From r95222066 at ntu.edu.tw Tue Nov 13 08:01:22 2007 From: r95222066 at ntu.edu.tw (r95222066 at ntu.edu.tw) Date: Tue, 13 Nov 2007 15:01:22 +0800 Subject: [Wannier] Does wannier90 program need wfc file generated from pwscf process of scf or nscf ? Message-ID: <20071113150122.jsr87qk28soc0gg8@wmail1.cc.ntu.edu.tw> Hello, I want to ask if wannier90 package need the wavefunction files which are generated from PWSCF scf process or nscf process ? Because I have a system which has 66 atoms and I want to calculate the magnetic properties , and when I used more nodes to run pwscf program for faster , the wfc (wavefunctions) will be larger proportional to nodes I used. Thats situation confused me . Thanks a lot! Bing-Hong Chen From marzari at MIT.EDU Tue Nov 13 15:30:50 2007 From: marzari at MIT.EDU (Nicola Marzari) Date: Tue, 13 Nov 2007 09:30:50 -0500 Subject: [Wannier] Does wannier90 program need wfc file generated from pwscf process of scf or nscf ? In-Reply-To: <20071113150122.jsr87qk28soc0gg8@wmail1.cc.ntu.edu.tw> References: <20071113150122.jsr87qk28soc0gg8@wmail1.cc.ntu.edu.tw> Message-ID: <4739B51A.3020207@mit.edu> If you are using PWSCF, indeed you will need the wavefunctions for the pw2wannier intermediate step. nicola r95222066 at ntu.edu.tw wrote: > Hello, > > I want to ask if wannier90 package need the wavefunction files > which are generated from PWSCF scf process or nscf process ? > Because I have a system which has 66 atoms and I want to > calculate the magnetic properties , and when I used more nodes to run > pwscf program for faster , the wfc (wavefunctions) will be larger > proportional to nodes I used. Thats situation confused me . > > Thanks a lot! > > Bing-Hong Chen > _______________________________________________ > Wannier mailing list > Wannier at quantum-espresso.org > http://www.democritos.it/mailman/listinfo/wannier -- --------------------------------------------------------------------- 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 levuser at gmail.com Sat Nov 17 21:01:34 2007 From: levuser at gmail.com (Sergey Levchenko) Date: Sat, 17 Nov 2007 12:01:34 -0800 Subject: [Wannier] Plotting several wannier functions in one unit cell? Message-ID: Hi! Is there a way to plot several different Wannier functions in one unit cell? When I request several functions using wannier_plot_list, several different xsf files are generated. Thanks! Sergey Levchenko. From jry20 at cam.ac.uk Sat Nov 17 22:45:00 2007 From: jry20 at cam.ac.uk (Jonathan Yates) Date: Sat, 17 Nov 2007 21:45:00 +0000 (GMT) Subject: [Wannier] Plotting several wannier functions in one unit cell? In-Reply-To: References: Message-ID: On Sat, 17 Nov 2007, Sergey Levchenko wrote: > Is there a way to plot several different Wannier functions in one unit > cell? When I request several functions using wannier_plot_list, > several different xsf files are generated. Sergey, The quick answer to your question is that I do not know an easy way to do this. But I can appreciate that it would be a useful thing to be able to do.... The xsf format is intended to be read by Xcrysden. As I understand it Xcrysden is not able to plot multiple isosurfaces (ie several WF at once). You could put several WF in a single xsf file - however, you are then only able to plot a single isosurface over a linear combination of the WF - which generally isn't helpful. In the near future we will add support for cube format. This is read by several programs (jmol, vmd) and may help. Yours Jonathan btw. If you have any suggestions for good crystal visualisation programs that support multiple isosurfaces, we'd be interested to hear them - even more so if someone will volunteer to write the wannier90 interface... -- Dr Jonathan Yates | Theory of Condensed Matter Group Corpus Christi College | Cavendish Laboratory Cambridge, CB2 1RH, UK | Cambridge, CB3 OHE, UK email jry20 at cam.ac.uk | Tel +44 (0)1223 337461 From wieferink at uni-muenster.de Mon Nov 26 10:02:08 2007 From: wieferink at uni-muenster.de (=?iso-8859-1?q?J=FCrgen_Wieferink?=) Date: Mon, 26 Nov 2007 10:02:08 +0100 Subject: [Wannier] Library mode Message-ID: <200711261002.08705.wieferink@uni-muenster.de> Hello, to understand the adsorption of organic molecules on silicon and silicon carbide surfaces I have interfaced our density functional code to wannier90 (v1.0.2) as a post-processing tool. It is really great how little work is needed to get centres and plots of Wannier functions. Nevertheless, I have a few questions I don't know where else to ask. One question at this stage: Plotting the Wannier functions only works when I write \psi_{n,k}(r) instead of u_{n,k}(r) to UNK000*.1. Is this intentional or do I do anything wrong somewhere else? On the long run, I want to use our own visualization methods. So I need the information about the centres and the unitary transformations within our code. Using the library mode seems to be the easiest way. Thus, I have had a look into the source code and have a few questions. One has to provide the M_matrix to wannier_run() without an explicit reference to nnlist and nncell. Do I have to rely on the fact that wannier90 will generate these lists in exactly the same order? And secondly it seems that the centres and spreads are not really passed to the caller after being reset. I could fix this for myself by making rave and r2ave global and passing it to wann_centres and friends. Or is this fixed in a more recent release? Or am I wrong anyway? Juergen -- Juergen Wieferink AG Pollmann Institute of Solid State Physics University of Muenster From a.mostofi at imperial.ac.uk Wed Nov 28 01:57:40 2007 From: a.mostofi at imperial.ac.uk (Arash Mostofi) Date: Wed, 28 Nov 2007 00:57:40 +0000 Subject: [Wannier] Library mode In-Reply-To: <200711261002.08705.wieferink@uni-muenster.de> References: <200711261002.08705.wieferink@uni-muenster.de> Message-ID: <474CBD04.1000605@imperial.ac.uk> Dear Jurgen Wannier90 assumes that it is the u_nk(r) and not the \psi_nk(r) that are written to the UNK* files. If you look at plot.F90 in the src directory of the distribution, in the subroutine plot_wannier, the e^{ik.r} is explicitly factored in when constructing the MLWFs from the u_nk(r). I guess this leaves us wondering why you get the behaviour you observe. You're right about the centres and the spreads when using library mode. In v1.0.2 they are just set to zero before returning to the calling program. This will be fixed in the next release (v1.1), which will be sometime next month. As regards nnlist and nncell, yes, these are re-calculated in wannier_run in exactly the same way in which they are calculated in wannier_setup, so there is no need to pass them back as arguments. Incidentally, what is the code to which you've interfaced Wannier90? Cheers Arash -- :------------------------------------------------------------: : Dr. Arash A. Mostofi :: a.mostofi at imperial.ac.uk : : Lecturer and RCUK Fellow :: : : Depts. of Materials & Physics :: : : Imperial College London :: T +44 (0)207 594 6753 : : London SW7 2AZ, United Kingdom :: F +44 (0)207 594 6757 : :------- http://www.cmth.ph.ic.ac.uk/people/a.mostofi -------: J?rgen Wieferink wrote: > Hello, > > to understand the adsorption of organic molecules on silicon and > silicon carbide surfaces I have interfaced our density functional > code to wannier90 (v1.0.2) as a post-processing tool. It is really > great how little work is needed to get centres and plots of Wannier > functions. Nevertheless, I have a few questions I don't know where > else to ask. > > One question at this stage: Plotting the Wannier functions only > works when I write \psi_{n,k}(r) instead of u_{n,k}(r) to UNK000*.1. > Is this intentional or do I do anything wrong somewhere else? > > > On the long run, I want to use our own visualization methods. So I > need the information about the centres and the unitary > transformations within our code. Using the library mode seems to be > the easiest way. Thus, I have had a look into the source code and > have a few questions. > > One has to provide the M_matrix to wannier_run() without an explicit > reference to nnlist and nncell. Do I have to rely on the fact that > wannier90 will generate these lists in exactly the same order? And > secondly it seems that the centres and spreads are not really passed > to the caller after being reset. I could fix this for myself by > making rave and r2ave global and passing it to wann_centres and > friends. Or is this fixed in a more recent release? Or am I wrong > anyway? > > Juergen > > -- > Juergen Wieferink > AG Pollmann > Institute of Solid State Physics > University of Muenster > > _______________________________________________ > Wannier mailing list > Wannier at quantum-espresso.org > http://www.democritos.it/mailman/listinfo/wannier From wieferink at uni-muenster.de Wed Nov 28 15:34:33 2007 From: wieferink at uni-muenster.de (=?iso-8859-1?q?J=FCrgen_Wieferink?=) Date: Wed, 28 Nov 2007 15:34:33 +0100 Subject: [Wannier] Library mode In-Reply-To: <474CBD04.1000605@imperial.ac.uk> References: <200711261002.08705.wieferink@uni-muenster.de> <474CBD04.1000605@imperial.ac.uk> Message-ID: <200711281534.33641.wieferink@uni-muenster.de> Dear list, dear Arash, > Wannier90 assumes that it is the u_nk(r) and not the \psi_nk(r) that are > written to the UNK* files. If you look at plot.F90 in the src directory > of the distribution, in the subroutine plot_wannier, the e^{ik.r} is > explicitly factored in when constructing the MLWFs from the u_nk(r). I > guess this leaves us wondering why you get the behaviour you observe. Weird. But as soon as the library interface works fine, I will not need this any more. > You're right about the centres and the spreads when using library mode. > In v1.0.2 they are just set to zero before returning to the calling > program. This will be fixed in the next release (v1.1), which will be > sometime next month. Fine. Just for the case someone has use for it, I have attached a patch containing quick fixes to bugs I came across. > Incidentally, what is the code to which you've interfaced Wannier90? I'm using my own density functional theory code. It expands wave functions in Gaussian orbitals and local potentials as Fourier series. It does not have a name, but is described in Phys. Rev. B74, 205311 (2006). Wannier90 is the first external tool apart from Lapack and some FFT library to be interfaced. Thanks for your reply, Juergen -- Juergen Wieferink AG Pollmann Institute of Solid State Physics University of Muenster -------------- next part -------------- A non-text attachment was scrubbed... Name: wannier90-lib.patch Type: text/x-diff Size: 4958 bytes Desc: not available URL: