From simon.blackburn at umontreal.ca Tue Jul 7 15:51:35 2009 From: simon.blackburn at umontreal.ca (Blackburn Simon) Date: Tue, 7 Jul 2009 09:51:35 -0400 Subject: [Wannier] Negative spread for YBCO Message-ID: Hi, I am calculating wannier functions for a highly correlated material, but I got negative spreads for some functions : WF centre and spread 1 ( 1.396087, 1.951657, 4.321091 ) -67.06856766 WF centre and spread 2 ( 1.934788, 1.951531, 5.977162 ) 2.40616250 WF centre and spread 3 ( 1.934421, 1.951706, 5.977224 ) 2.39819002 WF centre and spread 4 ( 1.934972, 1.951188, 5.976614 ) 2.39891922 WF centre and spread 5 ( 1.959096, 1.952858, 6.037811 ) 8.83943628 WF centre and spread 6 ( 1.870143, 1.952592, 6.081024 ) 19.66047794 WF centre and spread 7 ( 1.972215, 1.937209, 5.988440 ) 21.95857517 WF centre and spread 8 ( 1.969876, 1.956152, 6.084227 ) 12.08353279 WF centre and spread 9 ( 1.959757, 1.947718, 6.065042 ) 9.55748623 WF centre and spread 10 ( 1.935713, 1.964057, 2.114624 ) 2.44523435 WF centre and spread 11 ( 2.031171, 2.025423, 1.941027 ) 4.06438871 WF centre and spread 12 ( 1.903267, 1.954871, 2.093530 ) 1.88270760 WF centre and spread 13 ( 1.917133, 1.936205, 2.150585 ) 1.62635224 WF centre and spread 14 ( 0.884877, 1.952352, 7.002234 ) -157.28293063 WF centre and spread 15 ( 1.613742, 1.951188, 8.931454 ) -28.16475747 WF centre and spread 16 ( 1.604912, 1.951940, 8.954071 ) -28.76398600 WF centre and spread 17 ( 1.607723, 1.951161, 8.930750 ) -28.99502259 I changed the functions and also I exclude low energy bands, but I still got negative spreads, however they were not necessarily for the same functions. I would like to know how come some spreads are negative and how to solve this problem. Sincerely, Simon Blackburn Here is the input file : num_iter = 10000 conv_window = 3 num_wann = 70 num_bands = 70 mp_grid = 6 6 6 begin projections random Y:l=0;l=1;l=2 Ba:l=0;l=1 Cu:l=1;l=2 O:l=0;l=1 end projections begin atoms_frac Y 5.0000000000E-01 5.0000000000E-01 5.0000000000E-01 Ba 5.0000000000E-01 5.0000000000E-01 1.8342796139E-01 Cu 5.1594053334E-41 3.8011886002E-41 2.2852404443E-23 Cu 6.7359629443E-17 6.6174517631E-17 3.5605414672E-01 O 2.9697060661E-17 2.9174577011E-17 1.5697390332E-01 O 5.0000000000E-01 7.0595436999E-17 3.7983938018E-01 O 1.0290749028E-16 5.0000000000E-01 3.7922374798E-01 O 3.1164471290E-17 5.0000000000E-01 2.2852404443E-23 Ba 5.0000000000E-01 5.0000000000E-01 8.1657203862E-01 Cu 1.2182433813E-16 1.1968098604E-16 6.4394585329E-01 O 1.5948690691E-16 1.5668092666E-16 8.4302609668E-01 O 5.0000000000E-01 1.1526006667E-16 6.2016061983E-01 O 1.4860541987E-16 5.0000000000E-01 6.2077625202E-01 end atoms_frac begin unit_cell_cart 3.8338465 0.0000000 0.0000000 0.0000000 3.9025064 0.0000000 0.0000000 0.0000000 11.8450852 end unit_cell_cart From jonathan.yates at materials.ox.ac.uk Tue Jul 7 17:24:33 2009 From: jonathan.yates at materials.ox.ac.uk (Jonathan Yates) Date: Tue, 07 Jul 2009 16:24:33 +0100 Subject: [Wannier] Negative spread for YBCO In-Reply-To: Message-ID: <20090707152433.F250414044@webmail222.herald.ox.ac.uk> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From marzari at MIT.EDU Tue Jul 7 19:04:45 2009 From: marzari at MIT.EDU (Nicola Marzari) Date: Tue, 07 Jul 2009 13:04:45 -0400 Subject: [Wannier] Negative spread for YBCO In-Reply-To: <20090707152433.F250414044@webmail222.herald.ox.ac.uk> References: <20090707152433.F250414044@webmail222.herald.ox.ac.uk> Message-ID: <4A53802D.9020902@mit.edu> Jonathan Yates wrote: > If that isn't the cause of your problem, then I suspect the next step is to put > the mmn,amn and eig files on a webserver - so that we can look at the problem in > more detail. Unless anyone else has seen, and cured, this type of problem before? Hi Simon, Jonathan - indeed, it's a phase problem, and I'm pretty sure I've seen it happen (12 years ago, so my memory might be hazy). The code uses definitions 31 and 32 from the 97 PRB to calculate the spread, but errors in phases do not cancel out when you subtract the square of 31 from 32 . What is an error in phase ? Well, let's take the simple example of a very localized wannier function centered somewhere in r0. If you use the equation in between 5 and 6, you'll see that M_nn^k,b is a constant ndependent of k: exp (-i b.r0) (from which Im ln M_nn^k,b gives you - b.r0, and eq 31 gives you r0 as the center). Now, suppose you are trying to get the WFs, and these are getting localized, but not perfectly, and the M_nn^k,b is not perfectly constant, and is such that for a vectot b in one direction it is just above the branch cut, and for b in the opposite direction it is just below the branch cut, then the sum of the two terms will be mismatched (there is a 2pi between the two of them), and instead of averaging to around r0, it averages to (r0+r0+lattice vector)/2, so it's really in the middle. That's why it is important to choose phase factors so that these problems do not arise, and that's what the sheet try to do in the code - picking up a good estimate of the center to guess the most appopriate phase factor. In all of this, the golden rule is to start from carefully chosen projections at the beginning - these usually solve all problems. nicola --------------------------------------------------------------------- Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu From sumanta.bhandary at fysik.uu.se Wed Jul 8 16:25:58 2009 From: sumanta.bhandary at fysik.uu.se (Sumanta Bhandary) Date: Wed, 08 Jul 2009 16:25:58 +0200 Subject: [Wannier] DOS and conductance calculations in W90 Message-ID: <20090708162558.m7vukxvzwkco8008@webmail6.uu.se> Dear Wannier90 users, I want to know how Density of states are calculated in wannier90 and how does it depend on frozen window. Also, is it possible to calculate the conductance and DOS for the full energy spectrum even when the inner window is set up to Fermi level ? Best Regards Sumanta Bhandary Uppsala University From marzari at MIT.EDU Wed Jul 8 16:39:28 2009 From: marzari at MIT.EDU (Nicola Marzari) Date: Wed, 08 Jul 2009 10:39:28 -0400 Subject: [Wannier] DOS and conductance calculations in W90 In-Reply-To: <20090708162558.m7vukxvzwkco8008@webmail6.uu.se> References: <20090708162558.m7vukxvzwkco8008@webmail6.uu.se> Message-ID: <4A54AFA0.708@mit.edu> Sumanta Bhandary wrote: > Dear Wannier90 users, > I want to know how Density of states are > calculated in wannier90 and how does it depend on frozen window. > Also, is it possible to calculate the conductance and DOS for the full > energy spectrum even when the inner window is set up to Fermi level ? The algorithms are those of Nardelli PRB 99, Calzolari Marzari Souza Nardelli PRB 2004, Lee, Nardelli, Marzari PRL 2005 . The DOS and conductance (let's think at a bulk system for the time being) will be those described by the WFs you have identified - if those reproduce the band structure of your system well or not above the inner window will depend on your system, on your choice of WFs, etc. Even below the inner window you can do a poor job in interpolating the band structure at k-points in between those you actually used in your Wannier calculation - a good study case would be a semiconducting CNT, using only the occupied subspace. The WFs interpolation would be quite poor, unless you were to disentanble the entire pi manifold. Have a close look at the PRL mentioned above. nic --------------------------------------------------------------------- Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu From sagarambavale at yahoo.co.in Fri Jul 10 09:29:53 2009 From: sagarambavale at yahoo.co.in (ambavale sagar) Date: Fri, 10 Jul 2009 12:59:53 +0530 (IST) Subject: [Wannier] mkl-library problem for want code Message-ID: <450169.117.qm@web94613.mail.in2.yahoo.com> Dear wannier users, I am trying to compile "want" code on my quad-core dual cpu intel xeon work station. I have working mkl-10.0.1.014 installed and pwscf is working well in parallel. When I tried to configure want-code it detected : The following libraries have been found: BLAS_LIBS= -lmkl_em64t -lguide -lpthread LAPACK_LIBS=-lmkl_lapack FFT_LIBS= -lfftw but make all gives the error: ...... ..... ftread.o want_init.o want_interfaces.o wf2_augment.o wfc_data.o wfc_info.o wfc_manager.o windows.o xsf.o zmatrix.o ../libs/libwant.a ../iotk/libiotk.a ../ctools/libctools.a -lmkl_lapack -lmkl_em64t -lguide -lpthread -lfftw /opt/intel/mkl/10.0.1.014/lib/em64t/libmkl_lapack.so: undefined reference to `mkl_serv_load_fun' /opt/intel/mkl/10.0.1.014/lib/em64t/libmkl_lapack.so: undefined reference to `mkl_serv_load_dll' make[1]: *** [disentangle.x] Error 1 make[1]: Leaving directory `/home/phy/want-2.1.1/wannier' make: *** [wannier] Error 2 Searching the pwscf forum for same error, suggested me that for BLAS_LIBS=-lguide and -lpthread should not be detected. {-lmkl_sequential does not hellp} In config.msg of pwscf -lguide and -lpthread were not there, but for want-code they are deteced. Can you suggest that how can i install want-code successfully? Regards Sagar Ambavale Phd student M.S. University of Baroda India Looking for local information? Find it on Yahoo! Local http://in.local.yahoo.com/ -------------- next part -------------- An HTML attachment was scrubbed... URL: From edouard.duriau at gmail.com Sun Jul 26 15:11:44 2009 From: edouard.duriau at gmail.com (Edouard Duriau) Date: Sun, 26 Jul 2009 15:11:44 +0200 Subject: [Wannier] Problems with test cases Message-ID: <66650c2d0907260611v190cf4f6v74e8b80e230d9e38@mail.gmail.com> *Dear all, Can somebody please help me? I just built Wannier90 on my machine with intel duo core, gfortran-4.4, Linux (Ubuntu 9.04) The building went on correctly (see results below from the terminal window)* eduriau at VaioEd-Linux:~/WORK/Wannier90/wannier90-1.1$ sudo make (cd src ; make prog) make[1]: Entering directory `/home/eduriau/WORK/Wannier90/wannier90-1.1/src' gfortran-4.4 wannier_prog.F90 -O2 constants.o io.o utility.o parameters.o hamiltonian.o overlap.o kmesh.o disentangle.o wannierise.o plot.o transport.o -L/usr/lib -llapack -lf77blas -lcblas -latlas -o ../wannier90.x make[1]: Leaving directory `/home/eduriau/WORK/Wannier90/wannier90-1.1/src' eduriau at VaioEd-Linux:~/WORK/Wannier90/wannier90-1.1$ ls CHANGE.log doc LICENCE make.sys README src utility config examples Makefile make.sys~ README.install tests wannier90.x *After building Wannier90, I ran the test cases with the "make test" command but I obtained the following results: *eduriau at VaioEd-Linux:~/WORK/Wannier90/wannier90-1.1$ sudo make test (cd src ; make prog) make[1]: Entering directory `/home/eduriau/WORK/Wannier90/wannier90-1.1/src' gfortran-4.4 wannier_prog.F90 -O2 constants.o io.o utility.o parameters.o hamiltonian.o overlap.o kmesh.o disentangle.o wannierise.o plot.o transport.o -L/usr/lib -llapack -lf77blas -lcblas -latlas -o ../wannier90.x make[1]: Leaving directory `/home/eduriau/WORK/Wannier90/wannier90-1.1/src' (cd tests ; make test ) make[1]: Entering directory `/home/eduriau/WORK/Wannier90/wannier90-1.1/tests' ./run_test.pl Running test set of inputs Starting test1 Valence States of GaAs with a 2x2x2 kpoint mesh STOP wannier90 error: examine the output/error file for details Test Failed. Starting test2 4 MLWF for Li using disentanglement and a 4x4x4 grid Plotting Interpolated Band Structure STOP wannier90 error: examine the output/error file for details Test Failed. Tests Complete Examine the file /home/eduriau/WORK/Wannier90/wannier90-1.1/tests/wantest.log make[1]: Leaving directory `/home/eduriau/WORK/Wannier90/wannier90-1.1/tests'* For test 1: (extract from wannier.wout, I did not copy the header)* Starting a new Wannier90 calculation ... Time to get kmesh 0.052 (sec) Reading overlaps from wannier.mmn : Created on 25Feb2006 at 11:03:51 Reading projections from wannier.amn : Created on 25Feb2006 at 11:03:51 ERROR: unitarity of initial U nkp= 1 i= 1 j= 1 [u_matrix.transpose(u_matrix)]_ij= ************ 0.000000 Exiting....... Error in unitarity of initial U in overlap_project *For test 2: (extract from wannier.wout, I did not copy the header)* Starting a new Wannier90 calculation ... Time to get kmesh 0.200 (sec) Reading overlaps from wannier.mmn : Created on 25Feb2006 at 11:03:51 Reading projections from wannier.amn : Created on 25Feb2006 at 11:03:51 Time to read overlaps 0.120 (sec) *------------------------------- DISENTANGLE --------------------------------* +----------------------------------------------------------------------------+ | Energy Windows | | --------------- | | Outer: -8.00000 to 14.30000 (eV) | | Inner: -8.00000 to 6.70000 (eV) | +----------------------------------------------------------------------------+ Number of target bands to extract: 4 Unitarised projection of Wannier functions ------------------------------------------ A_mn = --> S = A.A^+ --> U = S^-1/2.A In dis_project... ERROR: unitarity of initial U nkp= 1 i= 1 j= 1 [u_matrix_opt.transpose(u_matrix_opt)]_ij= ************ 0.000000 Exiting....... dis_project: Error in unitarity of initial U in dis_project *Of course, the .amn and .mmn files were present. I also tried to run the examples but all of them returned the same problem in their respective .wout files.* *I am a recent user of Linux and I have no clue about what I did wrong and what I could do to make Wannier90 run correctely. Can you please give me some advise/help?* *Thanks a lot upfront, Ed * -------------- next part -------------- An HTML attachment was scrubbed... URL: From jonathan.yates at materials.ox.ac.uk Sun Jul 26 23:10:35 2009 From: jonathan.yates at materials.ox.ac.uk (Jonathan Yates) Date: Sun, 26 Jul 2009 22:10:35 +0100 Subject: [Wannier] Problems with test cases In-Reply-To: <66650c2d0907260611v190cf4f6v74e8b80e230d9e38@mail.gmail.com> References: <66650c2d0907260611v190cf4f6v74e8b80e230d9e38@mail.gmail.com> Message-ID: <8385EE14-52FE-4913-AEEE-6A92CD563A2D@materials.ox.ac.uk> Ed, I don't think you did anything obviously wrong. My guess is that there is a (subtle) incompatibility between the compiled W90 and the lapack/blas routines. This could be because the atlas routines you are linking against have been compiled with a version of gfortran other than the one you are using to compile W90. Or it could be a totally different reason - either way every time I've seen this behavior it's been a linking blas issue. My suggestion would be to compile the versions of lapack and blas from netlib (www.netlib.org/lapack/). The blas is not optimised - but it compiles quickly and is robust. If that works try compiling your own atlas blas (if you haven't already). If not, try swapping the compiler for an older version of gfortran, or g95. Yours Jonathan On 26 Jul 2009, at 14:11, Edouard Duriau wrote: > Dear all, > > > Can somebody please help me? > > I just built Wannier90 on my machine with intel duo core, > gfortran-4.4, Linux (Ubuntu 9.04) > The building went on correctly (see results below from the terminal > window) -- Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK tel: +44 (0)1865 283225 http://users.ox.ac.uk/~oums0549/ From edouard.duriau at gmail.com Tue Jul 28 07:47:34 2009 From: edouard.duriau at gmail.com (Edouard Duriau) Date: Tue, 28 Jul 2009 07:47:34 +0200 Subject: [Wannier] Problems with test cases In-Reply-To: <8385EE14-52FE-4913-AEEE-6A92CD563A2D@materials.ox.ac.uk> References: <66650c2d0907260611v190cf4f6v74e8b80e230d9e38@mail.gmail.com> <8385EE14-52FE-4913-AEEE-6A92CD563A2D@materials.ox.ac.uk> Message-ID: <66650c2d0907272247l5c2c3c51ua57eca921a8cdbd3@mail.gmail.com> Thanks to both of you! I have cleaned everything and double checked the BLAS and LAPACK libs then I have successfully rebuilt Wannier90. Now it seems to work fine, thanks a lot for your help! Ed On Sun, Jul 26, 2009 at 11:10 PM, Jonathan Yates < jonathan.yates at materials.ox.ac.uk> wrote: > Ed, > > I don't think you did anything obviously wrong. > > My guess is that there is a (subtle) incompatibility between the > compiled W90 and the lapack/blas routines. This could be because the > atlas routines you are linking against have been compiled with a > version of gfortran other than the one you are using to compile W90. > Or it could be a totally different reason - either way every time I've > seen this behavior it's been a linking blas issue. > > My suggestion would be to compile the versions of lapack and blas > from netlib (www.netlib.org/lapack/). The blas is not optimised - but > it compiles quickly and is robust. If that works try compiling your > own atlas blas (if you haven't already). If not, try swapping the > compiler for an older version of gfortran, or g95. > > Yours > Jonathan > > > On 26 Jul 2009, at 14:11, Edouard Duriau wrote: > > > Dear all, > > > > > > Can somebody please help me? > > > > I just built Wannier90 on my machine with intel duo core, > > gfortran-4.4, Linux (Ubuntu 9.04) > > The building went on correctly (see results below from the terminal > > window) > > -- > Department of Materials, University of Oxford, Parks Road, Oxford, OX1 > 3PH, UK > tel: +44 (0)1865 283225 http://users.ox.ac.uk/~oums0549/ > > _______________________________________________ > Wannier mailing list > Wannier at quantum-espresso.org > http://www.democritos.it/mailman/listinfo/wannier > -------------- next part -------------- An HTML attachment was scrubbed... URL: From sumanta.bhandary at fysik.uu.se Wed Jul 29 10:34:18 2009 From: sumanta.bhandary at fysik.uu.se (Sumanta Bhandary) Date: Wed, 29 Jul 2009 10:34:18 +0200 Subject: [Wannier] Error in allocating wann_func in plot_wannier Message-ID: <20090729103418.ksj8nwi674kowo4g@webmail5.uu.se> Dear W90 users, I am new to wannier and trying to do parallel calculation . But the calculation stops saying ""Error in allocating wann_func in plot_wannier" . Could you please give me any suggestion ? Best Regards Sumanta Bhandary Uppsala University From jonathan.yates at materials.ox.ac.uk Wed Jul 29 11:47:11 2009 From: jonathan.yates at materials.ox.ac.uk (Jonathan Yates) Date: Wed, 29 Jul 2009 10:47:11 +0100 Subject: [Wannier] Error in allocating wann_func in plot_wannier In-Reply-To: <20090729103418.ksj8nwi674kowo4g@webmail5.uu.se> References: <20090729103418.ksj8nwi674kowo4g@webmail5.uu.se> Message-ID: On 29 Jul 2009, at 09:34, Sumanta Bhandary wrote: > Dear W90 users, > I am new to wannier and trying to do parallel > calculation . But the calculation stops saying ""Error in allocating > wann_func in plot_wannier" . Could you please give me any suggestion ? Dear Sumanta, Firstly, wannier90 is not a parallel program. It is fine to run your original ab-initio program in parallel - but then you should run wannier90 in serial. The error message means the program is running out of RAM when it tries to form the wannier functions on a regular grid for plotting. There are various things that can reduce the memory (actually, try running in serial first - that could be the problem). 1- The size of the grid. An important keyword is wannier_plot_supercell. The userguide and tutorial will explain this. If you're using pw2wannier90 (ie QE) then there is a logical flag reduce_unk which reduces the density of the grid by half (so memory by ~8) 2- The number of wannier functions to plot at one time. By default we plot all of them. Use wannier_plot_list to plot a subset. Yours Jonathan -- Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK tel: +44 (0)1865 283225 http://users.ox.ac.uk/~oums0549/ From sumanta.bhandary at fysik.uu.se Wed Jul 29 12:36:28 2009 From: sumanta.bhandary at fysik.uu.se (Sumanta Bhandary) Date: Wed, 29 Jul 2009 12:36:28 +0200 Subject: [Wannier] Error in allocating wann_func in plot_wannier Message-ID: <20090729123628.feeu2issiscwkcsk@webmail5.uu.se> Thanks Jonathan for your help . It worked . Regards Sumanta