[Wannier] School on "Wannier90 v3.0: new features and applications", 25-27 March 2020, Oxford UK
Dear Wannier list members, In case you did not receive the following announcement via the Psi_k list earlier in the month, please see details of the upcoming workshop. Note the deadline is very soon - if this is problematic please email wannier90.works...@gmail.com Jonathan We are pleased to announce a 3-day school "Wannier90 v3.0: new features and applications", which will be held at St. Edmund Hall, University of Oxford, UK, 25-27 March 2020. In this three-day school, we will present the state-of-the-art functionalities of the Wannier90 code, with a particular focus on the latest developments. Each day will involve lectures and hands-on tutorials. The registration fee, which includes three nights' accommodation (with breakfast) and a social dinner, is £120. For a preliminary schedule and how to register, please see http://www.wannier.org/events/school-oxford-2020/ The registration deadline is end of the day on Monday 3rd Feb On behalf of the Organising Committee (Arash Mostofi, Valerio Vitale, Giovanni Pizzi, Ivo Souza and Jonathan Yates) ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
Re: [Wannier] Question on " Unable to satisfy B1 with any of the first 12 shells"
Dear Houchen, Your unit cell has a long a_2 cell vector, and you are sampling the BZ very finely in this direction. This means that the k-points in this direction are very very close together. So the early shells of k-point are mostly in that direction. In fact we have to go out to the 578th k-point shell to find k-point corresponding to the a_3 direction. You might want to consider if this k-point sampling is necessary for this system. So to make your calculation “run” the only default parameter you need to change is to set Search_shells=600 Note that if you set iprint=3 in the output file you will get more information on the search for the b vectors, which is useful for understanding issues such as this. I have copied the output of W90 using Search_shells=600 below Yours Jonathan ++ |Shell # Nearest-Neighbours| |- | | 1 2 | | 3 2 | | 4 2 | |578 2 | |579 2 | |586 2 | ++ | Completeness relation is fully satisfied [Eq. (B1), PRB 56, 12847 (1997)] | ++ | b_k Vectors (Ang^-1) and Weights (Ang^2) | | | |No. b_k(x) b_k(y) b_k(z)w_b | |--- | | 1 0.000.027886 -0.002218 637.399983| | 2 0.00 -0.0278860.002218 637.399983| | 3 0.074481 -0.001091 -0.01959882.840923| | 4-0.0744810.0010910.01959882.840923| | 5 0.074481 -0.028977 -0.017380 1.416631| | 6-0.0744810.0289770.017380 1.416631| | 7 0.2234420.0803840.856883 0.330359| | 8-0.223442 -0.080384 -0.856883 0.330359| | 9 0.2234420.0524990.859101 0.276083| |10-0.223442 -0.052499 -0.859101 0.276083| |11 0.2979220.0792940.837285 0.026087| |12-0.297922 -0.079294 -0.837285 0.026087| ++ | b_k Directions (Ang^-1) | | --- | |No. x y z | |--- | | 1 0.000.027886 -0.002218 | | 2 0.074481 -0.001091 -0.019598 | | 3 0.074481 -0.028977 -0.017380 | | 4 0.2234420.0803840.856883 | | 5 0.2234420.0524990.859101 | | 6 0.2979220.0792940.837285 | ++ > On 11 Oct 2020, at 11:29, Li, Houchen wrote: > > Dear all, > > I am trying to use VASP&wannier90 to calculate monolayer beta-RhBi2 band > structure. > Then I have the problem with "unable to satisfy Bi with any of xx shells" > error. While the bulk band structure works well. > > I looked for possible solutions, and I have tried to change the parameter > "k_mesh_tol" and "search_shell" both to much bigger values. It still does not > work. > > Some related codes are attached. Many thanks for potential solutions. > > Houchen > > > ___ > Wannier mailing list > Wannier@lists.quantum-espresso.org > https://lists.quantum-espresso.org/mailman/listinfo/wannier ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
Re: [Wannier] Post90w.x Allocation Error?
Dear Stephen, One way to trigger this error is if your calculation is trying to use more RAM than is available on your machine. An AHC calculation will need more RAM than a simple band-structure - which could explain why one calculation ran and the other failed. Interpolation needs matrices which are of the order of num_wann x num_wann x num_kpoints. These are not distributed - and so they will add up. You have num_wann=110 and num_kpoints=10^3, both reasonably large. Perhaps you could try a 8x8x8 grid as this should roughly half the memory requirement. If it runs monitor the job to see if it confirms the large memory hypothesis. If you need the accuracy of a 10^3 grid, then you would need to find a machine with more RAM. btw - your second email contained most of the info I needed to suggest a fix. However, in future it would help if you said which version of Wannier90 you are running. Jonathan > On 27 Dec 2020, at 17:39, Stephen Zhang wrote: > > Hi everyone, > > I'm currently trying to plot the k-space berry curvature for Co3Sn2S2 using > Wannier90. When doing so, I've encountered the error below, > > "Error in allocating crdist_ws in ws_translate_dist" > > Looking at the source code, I'm not exactly sure what triggers this error, or > how to fix it. If anyone could provide me with some hint or ideas on what > might cause this error to appear, I would really appreciate it. > > Thanks! > Stephen > -- > University of California, Berkeley > Department of Letter and Sciences > ___ > Wannier mailing list > Wannier@lists.quantum-espresso.org > https://lists.quantum-espresso.org/mailman/listinfo/wannier ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
Re: [Wannier] Error in Wannier90: kmesh_get: something wrong, found too many nearest neighbours
Dear Soumyadeep, We’d need to see your win file in order to reproduce this. Usually this is caused by an inaccurate cell file. So I think you should check that your cell is correct before taking this further. Converting it to abc notion gives me: a = 7.047207 alpha = 73.598016 b = 7.055220 beta = 73.820822 c = 3.955864 gamma = 32.581161 spglib thinks your input cell has 8 symmetry operations (Fmmm). I think the usual structure of BaFe2As2 is I4/mmm with 16 operations. If your cell has accidentally broken the symmetry that can cause issues for the k-mesh routine. Jonathan > On 17 Apr 2021, at 05:24, Soumyadeep wrote: > > Dear All, > I am facing following error in a Wannier-90 calculation, > running wannier90 -pp for Ba22... kmesh_get: something wrong, found too > many nearest neighbours > Error on node 0: examine the output/error files for details > Error condition encountered during test: exit status = 1 > Aborting > > As per previous discussions in the mailing list I have changed the > parameter "kmesh_tol" from default (10xdefault, 100xdefault, > 0.1xdefault,0.01xdefault). But all of these efforts leads to same error > message. I have also changed the DFT kmesh from 16x16x16 to 16x16x6, but no > improvement. I have attached all the necessary files with the mail. > > Please help me overcome this error. > > with many thanks and best regards > Soumyadeep > --- > Soumyadeep Ghosh, > Senior Research Fellow, > Homi Bhabha National Institute (HBNI), > Raja Ramanna Centre for Advanced Technology, Indore, India-452013 > Mob: (+91)9424664553 > User Lab: 0731244-2580 > Email: soumyadeepghos...@gmail.com, soumyad...@rrcat.gov.in > ---___ > Wannier mailing list > Wannier@lists.quantum-espresso.org > https://lists.quantum-espresso.org/mailman/listinfo/wannier ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
Re: [Wannier] Errors when using wannier90 interface with VASP
Dear Guotao, Did you try what the error message from wannier90 suggested, and increase search_shells? Depending on the dimensions of your until cell you might need to make it fairly large. Note that in the most recent release the default value is 36, not 12 (as in version 2) If that doesn’t fix the issue you should post your input file. Jonathan > On 6 Oct 2021, at 07:39, 邱 国涛 wrote: > > Dear wannier-lists, > > I am currently using VASP with wannier90 to build wannier function for a > nanoribbon structures. However, some errors occurs during the process. > > At the beginning I used K-mesh equals to 1 8 1 and the *.wout file shows the > error like: > > Unable to satisfy B1 with any of the first 12 shells > Your cell might be very long, or you may have an irregular MP grid > Try increasing the parameter search_shells in the win file (default=12) > > I googled it and someone told me to set kmesh_tol=0.1 but the error exist > still. > > Then I try to decrease my k-mesh to 1 5 1 and the process has successfully > carried out but with very poor converge quality like this: > > WF centre and spread 126 ( -4.798235,-12.074878, 8.737732 ) 242.32368440 > WF centre and spread 127 ( -4.799080,-12.078331, 8.738805 ) 242.43889057 > WF centre and spread 128 ( -4.798520,-12.067943, 8.739090 ) 242.40374260 > WF centre and spread 129 ( -4.799015,-12.078600, 8.738088 ) 242.37103635 > WF centre and spread 130 ( -4.798828,-12.068568, 8.738568 ) 242.40592147 > > Can someone tell me how to fix these problems and build the wannier function > with good quality? > > Thank you very much. > > Best, > Guotao Qiu > ___ > Wannier mailing list > Wannier@lists.quantum-espresso.org > https://lists.quantum-espresso.org/mailman/listinfo/wannier ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
Re: [Wannier] Discrepancy between Wannier90 and DFT band structures
Hi, I’ve had a quick look at your files. I can’t say for sure what is wrong with your calculation. But let me make an observation that might help. Wannier band structures are an interpolation. i.e. when there is no disentanglement (an isolated manifold of bands) at any kpoint that is part of the mesh used to construct the WF, the Wannier interpolated eigenvalues will agree exactly with the original DFT calculation. The Gamma point should always be part of the mesh used to construct the WF, so the band-structures at Gamma should match. For the set of 17 WF: Am I right that the DFT bands form an isolated manifold, so no disentanglement is need? If that is true the red and black bands should match at Gamma. They clearly don’t. You need to understand this point before moving on to the set of 5 WF. [the more I look at NCP17.png I have the feeling that there might be nothing “wrong” with your WF, but that the DFT groundstate used to compute the WF and that used to compute the bands are different in someway e.g. xc functional - just a wild guess..] Jonathan > On 14 Apr 2022, at 15:48, Azin Kazemi Moridani > wrote: > > Dear All, > > I am trying to generate a Wannier90 tight-binding model for Sr2FeO4 using > Wannier90 interfaced with ABINIT > . The > overall > behavior > of the bands produced by Wannier90 and ABINIT agree with one another, but > there are some discrepancies almost everywhere, especially > around the Fermi energy. I have tried to investigate the effect of varying > different parameters, such as : size of the KPOINT mesh, number of > iterations, energy window (with and without a frozen window, excluding > bands,..), and the set of projections (either > only on Iron d-orbitals, or on Iron d-orbitals plus Oxygen p-orbitals). In > the disentangled case of Iron d-orbitals plus Oxygen p-orbitals, the mapping > is expected to be exactly on top of the DFT band structure. I performed the > calculations with two > > pseudopotentials, > > Norm-Conserving (NC) and > Projector-Augmented Wave (PAW) > . > > In all the cases, convergence is > > achieved > but the disagreement persists. > > > I would appreciate it a lot if anyone could help me improve > my results. The > “wannier90.win" and "wannier90.wout” > files > > alongside the band structures in the case of > > five Iron d-orbitals > > (NC5.PNG) and > five > Iron d-orbitals plus > > twelve (4 * 3) Oxygen p-orbitals (NC17.PNG) can be found here: > https://drive.google.com/drive/folders/1WDjjKl22XXsIW7sdPIbWEsz29ZnkMTGO?usp=sharing > > > Thank you so much in advance. > > > Best Regards, > Mohaddeseh Kazemi-Moridani > Department of Physics > Université de Montréal > > > ___ > Wannier mailing list > Wannier@lists.quantum-espresso.org > https://lists.quantum-espresso.org/mailman/listinfo/wannier ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
Re: [Wannier] Trouble with kmesh/ bvectors
Dear Payden, Your unit cell deviates very slightly from cubic symmetry. You have two choices: 1- Impose cubic symmetry (e.g. in the initial DFT calculation). Wannier90 will then be able to find the 6 b-vectors needed with no trouble. 