[Wien] how to find Stoner parameter using wien2k
Dear wien2k users, According to the Stoner band theory of magnetism, one can able to predict the magnetic ground state of the material as it is paramagnetic or ferromagnetic using product of the non magnetic density of states around the Fermi level (N(E)) and the Stoner parameter (I). As given in some of the literature, the Stoner parameter can be described as the exchange integral and it could be found using LSDA and LMTO methods. see the fallowing literature... 1. http://prb.aps.org/pdf/PRB/v16/i1/p255_1 2. http://jap.aip.org/resource/1/japiau/v89/i11/p6889_s1 We are doing our calculations in Wein2K package using the GGA approximation and we want to know how the stoner parameter could be found according to the my calculations. As non magnetic calculations gives the total density of states at around the Fermi level, it is very helpful for us to predict the material is magnetic or not using the Stoner parameter. therefore, please suggest us how to find the Stoner parameter using wein2k package. -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/ecde5d8a/attachment.htm
[Wien] Would like to have some guidance from you
Dear professor Blaha and Wien2k users, Thanks to the Wien2k Workshop, now that I can do some proper calculation but since i am new to Wien2k and I would like to have some guidance in my current calculation. I am doing a supercell (3x3x1) calculation (for publishing quality) to a layered structure perovskite A2BO4 compound. It is of K2NiF2 structure (SG I4/mmm), its resistivity is at around 0.001 Ohm-m at room temp so it is semi-conductive. Half of the A atoms (A atom is Sr in this case, and the B atom is Co) are replace by an Rare Earth atom so the whole supercell is of 90 atoms in size in my struct file. The original unit cell parameters were a=b=3.8 Amstrong and c=12.3 Amstrong. I am using a PC with 2 core Xeon 2.4GHz, 4G RAM running latest Susie, latest Wien2k and ifort11. The calculation is Spin Polarized (and not an anti-ferromagnetic as I know that this compound is ferromagnetic), I used 200 k-points for the current calculation. The initialization seems working fine (thanks to professor Blaha's guidance in the workshop). Now it seems that the calculation is a bit too much for this computer - it runs painfully slow and after two days, it only just displays to me two lines: LAPW0 END and LAPW1 END. I am wondering: for this calculation, can I use less k-points say 50 (as it is a big cell) and the result would still be good enough for paper publishing? What kind of speed up do I expect if I am to use a 50 k-pints instead of 200 k-points? Would you think that a computer with i7 core and 16GB of RAM would be able to run this calculation without too much trouble (I am trying to convince my boss to purchase a new PC for me for WIEN2k)? Any comments/feedback would be greatly appreciated. Thank you for your time. kind regards, Qiwen ** Dr QiWen YAO JSPS Fellow Multifunctional Materials Group Optical and Electronic Materials Unit Environment and Energy Materials Research Division National Institute for Materials Science 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan Phone: +81-29-851-3354, ext. no. 6482, Fax: +81-29-859-2501 **
[Wien] Would like to have some guidance from you
Thank you very much for the detailed explanation again Professor Blaha. I will follow your suggestion and stop the current calculation, and starting from the simple cases with a lower RKmax. PS: the cpu usage for the lapw1c is mostly at around 100% and the memory at around 77% in 'top' In the case.output1dn(up) I can see that the CUP time in interstitial (total) 15.44(15.29), and the cpu time in dstart: 104.55(103.89) - are we talking about the wall time is the dstart time? (I can not see anywhere says 'wall-time' in both files)- so this means my computer does not have enough memory (for this calculation)? Thank you very much for your time. kind regards, Qiwen --Original Message-- From:Peter Blahapblaha at theochem.tuwien.ac.at To:A Mailing list for WIEN2k userswien at zeus.theochem.tuwien.ac.at Cc: Subject:Re: [Wien] Would like to have some guidance from you Date:07/07/2011 09:02:27 AM(+0200) As a beginner, you need to gain experience. So first start with pure A2BO4. Does it run properly, check timings,... Then create first a smaller supercell, eg. a 2x2x1 cell. Does it run ? timings (4 times as many atoms, -- without iterative diag. ~ 64 times as long cpu-time in lapw1. However, in the supercell you can use 4 times less k-points, s the total effort does NOT increase that much. For the large cell: use the top command to check the performance of lapw1 or lapw2. In particular you should see of it uses the cpu by nearly 100 (or 200 when OMP_NUM_THREAD=2) %, or it does not becuase it does not have enough memory and pages. Also check the timings listed in case.output1up/dn. Is the cpu-time and wall-time similar ? If wall-time is much larger, your computer does not have enough memory. k-points: You heard at the workshop about k-points: metal-insulator or small-large cells. most likely I would not start out with so many k-points, but I don't know the details of your system. Of course 50-kpoints will run 4 times faster than 200 k. Another hint: For large cells you do not want to start out with RKMAX=7 (the default). Start the calculation with RKMax=6 or even 5.5. Run to scf, it should be MUCH faster (10-100 times !!!), but then don't stop, but increase RKMax and compare eg. the forces on all atoms, magnetic moments, DOS, (... what ever you are interested in) to find a reasonable RKMAX. And finally: For these large cells, iterative diagonalization should be used. I guess the most important message is: Start out with smaller problems. Experiment with parameters like k-points, RKmax, -it to get experience. Of course, a modern PC (or 4 of them for k-parallel runs) will help. But checkout the real bottleneck (memory ?) before. Am 07.07.2011 02:58, schrieb Dr Qi Wen YAO: Dear professor Blaha and Wien2k users, Thanks to the Wien2k Workshop, now that I can do some proper calculation but since i am new to Wien2k and I would like to have some guidance in my current calculation. I am doing a supercell (3x3x1) calculation (for publishing quality) to a layered structure perovskite A2BO4 compound. It is of K2NiF2 structure (SG I4/mmm), its resistivity is at around 0.001 Ohm-m at room temp so it is semi-conductive. Half of the A atoms (A atom is Sr in this case, and the B atom is Co) are replace by an Rare Earth atom so the whole supercell is of 90 atoms in size in my struct file. The original unit cell parameters were a=b=3.8 Amstrong and c=12.3 Amstrong. I am using a PC with 2 core Xeon 2.4GHz, 4G RAM running latest Susie, latest Wien2k and ifort11. The calculation is Spin Polarized (and not an anti-ferromagnetic as I know that this compound is ferromagnetic), I used 200 k-points for the current calculation. The initialization seems working fine (thanks to professor Blaha's guidance in the workshop). Now it seems that the calculation is a bit too much for this computer - it runs painfully slow and after two days, it only just displays to me two lines: LAPW0 END and LAPW1 END. I am wondering: for this calculation, can I use less k-points say 50 (as it is a big cell) and the result would still be good enough for paper publishing? What kind of speed up do I expect if I am to use a 50 k-pints instead of 200 k-points? Would you think that a computer with i7 core and 16GB of RAM would be able to run this calculation without too much trouble (I am trying to convince my boss to purchase a new PC for me for WIEN2k)? Any comments/feedback would be greatly appreciated. Thank you for your time. kind regards, Qiwen ** Dr QiWen YAO JSPS Fellow Multifunctional Materials Group Optical and Electronic Materials Unit Environment and Energy Materials Research Division National Institute for Materials Science 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan Phone: +81-29-851-3354, ext. no. 6482, Fax: +81-29-859-2501
[Wien] compile time error during installation of wien2k 11 with the latest compilers
Dear Wien2k users, We are trying to install wien2k 11 in a Compaq Laptop having core2duo processor by using the latest compilers (ifort+mkl). The used OPTION is given below: FOPT:-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -traceback LDFLAGS:-L/home/avijitghosh/intel/composerxe-2011.4.191/mkl/lib/ia32 -pthread DPARALLEL:'-DParallel' R_LIBS:-lmkl_lapack95 -lmkl_intel -lmkl_intel_thread -lmkl_core -openmp -lpthread using these OPTIONS when we compile all the programs the following errors appeared: Compile time errors (if any) were: SRC_lapw0/compile.msg:W2kinit.F(28): error #5102: Cannot open include file ' mkl_vml.fi' SRC_lapw0/compile.msg:make[1]: *** [W2kinit.o] Error 1 SRC_lapw0/compile.msg:make: *** [seq] Error 2 We are not able to figure out why these errors appear as *mkl_vml.fi is in the include directory of mkl. *Another thing may be worthy to mention here that although we are using core2duo processor, during installation of ifort , the compiler choose the ia32 environment by itself. Any response in this regard will be very helpful for us. Thanks in advance. with regards, -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/94733752/attachment.htm
[Wien] Would like to have some guidance from you
PS: the cpu usage for the lapw1c is mostly at around 100% and the memory at around 77% in 'top' In the case.output1dn(up) I can see that the CUP time in interstitial (total) 15.44(15.29), and the cpu time in dstart: 104.55(103.89) - are we talking about the wall time is the dstart time? (I can not see anywhere says 'wall-time' in both files)- so this means my computer does not have enough memory (for this calculation)? in case.output1up there are lines like: TIME HAMILT (CPU) = 0.4, HNS = 0.5, HORB = 0.0, DIAG = 0.9 TIME HAMILT (WALL) = 0.4, HNS = 0.4, HORB = 0.0, DIAG = 0.8 and they should be similar You can thus also find detailed timing information for each individual k-point and clearly less k-points take less time -- P.Blaha -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-15671 FAX: +43-1-58801-15698 Email: blaha at theochem.tuwien.ac.atWWW: http://info.tuwien.ac.at/theochem/ --
[Wien] compile time error during installation of wien2k 11 with the latest compilers
Probably because you installed a 32-bit Linux version on that laptop ??? and do you have the recommended source compilervars.sh ... line in your .bashrc ?? Am 07.07.2011 11:56, schrieb shamik chakrabarti: Dear Wien2k users, We are trying to install wien2k 11 in a Compaq Laptop having core2duo processor by using the latest compilers (ifort+mkl). The used OPTION is given below: FOPT:-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -traceback LDFLAGS:-L/home/avijitghosh/intel/composerxe-2011.4.191/mkl/lib/ia32 -pthread DPARALLEL:'-DParallel' R_LIBS:-lmkl_lapack95 -lmkl_intel -lmkl_intel_thread -lmkl_core -openmp -lpthread using these OPTIONS when we compile all the programs the following errors appeared: Compile time errors (if any) were: SRC_lapw0/compile.msg:W2kinit.F(28): error #5102: Cannot open include file 'mkl_vml.fi http://mkl_vml.fi' SRC_lapw0/compile.msg:make[1]: *** [W2kinit.o] Error 1 SRC_lapw0/compile.msg:make: *** [seq] Error 2 We are not able to figure out why these errors appear as *mkl_vml.fi http://mkl_vml.fi is in the include directory of mkl. *Another thing may be worthy to mention here that although we are using core2duo processor, during installation of ifort , the compiler choose the ia32 environment by itself. Any response in this regard will be very helpful for us. Thanks in advance. with regards, -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- P.Blaha -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-15671 FAX: +43-1-58801-15698 Email: blaha at theochem.tuwien.ac.atWWW: http://info.tuwien.ac.at/theochem/ --
[Wien] Charge Convergence is not achieved
Dear Peter Blaha Sir, Indeed by increasing number of K points we got the convergence. Sir I have now some basic queries on this topic. You have said that sometimes you cannot reach (easily) arbitrary convergence why in some cases we can not reach convergence up to our desired limit?...is it the limitation of DFT?or it means that the feasibility of the solution is only up to the achieved convergence? Thanking you, with best regards, On Wed, Jul 6, 2011 at 6:13 PM, Peter Blaha pblaha at theochem.tuwien.ac.atwrote: For just 3 atoms/cell (and metallid ??) 1000 or 5000 k are still not much. Better k-mesh should improve convergence. And sometimes you cannot reach (easily) arbitrary convergence. What about E-tot, ... ? Am 06.07.2011 07:48, schrieb Shamik Chakrabarti: Dear wien2k users, We have done volume optimization of a structure having space group no. 225 (Fm3m) and 3 inequivalent atoms per unit cell. We have taken the least energy lattice parameters for spin polarized SCF calculations. However, the last calculation (spin polarized SCF) has not been converging at all. We have set the convergence criteria of charge to 0.0001and it reached up to 0.006. After that its is fluctuating sinusoidally at this value (even at around 400th iteration). We have also tried using the fallowing steps, such as 1. increasing the RmaxKmax value from 7 to 9 2. changing mixer values ranging from 0.01 to 0.3 (using MSEC mixing scheme...wien2k 10) 3. increasing the K-points from 1000 to 5000 but unfortunately we haven't got the desired convergence yet. Therefore, it is very helpful for us to have a suggestion for this problem. -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA __**_ Wien mailing list Wien at zeus.theochem.tuwien.ac.**at Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.**ac.at/mailman/listinfo/wienhttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- P.Blaha --**--** -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-15671 FAX: +43-1-58801-15698 Email: blaha at theochem.tuwien.ac.atWWW: http://info.tuwien.ac.at/** theochem/ http://info.tuwien.ac.at/theochem/ --**--** -- __**_ Wien mailing list Wien at zeus.theochem.tuwien.ac.**at Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.**ac.at/mailman/listinfo/wienhttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/e69d3bee/attachment.htm
[Wien] compile time error during installation of wien2k 11 with the latest compilers
Dear Peter Blaha Sir, Yes we have installed 32 bit linux version on that laptop. do you have the recommended source compilervars.sh ... line in your .bashrc ?? No Sir, we do not have such line in .bashrc.this file only contains the lines added by Xcrysden during its installation and also the lines: # Source global definitions if [ -f /etc/bashrc ]; then . /etc/bashrc fi #User specific aliases and functions with best regards, On Thu, Jul 7, 2011 at 3:38 PM, Peter Blaha pblaha at theochem.tuwien.ac.atwrote: Probably because you installed a 32-bit Linux version on that laptop ??? and do you have the recommended source compilervars.sh ... line in your .bashrc ?? Am 07.07.2011 11:56, schrieb shamik chakrabarti: Dear Wien2k users, We are trying to install wien2k 11 in a Compaq Laptop having core2duo processor by using the latest compilers (ifort+mkl). The used OPTION is given below: FOPT:-FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -traceback LDFLAGS:-L/home/avijitghosh/**intel/composerxe-2011.4.191/**mkl/lib/ia32 -pthread DPARALLEL:'-DParallel' R_LIBS:-lmkl_lapack95 -lmkl_intel -lmkl_intel_thread -lmkl_core -openmp -lpthread using these OPTIONS when we compile all the programs the following errors appeared: Compile time errors (if any) were: SRC_lapw0/compile.msg:W2kinit.**F(28): error #5102: Cannot open include file 'mkl_vml.fi http://mkl_vml.fi' SRC_lapw0/compile.msg:make[1]: *** [W2kinit.o] Error 1 SRC_lapw0/compile.msg:make: *** [seq] Error 2 We are not able to figure out why these errors appear as *mkl_vml.fi http://mkl_vml.fi is in the include directory of mkl. *Another thing may be worthy to mention here that although we are using core2duo processor, during installation of ifort , the compiler choose the ia32 environment by itself. Any response in this regard will be very helpful for us. Thanks in advance. with regards, -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA __**_ Wien mailing list Wien at zeus.theochem.tuwien.ac.**at Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.**ac.at/mailman/listinfo/wienhttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- P.Blaha --**--** -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-15671 FAX: +43-1-58801-15698 Email: blaha at theochem.tuwien.ac.atWWW: http://info.tuwien.ac.at/** theochem/ http://info.tuwien.ac.at/theochem/ --**--** -- __**_ Wien mailing list Wien at zeus.theochem.tuwien.ac.**at Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.**ac.at/mailman/listinfo/wienhttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/95ff7166/attachment.htm
[Wien] how to find Stoner parameter using wien2k
Hi Shamik, Sayed's advice is good. There are also some procedures based on Fixed spin moment calculations. You will have to think carefully about whether your system is metallic or locaised type. The links in the appendix to this paper should help: Ylvisaker and Pickett, PHYSICAL REVIEW B 79, 035103 2009 Best, David. 2011/7/6 Seyed Javad Hashemifar hashemifar at cc.iut.ac.ir I have no experience on calculation of stoner parameter and interesting to learn it. However, by doing a simple spin polarized calculations with nonzero and parallel initial moments on your system, you may easily find whether your system prefers ferromagnetism or not. The initial magnetic moments are controlled in case.inst file. SJ Hashemifar == Seyed Javad Hashemifar Physics Department, Isfahan University of Technology 84156-83111 Isfahan, Iran Tel: +98 311 391 2375 Fax:+98 311 3912376 Email: hashemifar at cc.iut.ac.ir Homepage: http://hashemifar.iut.ac.ir --- 2011/7/6 Shamik Chakrabarti shamikiitkgp at gmail.com Dear wien2k users, According to the Stoner band theory of magnetism, one can able to predict the magnetic ground state of the material as it is paramagnetic or ferromagnetic using product of the non magnetic density of states around the Fermi level (N(E)) and the Stoner parameter (I). As given in some of the literature, the Stoner parameter can be described as the exchange integral and it could be found using LSDA and LMTO methods. see the fallowing literature... 1. http://prb.aps.org/pdf/PRB/v16/i1/p255_1 2. http://jap.aip.org/resource/1/japiau/v89/i11/p6889_s1 We are doing our calculations in Wein2K package using the GGA approximation and we want to know how the stoner parameter could be found according to the my calculations. As non magnetic calculations gives the total density of states at around the Fermi level, it is very helpful for us to predict the material is magnetic or not using the Stoner parameter. therefore, please suggest us how to find the Stoner parameter using wein2k package. -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/85a8296c/attachment.htm
[Wien] Charge Convergence is not achieved
2011/7/7 Shamik Chakrabarti shamikiitkgp at gmail.com: Dear Peter Blaha Sir, ?? ? ? ? ? ? ? ? ? ? ? Indeed by increasing number of K points we got the convergence. Sir I have now some basic queries on this topic. You have said that ?? ? ? ? ? ? ? ? ? ?sometimes you cannot reach (easily) arbitrary convergence why in some cases we can not reach convergence up to our desired limit?...is it the limitation of DFT?or it means that the feasibility of the solution is only up to the achieved convergence? This is in fact a deep, and very good question, at least in my opinion. Unfortunately that does not mean that there is a good answer to it! With the perfect functional convergence should (I believe, others may disagree) always be good. With a very imperfect functional it is quite possible that a DFT calculation will not converge, i.e. it is unfeasible. Empirically many (but not all) metals do not converge well with small numbers of k-points, but some others do. WhyI do not understand as I cannot write down a mathematical analysis to explain this and do not believe that there is a formal analysis in the literature, it is just empirical knowledge (folklore). -- Laurence Marks Department of Materials Science and Engineering MSE Rm 2036 Cook Hall 2220 N Campus Drive Northwestern University Evanston, IL 60208, USA Tel: (847) 491-3996 Fax: (847) 491-7820 email: L-marks at northwestern dot edu Web: www.numis.northwestern.edu Chair, Commission on Electron Crystallography of IUCR www.numis.northwestern.edu/ Research is to see what everybody else has seen, and to think what nobody else has thought Albert Szent-Gyorgi
[Wien] Charge Convergence is not achieved
Dear Laurence Marks Sir, Thank you very much for your reply.yes the question may not have a very good answerif in any calculation we are not getting the desired convergence, it may happen that: (1) Our chosen functional is not appropriate for the system (2) The system (say while trying to predict a new material!) may not be feasible at all.. Probably all we can say that if we are able to achieve desired convergence (say 0.0001) we can say that we have used the appropriate functional for the system and the system is (may be) feasible (at least theoretically!! ). Sir please correct me if I am wrong in my concept. with best regards, On Thu, Jul 7, 2011 at 6:12 PM, Laurence Marks L-marks at northwestern.eduwrote: 2011/7/7 Shamik Chakrabarti shamikiitkgp at gmail.com: Dear Peter Blaha Sir, Indeed by increasing number of K points we got the convergence. Sir I have now some basic queries on this topic. You have said that sometimes you cannot reach (easily) arbitrary convergence why in some cases we can not reach convergence up to our desired limit?...is it the limitation of DFT?or it means that the feasibility of the solution is only up to the achieved convergence? This is in fact a deep, and very good question, at least in my opinion. Unfortunately that does not mean that there is a good answer to it! With the perfect functional convergence should (I believe, others may disagree) always be good. With a very imperfect functional it is quite possible that a DFT calculation will not converge, i.e. it is unfeasible. Empirically many (but not all) metals do not converge well with small numbers of k-points, but some others do. WhyI do not understand as I cannot write down a mathematical analysis to explain this and do not believe that there is a formal analysis in the literature, it is just empirical knowledge (folklore). -- Laurence Marks Department of Materials Science and Engineering MSE Rm 2036 Cook Hall 2220 N Campus Drive Northwestern University Evanston, IL 60208, USA Tel: (847) 491-3996 Fax: (847) 491-7820 email: L-marks at northwestern dot edu Web: www.numis.northwestern.edu Chair, Commission on Electron Crystallography of IUCR www.numis.northwestern.edu/ Research is to see what everybody else has seen, and to think what nobody else has thought Albert Szent-Gyorgi ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/ed69a138/attachment.htm
[Wien] Charge Convergence is not achieved
Basically, for a metal the convergence depends on the details of the bandstructure around EF and on the method to determine EF and the occupation of all eigenvalues. Suppose you have two bands crossing EF, one has A character, the other one B. Now you start with a coarse k-mesh and represent the band with only a few k-points, such that the weight (number of electrons) for each eigenvalue E_n_k is large (e.g 0.1 e) At some iteration it can happen that E_n1_k1 is just a tiny little bit lower than E_n2_k2 (k1 and k2 come from different bands) and both are close to EF. Than E_n1_k1 is fully occupied, while E_n2_k2 is completely empty when using the TETRA method (because this interpolates only within the same band n!) and thus you get more charge at atom A. Even when the mixer now adds only very little of this new density, it may lead to a potential where E_n_k1 is now HIGHER than E_n_k2, and thus in the next iteration we get a density which has 0.1 e more at site B (and not A). Thus the newly generated charge densities differ by a huge (0.1 e) amount from the previous one. If you now increase the k-mesh, the weight of an individual k-point will go down (eg. be only 0.01 e) and thus such oszillations will be an order of magnitude smaller. In addition, an integration (TETRAHEDRON method) becomes better with more sample points and convergence will be better. On the other hand when using TEMP(S) instead of TETRA, you may be able to damp these oszillations, since the occupation depends only on the energy, but not on the topology of the bands (i.e. which eigenvalues are connected to each other via band n and k-index k). This is a clear advantage of TEMP, however, you run into the problem that a final solution eventually has ALWAYS some occupation of unoccupied states, which should be zero for an exact method (and you may even loose or greatly reduce your magnetic moment). Basically, there is no absolute rule and convergence has to be checked for each individual case because you do not know the band-details. Of coarse there are general considerations like: bad - good convergence metal - nonmetal flat bands- steep bands at EF, or equivalently elements with f,d-states at EF - no d,f states at EF many non-equivalent atoms of the same type -onyl ONE equivalent atom on nuclear charge Z Some examples derived from those rules: fcc Cu converges very quick (only ONE very STEEP S-like band at EF), bcc V is more difficult (MANY D-BANDS cross EF). fcc Ni is even worse, because of spin polarization you DOUBLE the number of bands at EF and one can easily shuffle electrons from spin-up to dn,... A supercell or surface of Ni becomes even worse, because you may have several different Ni atoms (surface, sub-surface, bulk) and thus have with X-layers X-TIMES as many bands around EF, all of them VERY SIMILAR (because they are all Ni), but still clearly distinct (surface,). Am 07.07.2011 14:42, schrieb Laurence Marks: 2011/7/7 Shamik Chakrabartishamikiitkgp at gmail.com: Dear Peter Blaha Sir, Indeed by increasing number of K points we got the convergence. Sir I have now some basic queries on this topic. You have said that sometimes you cannot reach (easily) arbitrary convergence why in some cases we can not reach convergence up to our desired limit?...is it the limitation of DFT?or it means that the feasibility of the solution is only up to the achieved convergence? This is in fact a deep, and very good question, at least in my opinion. Unfortunately that does not mean that there is a good answer to it! With the perfect functional convergence should (I believe, others maydisagree) always be good. With a very imperfect functional it is quitepossible that a DFT calculation will not converge, i.e. it isunfeasible. Empirically many (but not all) metals do not converge wellwith small numbers of k-points, but some others do. WhyI do notunderstand as I cannot write down a mathematical analysis to explainthis and do not believe that there is a formal analysis in theliterature, it is just empirical knowledge (folklore). -- Laurence MarksDepartment of Materials Science and EngineeringMSE Rm 2036 Cook Hall2220 N Campus DriveNorthwestern UniversityEvanston, IL 60208, USATel: (847) 491-3996 Fax: (847) 491-7820email: L-marks at northwestern dot eduWeb: www.numis.northwestern.eduChair, Commission on Electron Crystallography of IUCRwww.numis.northwestern.edu/Research is to see what everybody else has seen, and to think whatnobody else has thoughtAlbert Szent-Gyorgi___Wien mailing listWien at zeus.theochem.tuwien.ac.athttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- P.Blaha
[Wien] Charge Convergence is not achieved
The problem of functional could for instance happen for 4f-compounds. In LDA/GGA all 4f bands will be around EF (which is physically wrong) and convergence is naturally very different (14 extremely FLAT bands/atom !!). Using open core or LDA+U or Hybrid-DFT you remove the failure of GGA and immediately also convergence should improve (since you have only a FEW WIDE s,d-bands at EF). Am 07.07.2011 15:05, schrieb shamik chakrabarti: Dear Laurence Marks Sir, Thank you very much for your reply.yes the question may not have a very good answerif in any calculation we are not getting the desired convergence, it may happen that: (1) Our chosen functional is not appropriate for the system (2) The system (say while trying to predict a new material!) may not be feasible at all.. Probably all we can say that if we are able to achieve desired convergence (say 0.0001) we can say that we have used the appropriate functional for the system and the system is (may be) feasible (at least theoretically!! ). Sir please correct me if I am wrong in my concept. with best regards, On Thu, Jul 7, 2011 at 6:12 PM, Laurence Marks L-marks at northwestern.edu mailto:L-marks at northwestern.edu wrote: 2011/7/7 Shamik Chakrabarti shamikiitkgp at gmail.com mailto:shamikiitkgp at gmail.com: Dear Peter Blaha Sir, Indeed by increasing number of K points we got the convergence. Sir I have now some basic queries on this topic. You have said that sometimes you cannot reach (easily) arbitrary convergence why in some cases we can not reach convergence up to our desired limit?...is it the limitation of DFT?or it means that the feasibility of the solution is only up to the achieved convergence? This is in fact a deep, and very good question, at least in my opinion. Unfortunately that does not mean that there is a good answer to it! With the perfect functional convergence should (I believe, others may disagree) always be good. With a very imperfect functional it is quite possible that a DFT calculation will not converge, i.e. it is unfeasible. Empirically many (but not all) metals do not converge well with small numbers of k-points, but some others do. WhyI do not understand as I cannot write down a mathematical analysis to explain this and do not believe that there is a formal analysis in the literature, it is just empirical knowledge (folklore). -- Laurence Marks Department of Materials Science and Engineering MSE Rm 2036 Cook Hall 2220 N Campus Drive Northwestern University Evanston, IL 60208, USA Tel: (847) 491-3996 Fax: (847) 491-7820 email: L-marks at northwestern dot edu Web: www.numis.northwestern.edu http://www.numis.northwestern.edu Chair, Commission on Electron Crystallography of IUCR www.numis.northwestern.edu/ http://www.numis.northwestern.edu/ Research is to see what everybody else has seen, and to think what nobody else has thought Albert Szent-Gyorgi ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at mailto:Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- P.Blaha -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-15671 FAX: +43-1-58801-15698 Email: blaha at theochem.tuwien.ac.atWWW: http://info.tuwien.ac.at/theochem/ --
[Wien] Charge Convergence is not achieved
://zeus.theochem.tuwien.**ac.at/mailman/listinfo/wienhttp://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien -- Shamik Chakrabarti Research Scholar Dept. of Physics Meteorology Material Processing Solid State Ionics Lab IIT Kharagpur Kharagpur 721302 INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20110707/3d89e0a2/attachment.htm
[Wien] Charge Convergence is not achieved
While what you say is probably right (it normally is), for this to prevent convergence one has to hypothesize that these occupancy changes correspond to excitation of different eigenvectors of the charge density (not just the wavefunctions) with respect to the appropriate dielectric response matrix, so near the solution the mixing Jacobian only obtains information about the off-diagonal terms, minimal information about the diagonal terms. (In effect the Jacobian then become very ill-conditioned.) In effect this implies that in hard cases this response matrix varies rapidly with k so needs to be finely sampled to include all the different eigenvectors and avoid ill-conditioning. While this is a very reasonable hypothesis, proving it is not so simple N.B., with respect to the comment by Shamik Chakrabarti, yes, if the mixing has converged then this is a sufficient condition that you have satisifed the KS equations for the functional used, i.e. found a variational minimum and the problem is feasible. If the problem does not converge it may be unfeasible, and all that is being found is a trap not a fixed-point variational minimum. Or it is too hard for the mixer. On Thu, Jul 7, 2011 at 8:24 AM, Peter Blaha pblaha at theochem.tuwien.ac.at wrote: Basically, for a metal the convergence depends on the details of the bandstructure around EF and on the method to determine EF and the occupation of all eigenvalues. Suppose you have two bands crossing EF, one has A character, the other one B. Now you start with a coarse k-mesh and represent the band with only a few k-points, such that the weight (number of electrons) for each eigenvalue E_n_k is large (e.g 0.1 e) At some iteration it can happen that E_n1_k1 is just a tiny little bit lower than E_n2_k2 (k1 and k2 come from different bands) and both are close to EF. Than E_n1_k1 is fully occupied, while E_n2_k2 is completely empty when using the TETRA method (because this interpolates only within the same band n!) and thus you get more charge at atom A. Even when the mixer now adds only very little of this new density, it may lead to a potential where E_n_k1 is now HIGHER than E_n_k2, and thus in the next iteration we get a density which has 0.1 e more at site B (and not A). Thus the newly generated charge densities differ by a huge (0.1 e) amount from the previous one. If you now increase the k-mesh, the weight of an individual k-point will go down (eg. be only 0.01 e) and thus such oszillations will be an order of magnitude smaller. In addition, an integration (TETRAHEDRON method) becomes better with more sample points and convergence will be better. On the other hand when using TEMP(S) instead of TETRA, you may be able to damp these oszillations, since the occupation depends only on the energy, but not on the topology of the bands (i.e. which eigenvalues are connected to each other via band n and k-index k). This is a clear advantage of TEMP, however, you run into the problem that a final solution eventually has ALWAYS some occupation of unoccupied states, which should be zero for an exact method (and you may even loose or greatly reduce your magnetic moment). Basically, there is no absolute rule and convergence has to be checked for each individual case because you do not know the band-details. Of coarse there are general considerations like: bad ? ? ? ? ? - ? ? ? ? ? good convergence metal ? ? ? ? - ? ? ? ? ? nonmetal flat bands ? ?- ? ? ? ? ? steep bands ? at EF, or equivalently elements with f,d-states at EF - ? ? ? ? no d,f states at EF many non-equivalent atoms of the same type - ? ?onyl ONE equivalent atom on nuclear charge Z Some examples derived from those rules: fcc Cu converges very quick (only ONE very STEEP S-like band at EF), bcc V is more difficult (MANY D-BANDS cross EF). fcc Ni is even worse, because of spin polarization you DOUBLE the number of bands at EF and one can easily shuffle electrons from spin-up to dn,... A supercell or surface of Ni becomes even worse, because you may have several different Ni atoms (surface, sub-surface, bulk) and thus have with X-layers X-TIMES as many bands around EF, all of them VERY SIMILAR (because they are all Ni), but still clearly distinct (surface,). Am 07.07.2011 14:42, schrieb Laurence Marks: 2011/7/7 Shamik Chakrabartishamikiitkgp at gmail.com: ?Dear Peter Blaha Sir, ? ? ? ? ? ? ? ? ? ? ? ? ?Indeed by increasing number of K points we got the ?convergence. Sir I have now some basic queries on this topic. You have said ?that ? ? ? ? ? ? ? ? ? ? ? sometimes you cannot reach (easily) arbitrary ?convergence ?why in some cases we can not reach convergence up to our desired limit?...is ?it the limitation of DFT?or it means that the feasibility of the ?solution is only up to the achieved convergence? This is in fact a deep, and very good question, at least in my opinion. Unfortunately that does not
[Wien] Charge Convergence is not achieved
In addition to what Peter said (use more k-points and/or TEMP or TEMPS, perhaps just 0.0018 or even 0.001 for the later), in principle it might help a little to: a) Reduce the Greed (mixing factor) to 0.1 b) Increase the number of memory steps (nuse) to 16 (the code will not let you go too high) c) At least in version 11.1 with MSEC3 you can increase the regularization, for instance with DIAG XXX where XXX can be increased to 1E-3 or perhaps even 1e-2 (but please be careful). On Thu, Jul 7, 2011 at 9:24 AM, Laurence Marks L-marks at northwestern.edu wrote: On Thu, Jul 7, 2011 at 8:24 AM, Peter Blaha pblaha at theochem.tuwien.ac.at wrote: Basically, for a metal the convergence depends on the details of the bandstructure around EF and on the method to determine EF and the occupation of all eigenvalues. Suppose you have two bands crossing EF, one has A character, the other one B. Now you start with a coarse k-mesh and represent the band with only a few k-points, such that the weight (number of electrons) for each eigenvalue E_n_k is large (e.g 0.1 e) At some iteration it can happen that E_n1_k1 is just a tiny little bit lower than E_n2_k2 (k1 and k2 come from different bands) and both are close to EF. Than E_n1_k1 is fully occupied, while E_n2_k2 is completely empty when using the TETRA method (because this interpolates only within the same band n!) and thus you get more charge at atom A. Even when the mixer now adds only very little of this new density, it may lead to a potential where E_n_k1 is now HIGHER than E_n_k2, and thus in the next iteration we get a density which has 0.1 e more at site B (and not A). Thus the newly generated charge densities differ by a huge (0.1 e) amount from the previous one. If you now increase the k-mesh, the weight of an individual k-point will go down (eg. be only 0.01 e) and thus such oszillations will be an order of magnitude smaller. In addition, an integration (TETRAHEDRON method) becomes better with more sample points and convergence will be better. On the other hand when using TEMP(S) instead of TETRA, you may be able to damp these oszillations, since the occupation depends only on the energy, but not on the topology of the bands (i.e. which eigenvalues are connected to each other via band n and k-index k). This is a clear advantage of TEMP, however, you run into the problem that a final solution eventually has ALWAYS some occupation of unoccupied states, which should be zero for an exact method (and you may even loose or greatly reduce your magnetic moment). Basically, there is no absolute rule and convergence has to be checked for each individual case because you do not know the band-details. Of coarse there are general considerations like: bad ? ? ? ? ? - ? ? ? ? ? good convergence metal ? ? ? ? - ? ? ? ? ? nonmetal flat bands ? ?- ? ? ? ? ? steep bands ? at EF, or equivalently elements with f,d-states at EF - ? ? ? ? no d,f states at EF many non-equivalent atoms of the same type - ? ?onyl ONE equivalent atom on nuclear charge Z Some examples derived from those rules: fcc Cu converges very quick (only ONE very STEEP S-like band at EF), bcc V is more difficult (MANY D-BANDS cross EF). fcc Ni is even worse, because of spin polarization you DOUBLE the number of bands at EF and one can easily shuffle electrons from spin-up to dn,... A supercell or surface of Ni becomes even worse, because you may have several different Ni atoms (surface, sub-surface, bulk) and thus have with X-layers X-TIMES as many bands around EF, all of them VERY SIMILAR (because they are all Ni), but still clearly distinct (surface,). Am 07.07.2011 14:42, schrieb Laurence Marks: 2011/7/7 Shamik Chakrabartishamikiitkgp at gmail.com: ?Dear Peter Blaha Sir, ? ? ? ? ? ? ? ? ? ? ? ? ?Indeed by increasing number of K points we got the ?convergence. Sir I have now some basic queries on this topic. You have said ?that ? ? ? ? ? ? ? ? ? ? ? sometimes you cannot reach (easily) arbitrary ?convergence ?why in some cases we can not reach convergence up to our desired limit?...is ?it the limitation of DFT?or it means that the feasibility of the ?solution is only up to the achieved convergence? This is in fact a deep, and very good question, at least in my opinion. Unfortunately that does not mean that there is a good answer to it! With the perfect functional convergence should (I believe, others maydisagree) always be good. With a very imperfect functional it is quitepossible that a DFT calculation will not converge, i.e. it isunfeasible. Empirically many (but not all) metals do not converge wellwith small numbers of k-points, but some others do. WhyI do notunderstand as I cannot write down a mathematical analysis to explainthis and do not believe that there is a formal analysis in theliterature, it is just empirical knowledge (folklore). -- Laurence MarksDepartment of Materials