes more solid-like and/or metallic. But at the beginning this extra electron is likely to be carried by a isolated level in the gap, I guess). A bad convergence is probably due to swapping between competing states, where this extra electron is differently allocated in space. This might be a manifestation of a genuine physical problem in a sense that, given the symmetry of your system, the level in question may want to be degenerate (jus a guess...) whereas you populate it with a single electron. If this is indeed the case, a small distortion may help to fix the issue... However in real world you'd need to allow spin-orbit interaction to get it right. By setting spin polarization off you cheat the system and tell it to search only among doubly degenerate states. The result may converge better, but place the energy level at a wrong place. It could be useful to have a look at where the levels are (where your extra charge is sitting) and try to make sense out of it. If not caring about all above, but just technically aiming at convergence, I'd suggest you to i) reduce the mixing factor considerably, ii) first set ElectronicTemperature higher (say to 600 K) to facilitate convergence, and then gradually decrease it. Moreover the MeshCutoff of 150 Ry might be not high enough to get a good physical result, and might also affect the convergence. Good luck Andrei Postnikov Hello everybody, I am trying to perform a calculation of a charged system composed of a molecule of Si5H12 with an extra electron (and in the future also one less electron) so I set: NetCharge -1.00 SpinPolarized .true. FixSpin .true. TotalSpin 1.0 But the calculation even for this small system does not converge. If I put SpinPolarization .false. the calculation converges rapidly but I am not sure that it's correct. recently I added also: SimulateDoping .true. It helps only slightly but convergence is still far from beying achieved. These are the other parameters of my input file: -- SystemName SiH SystemLabel SiH #(1) SPECIES AND ATOMS NumberOfAtoms 17 NumberOfSpecies 2 %block ChemicalSpeciesLabel 1 14 Si # Species index, atomic number, species label 2 1 H # Species index, atomic number, species label %endblock ChemicalSpeciesLabel #(2) FUNCTIONAL xc.functional GGA xc.authors PBE SpinPolarized .true. FixSpin .true. TotalSpin 1.0 MeshCutoff150 Ry NetCharge -1.00 SimulateDoping .true. #(3) BASIS PAO.SplitTailNorm .true. #(4) SCF OPTIONS MaxSCFIterations 4000 DM.MixingWeight0.01 DM.Tolerance 1.d-4 DM.NumberPulay 3 SolutionMethod Diagon ElectronicTemperature 300 K #(5) MD OPTIONS MD.TypeOfRun CG MD.NumCGsteps 1000 #(6) OUTPUT OPTIONS #(7) CRYSTAL STRUCTURE LatticeConstant 30.0 Ang %block LatticeVectors 1.000 0.000 0.000 0.000 1.000 0.000 0.000 0.000 1.000 %endblock LatticeVectors #(8) ATOMIC COORDINATES AtomicCoordinatesFormat NotScaledCartesianAng AtomicCoordinatesAndAtomicSpecies Si-H.siesta -- I hope someone has a hint. Thanck you very much in advance. L.
Re: [SIESTA-L] Convergence for charged system...
Dear Abraham, first of all thank you very much for your reply, it was very usefull and I was able to make my small test calculations converge. But now I moved to larger systems and I get very anoying runtime errors that make my calculations crash. The runtime errors I get are the following: forrtl: severe (174): SIGSEGV, segmentation fault occurred Image PCRoutineLineSource mca_pml_ob1.so 2B0077DBE1C8 Unknown Unknown Unknown mca_pml_ob1.so 2B0077DBE493 Unknown Unknown Unknown mca_btl_sm.so 2B0078C462AC Unknown Unknown Unknown libopen-pal.so.0 2B007319D427 Unknown Unknown Unknown libmpi.so.02B0072C8B596 Unknown Unknown Unknown libmpi.so.02B0072CBA83E Unknown Unknown Unknown libmpi_f77.so.02B0072A4744F Unknown Unknown Unknown siesta_mod2004BA2B0 redistribxy_ 657 fft3d.F siesta_mod2004ABAE3 fft_ 196 fft3d.F siesta_mod2005D89DE poison_ 109 poison.F siesta_mod2004A1AD2 dhscf_535 dhscf.F siesta_mod2005FEB57 m_setup_hamiltoni 214 setup_hamiltonian.F siesta_mod20063D68A m_siesta_forces_m 78 siesta_forces.F siesta_mod20063D86C MAIN__ 23 siesta.F siesta_mod20044CEEC Unknown Unknown Unknown libc.so.6 0030FFE1D994 Unknown Unknown Unknown siesta_mod20044CDF9 Unknown Unknown Unknown [compute-2-18.local:989144] [[59087,0],1]-[[59087,1],12] mca_oob_tcp_msg_recv: readv failed: Connection reset by peer (104) -- mpiexec has exited due to process rank 12 with PID 989149 on node compute-2-18 exiting without calling finalize. This may have caused other processes in the application to be terminated by signals sent by mpiexec (as reported here). -- I already encoutered segmentation faults and I got through them by increasing the number of cores, so that is what I am going to do now as well... but this isn't something I can keep doing since now I am dealing with only 72 atoms at a time but I will have to simulate much bigger systems. I hope I am not asking to much but have you ever encoutered such erros? Have you any clue? It looks like if the problem is in the fft routine so I will start increasing and decreasing the MeshCutoff that now is set to 150 Ry for a 20x37x20 Angstrom supercell, what do you think? Thank you very mch for any help you can give me. With best regards, Luigi. 2013/5/30 Abraham Hmiel abehm...@gmail.com Hi Luigi, A few suggestions: 1) Your MaxCGsteps and MaxSCFIterations are both very large, convergence in many problems can be adequately approached by using a small number of CG updates, like 10-50 to perform a stepwise convergence, changing mixing parameters slightly between runs, observing how the forces change between CG steps and using UseSaveData = .true. to use saved DM's, XV's and so on. MaxSCFIterations could be around 100-250, much more than that and you're wasting your time, might I add. 2) Try using more Pulay convergence parameters. This will probably get you to convergence if you choose them right. You will have to modify them until you attain something that works. For starters, try: DM.NumberPulay 5 DM.MixingWeight 0.002 DM.NumberKick 12 DM.KickMixingWeight0.0766 I've found that oftentimes the Pulay kick will knock your SCF cycle into a convergence basin, especially with charged systems or hard pseudopotentials. Look in the manual to see how these are defined. 3) I'm not sure why SimulateDoping would help you, it seems unphysical to me in a molecule rather than a solid. Your meshcutoff also seems low, but it won't help your convergence problem. Look into converging that value once you can close your SCF loop. 4) 30x30x30A cell is a big box for a small molecule, perhaps there may be a benefit to using a smaller cell, since a Madelung correction term is applied for molecules? If you go to a 20x20x20 cell does the convergence issue get worse? That's all I can think of for now, definitely look into the Pulay convergence acceleration though. Warm regards, *Abraham Hmiel* Katherine Belz Groves Fellow in Nanoscience Xue Group, College of Nanoscale Science and Engineering at SUNY Albany http://abehmiel.net/about
[SIESTA-L] Convergence for charged system...
Hello everybody, I am trying to perform a calculation of a charged system composed of a molecule of Si5H12 with an extra electron (and in the future also one less electron) so I set: NetCharge -1.00 SpinPolarized .true. FixSpin .true. TotalSpin 1.0 But the calculation even for this small system does not converge. If I put SpinPolarization .false. the calculation converges rapidly but I am not sure that it's correct. recently I added also: SimulateDoping .true. It helps only slightly but convergence is still far from beying achieved. These are the other parameters of my input file: -- SystemName SiH SystemLabel SiH #(1) SPECIES AND ATOMS NumberOfAtoms 17 NumberOfSpecies 2 %block ChemicalSpeciesLabel 1 14 Si # Species index, atomic number, species label 2 1 H # Species index, atomic number, species label %endblock ChemicalSpeciesLabel #(2) FUNCTIONAL xc.functional GGA xc.authors PBE SpinPolarized .true. FixSpin .true. TotalSpin 1.0 MeshCutoff150 Ry NetCharge -1.00 SimulateDoping .true. #(3) BASIS PAO.SplitTailNorm .true. #(4) SCF OPTIONS MaxSCFIterations 4000 DM.MixingWeight0.01 DM.Tolerance 1.d-4 DM.NumberPulay 3 SolutionMethod Diagon ElectronicTemperature 300 K #(5) MD OPTIONS MD.TypeOfRun CG MD.NumCGsteps 1000 #(6) OUTPUT OPTIONS #(7) CRYSTAL STRUCTURE LatticeConstant 30.0 Ang %block LatticeVectors 1.000 0.000 0.000 0.000 1.000 0.000 0.000 0.000 1.000 %endblock LatticeVectors #(8) ATOMIC COORDINATES AtomicCoordinatesFormat NotScaledCartesianAng AtomicCoordinatesAndAtomicSpecies Si-H.siesta -- I hope someone has a hint. Thanck you very much in advance. L.
Re: [SIESTA-L] Convergence for charged system...
Hi Luigi, A few suggestions: 1) Your MaxCGsteps and MaxSCFIterations are both very large, convergence in many problems can be adequately approached by using a small number of CG updates, like 10-50 to perform a stepwise convergence, changing mixing parameters slightly between runs, observing how the forces change between CG steps and using UseSaveData = .true. to use saved DM's, XV's and so on. MaxSCFIterations could be around 100-250, much more than that and you're wasting your time, might I add. 2) Try using more Pulay convergence parameters. This will probably get you to convergence if you choose them right. You will have to modify them until you attain something that works. For starters, try: DM.NumberPulay 5 DM.MixingWeight 0.002 DM.NumberKick 12 DM.KickMixingWeight0.0766 I've found that oftentimes the Pulay kick will knock your SCF cycle into a convergence basin, especially with charged systems or hard pseudopotentials. Look in the manual to see how these are defined. 3) I'm not sure why SimulateDoping would help you, it seems unphysical to me in a molecule rather than a solid. Your meshcutoff also seems low, but it won't help your convergence problem. Look into converging that value once you can close your SCF loop. 4) 30x30x30A cell is a big box for a small molecule, perhaps there may be a benefit to using a smaller cell, since a Madelung correction term is applied for molecules? If you go to a 20x20x20 cell does the convergence issue get worse? That's all I can think of for now, definitely look into the Pulay convergence acceleration though. Warm regards, *Abraham Hmiel* Katherine Belz Groves Fellow in Nanoscience Xue Group, College of Nanoscale Science and Engineering at SUNY Albany http://abehmiel.net/about
Re: [SIESTA-L] Convergence???
Hi Gregorio you don't have any problem in the run, this unusual because of the large number of atoms in addition you calculated optical properties and band structure at the same time this requires a great time On Thu, Nov 22, 2012 at 3:54 PM, Gregorio García Moreno gjgar...@ujaen.eswrote: The optimization is running (I don't know slowly or quickly, it's depends). Note, the max atomic forces and siesta E_KS energy decrease as a function of cycles number (although there is some increases). I think it's ok (I didn't see the optimization parameters. But I think you are using a high DM. tolerence (1.d-5) compared with a low MessCutoff. I would recomend you to increase meshcutoff. Regards Gregorio I am doing force minimization using SIESTA and my calculation is still running (more than 4 days) i think this is convergence problem. Can any one help me in this regard. I have attached mu output file. With regards hakkim -- Gregorio García Moreno, PhD Modélisation des Systèmes Complexes LECIME UMR 7575, ENSCP Chimie Paristech 11, rue Pierre et Marie Curie - 75231 Paris Cedex 05 -- *Nibras Mossa Umran* *PhD,Research Scholar * *Panjab University ** * *Chandigarh** * *I**ndia.*
Re: [SIESTA-L] Convergence???
The optimization is running (I don't know slowly or quickly, it's depends). Note, the max atomic forces and siesta E_KS energy decrease as a function of cycles number (although there is some increases). I think it's ok (I didn't see the optimization parameters. But I think you are using a high DM. tolerence (1.d-5) compared with a low MessCutoff. I would recomend you to increase meshcutoff. Regards Gregorio I am doing force minimization using SIESTA and my calculation is still running (more than 4 days) i think this is convergence problem. Can any one help me in this regard. I have attached mu output file. With regards hakkim -- Gregorio García Moreno, PhD Modélisation des Systèmes Complexes LECIME UMR 7575, ENSCP Chimie Paristech 11, rue Pierre et Marie Curie - 75231 Paris Cedex 05
[SIESTA-L] Convergence problem in geometry optimization
Hello, all I am doing a geometry optimization of graphene which is passivated with Hydrogen atoms at the edge of the sheet, and a Hydrogen atom is adsorbed on the top of a carbon atom in the center of the grapheme layer. In the calculation, the convergence criterion is MD.MaxforceTol = 0.01 eV/Ang, but the best result obtained in my calculation is Max force = 0.012 eV/Ang. So, it is not fully converged. I have tried many kinds of methods, such as increasing the MeshCutoff (actually to 400 Ry), and increasing the ElectronicTemperature (to 35 meV). Unfortunately, all failed. I do not know what is the problem and how to fix it. Any comment is appreciated. I attach the input file and the geometry of the configuration. Thank you. Fei Mao Beijing Normal University adsorb.xyz Description: Binary data adsorb.fdf Description: Binary data
Re:[SIESTA-L] Convergence: DM.Tolerance vs. DM.Energy.Tolerance
Hi, You might also try the Broyden method, which by default operates with a self-adjusting mixing weight. Check the manual. Best, Roberto On Fri, 10 Aug 2012, shihitwh wrote: Hi, Diana You can try to change the parameters DM.MixingWeight DM.NumberPulay . Best Wishes! Shi At 2012-04-24 02:41:07,Diana Otalvaro d.otalv...@utwente.nl wrote: Hello everyone, I have a lot of problems converging the DM of a structure where I force antiparallel magnetization in the two Ni leads, and have a scattering region made of non-magnetic material. I force the AP magnetization by using %block DM.InitSpin. I wonder if the DM.Tolerance criterion is a stricter convergence criterion than DM.Energy.Tolerance. I suspect that if I were to use Energy convergence instead of DM, convergence it would be faster, since the energy is variational whereas the elements of DM are not. Am I correct? Even so, I don't know if I can bypass the DM.Tolerance tag. The manual tells me that DM.Energy.Tolerance is used in addition to DM.Tolerance, not instead of. In my calculations I have tried to decrease the DM.tolerance from the default 1e-4 to 0.0009 hoping that it would help me handle the system, in addition to other scf related tags as given below, to no avail. Below are the values for the last cycles of the scf iteration, before the program exits because the max.number of SCF iterations is reached. siesta: iscf Eharris(eV) E_KS(eV) FreeEng(eV) dDmax Ef(eV) siesta: 1486 -13262.5643 -13262.5643 -13262.5643 0.0271 -5.2142 siesta: 1487 -13262.5674 -13262.5673 -13262.5673 0.0272 -5.2149 siesta: 1488 -13262.5665 -13262.5665 -13262.5665 0.0272 -5.2140 siesta: 1489 -13262.5676 -13262.5675 -13262.5675 0.0272 -5.2150 siesta: 1490 -13262.5657 -13262.5658 -13262.5658 0.0270 -5.2144 siesta: 1491 -13262.5675 -13262.5675 -13262.5675 0.0273 -5.2159 siesta: 1492 -13262.5661 -13262.5661 -13262.5661 0.0270 -5.2141 siesta: 1493 -13262.5610 -13262.5611 -13262.5611 0.0268 -5.2125 siesta: 1494 -13262.5658 -13262.5656 -13262.5656 0.0272 -5.2137 siesta: 1495 -13262.5644 -13262.5644 -13262.5644 0.0271 -5.2134 siesta: 1496 -13262.5682 -13262.5681 -13262.5681 0.0271 -5.2159 siesta: 1497 -13262.5630 -13262.5631 -13262.5631 0.0273 -5.2108 siesta: 1498 -13262.5678 -13262.5676 -13262.5676 0.0269 -5.2157 siesta: 1499 -13262.5660 -13262.5660 -13262.5660 0.0273 -5.2152 siesta: E_KS(eV) = -13262.5660 SCF related tags in the input file: #Convergence of SCF- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations1500 # DM.MixSCF1 .true. DM.MixingWeight 0.0005 DM.NumberPulay 8 DM.UseSaveDM .true. # to use continuation files DM.Tolerance0.0009 #---for tough convergence--- DM.NumberKick 100 DM.KickMixingWeight0.01 #DM.Energy.Tolerance 10-4 eV #DM.Require.Energy.Convergence .false. Diana Ot??lvaro d.otalv...@utwente.nl Computational Material Science MESA+ Institute for Nanotechnology University of Twente Carre 4049 Postbus 217 NL-7500 AE Enschede tel: +31-53-489-2986
[SIESTA-L] Convergence: DM.Tolerance vs. DM.Energy.Tolerance
Hello everyone, I have a lot of problems converging the DM of a structure where I force antiparallel magnetization in the two Ni leads, and have a scattering region made of non-magnetic material. I force the AP magnetization by using %block DM.InitSpin. I wonder if the DM.Tolerance criterion is a stricter convergence criterion than DM.Energy.Tolerance. I suspect that if I were to use Energy convergence instead of DM, convergence it would be faster, since the energy is variational whereas the elements of DM are not. Am I correct? Even so, I don't know if I can bypass the DM.Tolerance tag. The manual tells me that DM.Energy.Tolerance is used in addition to DM.Tolerance, not instead of. In my calculations I have tried to decrease the DM.tolerance from the default 1e-4 to 0.0009 hoping that it would help me handle the system, in addition to other scf related tags as given below, to no avail. Below are the values for the last cycles of the scf iteration, before the program exits because the max.number of SCF iterations is reached. siesta: iscf Eharris(eV) E_KS(eV) FreeEng(eV) dDmax Ef(eV) siesta: 1486 -13262.5643 -13262.5643 -13262.5643 0.0271 -5.2142 siesta: 1487 -13262.5674 -13262.5673 -13262.5673 0.0272 -5.2149 siesta: 1488 -13262.5665 -13262.5665 -13262.5665 0.0272 -5.2140 siesta: 1489 -13262.5676 -13262.5675 -13262.5675 0.0272 -5.2150 siesta: 1490 -13262.5657 -13262.5658 -13262.5658 0.0270 -5.2144 siesta: 1491 -13262.5675 -13262.5675 -13262.5675 0.0273 -5.2159 siesta: 1492 -13262.5661 -13262.5661 -13262.5661 0.0270 -5.2141 siesta: 1493 -13262.5610 -13262.5611 -13262.5611 0.0268 -5.2125 siesta: 1494 -13262.5658 -13262.5656 -13262.5656 0.0272 -5.2137 siesta: 1495 -13262.5644 -13262.5644 -13262.5644 0.0271 -5.2134 siesta: 1496 -13262.5682 -13262.5681 -13262.5681 0.0271 -5.2159 siesta: 1497 -13262.5630 -13262.5631 -13262.5631 0.0273 -5.2108 siesta: 1498 -13262.5678 -13262.5676 -13262.5676 0.0269 -5.2157 siesta: 1499 -13262.5660 -13262.5660 -13262.5660 0.0273 -5.2152 siesta: E_KS(eV) = -13262.5660 SCF related tags in the input file: #Convergence of SCF- MeshCutoff 200.0 Ry # Default value PW cutoff for grid MaxSCFIterations1500 # DM.MixSCF1 .true. DM.MixingWeight 0.0005 DM.NumberPulay 8 DM.UseSaveDM .true. # to use continuation files DM.Tolerance0.0009 #---for tough convergence--- DM.NumberKick 100 DM.KickMixingWeight0.01 #DM.Energy.Tolerance10-4 eV #DM.Require.Energy.Convergence .false. Diana Otálvaro d.otalv...@utwente.nl Computational Material Science MESA+ Institute for Nanotechnology University of Twente Carre 4049 Postbus 217 NL-7500 AE Enschede tel: +31-53-489-2986
Re: [SIESTA-L] Convergence and dependance of parametres on Transiesta run
I am having the same problem. Could anybody please help us with these problem.. On Tue, Nov 8, 2011 at 5:27 AM, Arunkumar C R arunkumar...@gmail.comwrote: Dear Siesta users, I was working with a metallic system 2u-5u-2u where u stands for the unit cell consisting of N atoms. Many attempts have been made to make my transiesta run converged, but only siesta part does so. At this juncture, I would like to intervene your attention towards my problem. The parameters used for the job along with the output obtained are given below: Basis set used: SZP XC GGA PBE Smearing Order: MP , 1 DM.NumberPulay 6 DM.NumberBroyden0 DM.MixingWeight 0.050 DM.NumberKick 0 DM.KickMixingWeight 0.50 DM.Tolerance 1.00E-04 UseSaveData T MD.NumCGsteps 0 MD.TypeOfRun CG MD.VariableCell F MD.MaxCGDispl0.20 Bohr MD.MaxForceTol 0.05 eV/Ang Part of the output file: --- TRANSIESTA: Initializing lists From SIESTA: Efermi= -5.371 qc0 : 0.000 qcn0 : 392.013 nou,NGL,NGR: 2592 576 576 TranSiesta: Qsol,Qtot: 864.005 864.0001.000 transiesta: TSiscf Eharris(eV) E_KS(eV) FreeEng(eV) dDmax Ef(eV) transiesta: 1 -25275.6317 -25274.8408 -25274.8408 0.0185 -5.3710 timer: Routine,Calls,Time,% = IterSCF 22 21042.405 29.82 elaps: Routine,Calls,Wall,% = IterSCF 22 657.5811147.81 timer: Routine,Calls,Time,% = TS 1 48557.236 68.82 elaps: Routine,Calls,Wall,% = TS 1-630.071*** TranSiesta: Qsol,Qtot: 864.005 864.0001.000 transiesta: 2 -25275.6324 -25274.8409 -25274.8409 0.0185 -5.3710 TranSiesta: Qsol,Qtot: 864.006 864.0001.000 transiesta: 3 -25275.6410 -25274.8405 -25274.8405 0.0185 -5.3710 TranSiesta: Qsol,Qtot: 864.006 864.0001.000 transiesta: 4 -25275.6406 -25274.8406 -25274.8406 0.0185 -5.3710 TranSiesta: Qsol,Qtot: 864.006 864.0001.000 : : stops at transiesta: 7 The input parametres like DM.MixingWeight has been varied a number of times, also the temperature (with FD smearing) ; and the above case is the not so bad among my trials, which is wrong too. The major problem noticed is that dDmax is not decreasing and almost fixed and after few transiesta steps (within 8), the job stops. Also Qtot and Qsol differs by a small fraction, but never gets equal or reach convergence. Apart from many trials, I could not figure out the reason. Would some of you who has faced similar situation can consider my request. With thanks and regards arun Arunkumar C R
Re: [SIESTA-L] Convergence for large nanowires
Marcos, I changed versions and its now working fine. Thnaks, Pavan. div id=RTEContentdivGraduate Teaching Assistant/div div249,Physical Sciences,brOklahoma State Universitybr/div divStillwater,OK,USA-74075brPhone:4057445804br/div div /div div---/div divbr/ Help Ever Hurt Never/div divService to Humanity is Service to God/div div/div div---/div/div --- On Thu, 9/30/10, Marcos Veríssimo Alves marcos.verissimo.al...@gmail.com wrote: From: Marcos Veríssimo Alves marcos.verissimo.al...@gmail.com Subject: Re: [SIESTA-L] Convergence for large nanowires To: siesta-l@uam.es Date: Thursday, September 30, 2010, 11:08 AM Pavan, If I'm not mistaken there was a bug in the extrapolation of the DM in siesta 3.0-b - sometimes it would lead to such problems. However, I think this bug has been fixed in the latest release candidates. If you don't wish to change the version of Siesta, then you should disable the DM extrapolation. Marcos On Thu, Sep 30, 2010 at 5:41 PM, Pavan K. Pillalamarri topavanku...@yahoo.co.in wrote: Hi All, I am trying to perform calculations on ZnO nanowires. I have no problems with the smaller wires but the larger wires have SCF convergence issues. For larger wires (192 atom) I get too large free energy and dDmax values and SCF does not converge. I varied the DM.NumberPulay from 3-8 and the mixing weight from to about 0.01 and below but it did not help. I am attaching my output files. Would be glad if anyone could help. Thanks, Pavan. div id=RTEContentdivGraduate Teaching Assistant/div div249,Physical Sciences,brOklahoma State Universitybr/div divStillwater,OK,USA-74075brPhone:4057445804br/div div /div div---/div divbr/ Help Ever Hurt Never/div divService to Humanity is Service to God/div div/div div---/div/div
Re: [SIESTA-L] Convergence for large nanowires
Pavan, If I'm not mistaken there was a bug in the extrapolation of the DM in siesta 3.0-b - sometimes it would lead to such problems. However, I think this bug has been fixed in the latest release candidates. If you don't wish to change the version of Siesta, then you should disable the DM extrapolation. Marcos On Thu, Sep 30, 2010 at 5:41 PM, Pavan K. Pillalamarri topavanku...@yahoo.co.in wrote: Hi All, I am trying to perform calculations on ZnO nanowires. I have no problems with the smaller wires but the larger wires have SCF convergence issues. For larger wires (192 atom) I get too large free energy and dDmax values and SCF does not converge. I varied the DM.NumberPulay from 3-8 and the mixing weight from to about 0.01 and below but it did not help. I am attaching my output files. Would be glad if anyone could help. Thanks, Pavan. div id=RTEContentdivGraduate Teaching Assistant/div div249,Physical Sciences,brOklahoma State Universitybr/div divStillwater,OK,USA-74075brPhone:4057445804br/div div /div div---/div divbr/ Help Ever Hurt Never/div divService to Humanity is Service to God/div div/div div---/div/div