2- Work with the slightly distorted cell e.g. if you think this has some physical meaning. To get wannier90 to work with this cell we need to do two changes to precisions: a- increase the number of significant figures in the mesh specification e.g. 0.833 0.833 0.667 0.833 0.833 0.833 b- set kmesh_tol = 0.1 With these two changes, your distorted cell works fine for me. The result you get is that the set of 6 b-vectors which would be degenerate in the cubic case, split into a set of 4 and a set of 2. You then need a fourth set (albeit at low weight) to deal with the distortion. | b_k Vectors (Ang^-1) and Weights (Ang^2) | | | |No. b_k(x) b_k(y) b_k(z)w_b | |--- | | 1-0.530.2637750.00 7.186018| | 2 0.2637750.620.00 7.186018| | 3 0.53 -0.2637750.00 7.186018| | 4-0.263775 -0.620.00 7.186018| | 5 0.000.000.263775 7.186283| | 6 0.000.00 -0.263775 7.186283| | 7 0.263827 -0.2637120.00 0.000265| | 8-0.2638270.2637120.00 0.000265| Jonathan — Prof. Jonathan Yates Professor of Materials Modelling, Dept of Materials, University of Oxford Dean / Tutor for Materials Science, St Edmund Hall, Oxford. > On 3 Aug 2023, at 23:01, Payden Brown wrote: > > Hello all, > > When I run wannier90.x -pp I first got a "kmesh_get: something wrong, found > too many nearest neighbours" error. After looking through the previous forum > discussions I modified kmesh_tol in attempts to fix but now I get > "kmesh_get_bvector: Not enough bvectors found" and when I decrease kmesh_tol > even more I get "Non-symmetric k-point neighbours!". Please help! > > Thank you, > > Payden Brown > > Here is my .win input: > > begin projections > random > end projections > guiding_centres=.true. > > num_bands = 16 > num_wann = 16 > exclude_bands = 1,2,3,4,5,6,7,8,9,10 > > iprint = 2 > > dis_num_iter = 500 > dis_win_max = 9.0 > dis_froz_max = 9.0 > dis_conv_tol = 1.0d-9 > dis_conv_window = 20 > > num_iter = 5000 > conv_tol= 1.0d-9 > conv_window = 20 > optimisation = 0 > mp_grid = 6 6 6 > > > begin unit_cell_cart > Ang >3.970048210.0007929080000. > -0.0009393540003.9700481780000. >0.0.3.970048235000 > end unit_cell_cart > > write_u_matrices = .true. > write_xyz = .true. > > #restart = plot > bands_plot = .true. > bands_plot_format = gnuplot > bands_num_points = 100 > > > begin kpoint_path > G 0.000 0.000 0.000 X 0.000 0.500 0.000 > X 0.000 0.500 0.000 M 0.500 0.500 0.000 > M 0.500 0.500 0.000 G 0.000 0.000 0.000 > G 0.000 0.000 0.000 R 0.500 0.500 0.500 > R 0.500 0.500 0.500 X 0.000 0.500 0.000 > M 0.500 0.500 0.000 R 0.500 0.500 0.500 > end kpoint_path > > > begin atoms_frac > Ta 0.50000.50000.5000 > O 0.50000.0.5000 > O 0.0.50000.5000 > O 0.50000.50000. > Na 0.0.0. > end atoms_frac > > > begin kpoints > > 0. 0. 0. > 0. 0. 0.1667 > 0. 0. 0. > 0. 0. 0.5000 ... > > OUTPUT (mesh_tol unspecified): > > *-- K-MESH --* > > ++ > |Distance to Nearest-Neighbour Shells > | > | > | > | Shell Distance (Ang^-1)
Re: [Wannier] k-gradient implementation
Dear Yue, In support of Nicola’s comments: a long time I did some comparisons of the B1 approach from MV97, and the simpler 6 neighbour approach. I didn’t look at the resulting MLWF - rather I looked at the form of the position operator they lead to. Indeed, for the same k-point mesh the B1 approach gives a more accurate position operator - and also a more symmetric representation. See http://www.tcm.phy.cam.ac.uk/~jry20/wannier/pos_op.html Jonathan — Prof. Jonathan Yates Professor of Materials Modelling, Dept of Materials, University of Oxford Tutor for Materials Science, St Edmund Hall, Oxford. > On 4 Nov 2023, at 02:55, Nicola Marzari wrote: > > > Dear Yue, > > admittedly both are easy - but think e.g. at a fcc lattice - its reciprocal > lattice is bcc, 8 neighbours, and calculating the gradient using those 8 b_k > vectors will be more accurate, at a given sampling, than just using 3. > > nicola > > > > On 03/11/2023 23:55, Lun Yue wrote: >> Dear all, >> I have a question regarding the implementation of the k-gradient. >> 1) In Wannier90, it is implemented by constructing the weights such that the >> completeness relation is fully satisfied [Eq. (B1), PRB 56, 12847 (1997)]. >> 2) Another approach would be to calculate the numerical derivatives along >> the reciprocal lattice vectors (which is easy as the quantities are given in >> a Monkhorst-Pack grid), and then transform to the Cartesian coordinates >> using the metric tensor and the reciprocal lattice vectors. >> I am wondering why approach 1) was implemented over approach 2) in >> Wannier90. The second approach seems to be easier, or does approach 2) fail >> in some cases? >> Best regards, >> Lun Yue >> Louisiana State University >> ___ >> Wannier mailing list >> Wannier@lists.quantum-espresso.org >> https://lists.quantum-espresso.org/mailman/listinfo/wannier > > -- > -- > Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL > Director, National Centre for Competence in Research NCCR MARVEL, SNSF > Head, Laboratory for Materials Simulations, Paul Scherrer Institut > Contact info and websites at http://theossrv1.epfl.ch/Main/Contact > > ___ > Wannier mailing list > Wannier@lists.quantum-espresso.org > https://lists.quantum-espresso.org/mailman/listinfo/wannier ___ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier