You can subscribe to this list here.
2010 
_{Jan}

_{Feb}
(2) 
_{Mar}
(22) 
_{Apr}
(11) 
_{May}
(19) 
_{Jun}
(25) 
_{Jul}
(13) 
_{Aug}
(28) 
_{Sep}
(17) 
_{Oct}
(12) 
_{Nov}
(17) 
_{Dec}
(14) 

2011 
_{Jan}
(1) 
_{Feb}
(11) 
_{Mar}
(20) 
_{Apr}
(7) 
_{May}
(44) 
_{Jun}
(33) 
_{Jul}
(6) 
_{Aug}
(10) 
_{Sep}
(38) 
_{Oct}
(12) 
_{Nov}
(21) 
_{Dec}
(9) 
2012 
_{Jan}
(2) 
_{Feb}
(15) 
_{Mar}
(37) 
_{Apr}
(29) 
_{May}
(36) 
_{Jun}
(24) 
_{Jul}
(16) 
_{Aug}
(14) 
_{Sep}
(27) 
_{Oct}
(63) 
_{Nov}
(34) 
_{Dec}
(10) 
2013 
_{Jan}
(6) 
_{Feb}
(25) 
_{Mar}
(34) 
_{Apr}
(23) 
_{May}
(36) 
_{Jun}
(54) 
_{Jul}
(9) 
_{Aug}
(19) 
_{Sep}
(22) 
_{Oct}
(11) 
_{Nov}
(35) 
_{Dec}
(33) 
2014 
_{Jan}
(20) 
_{Feb}
(21) 
_{Mar}
(24) 
_{Apr}
(31) 
_{May}
(29) 
_{Jun}
(31) 
_{Jul}
(49) 
_{Aug}
(51) 
_{Sep}
(3) 
_{Oct}

_{Nov}

_{Dec}

S  M  T  W  T  F  S 





1

2

3

4

5

6

7
(2) 
8
(1) 
9

10

11

12

13

14

15

16

17

18

19

20

21

22

23
(4) 
24

25
(3) 
26

27
(1) 
28
(2) 
29
(4) 
30

31
(2) 
From: Prasad Matukumilli <prasadchoudary@gm...>  20130831 15:28:00

Dear Dr.Togo, I'm sorry I didn't mentioned about system as CNT earlier. Thanks for your time and attention. Prasad. On Sat, Aug 31, 2013 at 7:24 PM, Atsushi Togo <atz.togo@...> wrote: > Hi, > > You ought to have written at first that you are working for CNT. I > have quite no idea about CNT symmetry. So unfortunately I have no > reply to your questions. I guess your eigenvectors are often > degenerated if you use usual code that can not handle CNT symmetry > properly. If so, those degenerate bands may be resolved by some > appropriate way (that I don't know though) to make dot product scheme > work in better quality, but anyway I'm afraid of discussing more about > your problem. > > Togo > > On Fri, Aug 30, 2013 at 5:59 AM, Prasad Matukumilli > <prasadchoudary@...> wrote: > > Dear Dr.Togo, > > My system is CNT. For 6,6 cnt, dot product scheme working fine but it's > > failing for 10,10 cnt to sort the phonon modes. Combination of schemes > > might work well. Which method would you suggest to use along with dot > > product scheme? > > > > Using characters, how can I proceed to compute trace of the > > representation(which I dont know how to get that representation square > > matrix out of eigen vector) Is there a kind transformation needed? > > > > Prasad > > > > > > On Thu, Aug 29, 2013 at 7:57 PM, Atsushi Togo <atz.togo@...> > wrote: > >> > >> Hi, > >> > >> Without concrete example, I can't say anything on this problem. > >> Probably for complicate cases, a mixture of techniques would be > >> helpful, but I need to write code and test it to know what is most > >> robust using examples. Especially to handle degenerate bands would be > >> problematic. > >> On characters, phonon modes on a band path have the same set of > >> characters of the irreducible representation. > >> > >> Togo > >> > >> On Thu, Aug 29, 2013 at 11:08 PM, Prasad Matukumilli > >> <prasadchoudary@...> wrote: > >> > Dear Dr.Togo, > >> > I am facing trouble in sorting of phonons while plotting phonon > >> > dispersion. > >> > My present scheme similar to the one in phonopy (sorting according to > >> > best > >> > match by comparing dot product of current eigenvector with all next > >> > qpoints' > >> > eigenvectors) working almost fine when the system with small number of > >> > eigenvectors(due to less number of close matchings) but this scheme > >> > failing > >> > when eigenvectors number is little more(where many pairs of degenerate > >> > eigenvectors and contains more confusion for sorting). Also if at one > >> > qpoint there is a wrong matching, it leads to many wrong > predictions. I > >> > am > >> > looking for some solutions address this. > >> > > >> > Can you please elaborate a little regarding usage of characters of > >> > phonon > >> > modes for this. > >> > > >> > Thanks, > >> > Prasad > >> > > >> > > >> > On Thu, Aug 29, 2013 at 6:26 PM, Atsushi Togo <atz.togo@...> > >> > wrote: > >> >> > >> >> Hi Prasad, > >> >> > >> >> If you really expect more information, you should show your problem > in > >> >> details. > >> >> > >> >> Togo > >> >> > >> >> On Wed, Aug 28, 2013 at 9:29 PM, Prasad Matukumilli > >> >> <prasadchoudary@...> wrote: > >> >> > Dear Sebastian, > >> >> > > >> >> > Thank you for quick response. > >> >> > Yes, it's just taking dot product. I am also using the same in my > >> >> > code, > >> >> > but > >> >> > at the band connections this scheme is failing and resulting partly > >> >> > missing > >> >> > of some important branches from the dispersion. I would appreciate > if > >> >> > you > >> >> > can provide some pointers for other schemes apart this dot product > >> >> > scheme. > >> >> > I also tried taking comparison of slopes along with dot product > >> >> > scheme, > >> >> > but > >> >> > of no use and resulted in much incomplete dispersion. > >> >> > > >> >> > Thanks, > >> >> > Prasad > >> >> > > >> >> > > >> >> > On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen > >> >> > <sebastian@...> wrote: > >> >> >> > >> >> >> Dear Prasad > >> >> >> > >> >> >> > >> >> >> I think the function you are looking for is > >> >> >> "estimate_band_connection" > >> >> >> located in phonon/band_structure.py > >> >> >> > >> >> >> But basically it is just taking the dot product of all > eigenvectors > >> >> >> of > >> >> >> two > >> >> >> adjacent points ("metric" array in code below). The eigenvectors > >> >> >> with > >> >> >> largest dotproduct are most similar/"parallel" so they are assumed > >> >> >> to > >> >> >> belong > >> >> >> to same phonon branch. > >> >> >> > >> >> >> The code is: > >> >> >> def estimate_band_connection(prev_eigvecs, eigvecs, > >> >> >> prev_band_order): > >> >> >> metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) > >> >> >> connection_order = [] > >> >> >> indices = range(len(metric)) > >> >> >> indices.reverse() > >> >> >> for overlaps in metric: > >> >> >> maxval = 0 > >> >> >> for i in indices: > >> >> >> val = overlaps[i] > >> >> >> if i in connection_order: > >> >> >> continue > >> >> >> if val > maxval: > >> >> >> maxval = val > >> >> >> maxindex = i > >> >> >> connection_order.append(maxindex) > >> >> >> > >> >> >> band_order = [connection_order[x] for x in prev_band_order] > >> >> >> > >> >> >> return band_order > >> >> >> > >> >> >> > >> >> >> Cheers > >> >> >> Sebastian > >> >> >> > >> >> >> Sebastian Christensen > >> >> >> PhD student > >> >> >> Center for Materials Crystallography > >> >> >> Department of Chemistry & iNANO, Aarhus University > >> >> >> Office: 1511318 > >> >> >> > >> >> >> Langelandsgade 140, DK8000 Aarhus C, Denmark > >> >> >> Mail: sebastian@... > >> >> >> > >> >> >> ________________________________ > >> >> >> From: Prasad Matukumilli [prasadchoudary@...] > >> >> >> Sent: 27 August 2013 23:16 > >> >> >> To: Atsushi Togo > >> >> >> Cc: phonopyusers > >> >> >> Subject: Re: [Phonopyusers] method of sorting of phonon branches > >> >> >> > >> >> >> Dear Togo, > >> >> >> I have gone through the code, I failed to locate the lines for > >> >> >> comparison > >> >> >> of eigenvectors. As I am new to python, I request your help and > time > >> >> >> to > >> >> >> direct me to those lines. > >> >> >> > >> >> >> Thanks > >> >> >> Prasad. > >> >> >> > >> >> >> > >> >> >> On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@... > > > >> >> >> wrote: > >> >> >>> > >> >> >>> Hi, > >> >> >>> > >> >> >>> In the band connection in phonopy, eigenvectors at neighboring > >> >> >>> qpoints are compared and most similar ones are connected. It's > not > >> >> >>> implemented in phonopy as a band connection method, but you may > be > >> >> >>> able to use characters of phonon modes > >> >> >>> > >> >> >>> > >> >> >>> > >> >> >>> > http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes > >> >> >>> . > >> >> >>> In group velocity, a perturbation approach same as > Hellmann–Feynman > >> >> >>> theorem is used, so sorting is usually unnecessary. > >> >> >>> > http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . > >> >> >>> > >> >> >>> For more details, please read the code. > >> >> >>> > >> >> >>> Togo > >> >> >>> > >> >> >>> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli > >> >> >>> <prasadchoudary@...> wrote: > >> >> >>> > Dear Dr.Togo, > >> >> >>> > > >> >> >>> > I am working on to code 'sorting of phonon branches' resulted > out > >> >> >>> > from > >> >> >>> > the > >> >> >>> > eigen solver. > >> >> >>> > I request you to explain the outline of method behind the > sorting > >> >> >>> > of > >> >> >>> > phonon > >> >> >>> > branches which has been implemented in Phonopy for the features > >> >> >>> > of > >> >> >>> > band_connection or group velocity. > >> >> >>> > > >> >> >>> > Thank you. > >> >> >>> >  > >> >> >>> > ********************************** > >> >> >>> > Prasad Matukumilli > >> >> >>> > Research Scholar, > >> >> >>> > Advanced Technology Development Centre, > >> >> >>> > Indian Institute of Technology (IIT) > >> >> >>> > Kharagpur  721302, India > >> >> >>> > ********************************** > >> >> >>> > > >> >> >>> > > >> >> >>> > > >> >> >>> > > >> >> >>> > >  > >> >> >>> > Introducing Performance Central, a new site from SourceForge > and > >> >> >>> > AppDynamics. Performance Central is your source for news, > >> >> >>> > insights, > >> >> >>> > analysis and resources for efficient Application Performance > >> >> >>> > Management. > >> >> >>> > Visit us today! > >> >> >>> > > >> >> >>> > > >> >> >>> > > >> >> >>> > > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > >> >> >>> > _______________________________________________ > >> >> >>> > Phonopyusers mailing list > >> >> >>> > Phonopyusers@... > >> >> >>> > https://lists.sourceforge.net/lists/listinfo/phonopyusers > >> >> >>> > > >> >> >>> > >> >> >>> > >> >> >>> > >> >> >>>  > >> >> >>> Atsushi Togo > >> >> >>> http://atztogo.github.com/ > >> >> >>> atz.togo@... > >> >> >> > >> >> >> > >> >> >> > >> >> >> > >> >> >>  > >> >> >> ********************************** > >> >> >> Prasad Matukumilli > >> >> >> Research Scholar, > >> >> >> Advanced Technology Development Centre, > >> >> >> Indian Institute of Technology (IIT) > >> >> >> Kharagpur  721302, India > >> >> >> ********************************** > >> >> > > >> >> > > >> >> > > >> >> > > >> >> >  > >> >> > ********************************** > >> >> > Prasad Matukumilli > >> >> > Research Scholar, > >> >> > Advanced Technology Development Centre, > >> >> > Indian Institute of Technology (IIT) > >> >> > Kharagpur  721302, India > >> >> > ********************************** > >> >> > >> >> > >> >> > >> >>  > >> >> Atsushi Togo > >> >> http://atztogo.github.com/ > >> >> atz.togo@... > >> > > >> > > >> > > >> > > >> >  > >> > ********************************** > >> > Prasad Matukumilli > >> > Research Scholar, > >> > Advanced Technology Development Centre, > >> > Indian Institute of Technology (IIT) > >> > Kharagpur  721302, India > >> > ********************************** > >> > >> > >> > >>  > >> Atsushi Togo > >> http://atztogo.github.com/ > >> atz.togo@... > > > > > > > > > >  > > ********************************** > > Prasad Matukumilli > > Research Scholar, > > Advanced Technology Development Centre, > > Indian Institute of Technology (IIT) > > Kharagpur  721302, India > > ********************************** > > > >  > Atsushi Togo > http://atztogo.github.com/ > atz.togo@... >  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Atsushi Togo <atz.togo@gm...>  20130831 13:54:40

Hi, You ought to have written at first that you are working for CNT. I have quite no idea about CNT symmetry. So unfortunately I have no reply to your questions. I guess your eigenvectors are often degenerated if you use usual code that can not handle CNT symmetry properly. If so, those degenerate bands may be resolved by some appropriate way (that I don't know though) to make dot product scheme work in better quality, but anyway I'm afraid of discussing more about your problem. Togo On Fri, Aug 30, 2013 at 5:59 AM, Prasad Matukumilli <prasadchoudary@...> wrote: > Dear Dr.Togo, > My system is CNT. For 6,6 cnt, dot product scheme working fine but it's > failing for 10,10 cnt to sort the phonon modes. Combination of schemes > might work well. Which method would you suggest to use along with dot > product scheme? > > Using characters, how can I proceed to compute trace of the > representation(which I dont know how to get that representation square > matrix out of eigen vector) Is there a kind transformation needed? > > Prasad > > > On Thu, Aug 29, 2013 at 7:57 PM, Atsushi Togo <atz.togo@...> wrote: >> >> Hi, >> >> Without concrete example, I can't say anything on this problem. >> Probably for complicate cases, a mixture of techniques would be >> helpful, but I need to write code and test it to know what is most >> robust using examples. Especially to handle degenerate bands would be >> problematic. >> On characters, phonon modes on a band path have the same set of >> characters of the irreducible representation. >> >> Togo >> >> On Thu, Aug 29, 2013 at 11:08 PM, Prasad Matukumilli >> <prasadchoudary@...> wrote: >> > Dear Dr.Togo, >> > I am facing trouble in sorting of phonons while plotting phonon >> > dispersion. >> > My present scheme similar to the one in phonopy (sorting according to >> > best >> > match by comparing dot product of current eigenvector with all next >> > qpoints' >> > eigenvectors) working almost fine when the system with small number of >> > eigenvectors(due to less number of close matchings) but this scheme >> > failing >> > when eigenvectors number is little more(where many pairs of degenerate >> > eigenvectors and contains more confusion for sorting). Also if at one >> > qpoint there is a wrong matching, it leads to many wrong predictions. I >> > am >> > looking for some solutions address this. >> > >> > Can you please elaborate a little regarding usage of characters of >> > phonon >> > modes for this. >> > >> > Thanks, >> > Prasad >> > >> > >> > On Thu, Aug 29, 2013 at 6:26 PM, Atsushi Togo <atz.togo@...> >> > wrote: >> >> >> >> Hi Prasad, >> >> >> >> If you really expect more information, you should show your problem in >> >> details. >> >> >> >> Togo >> >> >> >> On Wed, Aug 28, 2013 at 9:29 PM, Prasad Matukumilli >> >> <prasadchoudary@...> wrote: >> >> > Dear Sebastian, >> >> > >> >> > Thank you for quick response. >> >> > Yes, it's just taking dot product. I am also using the same in my >> >> > code, >> >> > but >> >> > at the band connections this scheme is failing and resulting partly >> >> > missing >> >> > of some important branches from the dispersion. I would appreciate if >> >> > you >> >> > can provide some pointers for other schemes apart this dot product >> >> > scheme. >> >> > I also tried taking comparison of slopes along with dot product >> >> > scheme, >> >> > but >> >> > of no use and resulted in much incomplete dispersion. >> >> > >> >> > Thanks, >> >> > Prasad >> >> > >> >> > >> >> > On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen >> >> > <sebastian@...> wrote: >> >> >> >> >> >> Dear Prasad >> >> >> >> >> >> >> >> >> I think the function you are looking for is >> >> >> "estimate_band_connection" >> >> >> located in phonon/band_structure.py >> >> >> >> >> >> But basically it is just taking the dot product of all eigenvectors >> >> >> of >> >> >> two >> >> >> adjacent points ("metric" array in code below). The eigenvectors >> >> >> with >> >> >> largest dotproduct are most similar/"parallel" so they are assumed >> >> >> to >> >> >> belong >> >> >> to same phonon branch. >> >> >> >> >> >> The code is: >> >> >> def estimate_band_connection(prev_eigvecs, eigvecs, >> >> >> prev_band_order): >> >> >> metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) >> >> >> connection_order = [] >> >> >> indices = range(len(metric)) >> >> >> indices.reverse() >> >> >> for overlaps in metric: >> >> >> maxval = 0 >> >> >> for i in indices: >> >> >> val = overlaps[i] >> >> >> if i in connection_order: >> >> >> continue >> >> >> if val > maxval: >> >> >> maxval = val >> >> >> maxindex = i >> >> >> connection_order.append(maxindex) >> >> >> >> >> >> band_order = [connection_order[x] for x in prev_band_order] >> >> >> >> >> >> return band_order >> >> >> >> >> >> >> >> >> Cheers >> >> >> Sebastian >> >> >> >> >> >> Sebastian Christensen >> >> >> PhD student >> >> >> Center for Materials Crystallography >> >> >> Department of Chemistry & iNANO, Aarhus University >> >> >> Office: 1511318 >> >> >> >> >> >> Langelandsgade 140, DK8000 Aarhus C, Denmark >> >> >> Mail: sebastian@... >> >> >> >> >> >> ________________________________ >> >> >> From: Prasad Matukumilli [prasadchoudary@...] >> >> >> Sent: 27 August 2013 23:16 >> >> >> To: Atsushi Togo >> >> >> Cc: phonopyusers >> >> >> Subject: Re: [Phonopyusers] method of sorting of phonon branches >> >> >> >> >> >> Dear Togo, >> >> >> I have gone through the code, I failed to locate the lines for >> >> >> comparison >> >> >> of eigenvectors. As I am new to python, I request your help and time >> >> >> to >> >> >> direct me to those lines. >> >> >> >> >> >> Thanks >> >> >> Prasad. >> >> >> >> >> >> >> >> >> On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> >> >> >> wrote: >> >> >>> >> >> >>> Hi, >> >> >>> >> >> >>> In the band connection in phonopy, eigenvectors at neighboring >> >> >>> qpoints are compared and most similar ones are connected. It's not >> >> >>> implemented in phonopy as a band connection method, but you may be >> >> >>> able to use characters of phonon modes >> >> >>> >> >> >>> >> >> >>> >> >> >>> http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes >> >> >>> . >> >> >>> In group velocity, a perturbation approach same as Hellmann–Feynman >> >> >>> theorem is used, so sorting is usually unnecessary. >> >> >>> http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . >> >> >>> >> >> >>> For more details, please read the code. >> >> >>> >> >> >>> Togo >> >> >>> >> >> >>> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli >> >> >>> <prasadchoudary@...> wrote: >> >> >>> > Dear Dr.Togo, >> >> >>> > >> >> >>> > I am working on to code 'sorting of phonon branches' resulted out >> >> >>> > from >> >> >>> > the >> >> >>> > eigen solver. >> >> >>> > I request you to explain the outline of method behind the sorting >> >> >>> > of >> >> >>> > phonon >> >> >>> > branches which has been implemented in Phonopy for the features >> >> >>> > of >> >> >>> > band_connection or group velocity. >> >> >>> > >> >> >>> > Thank you. >> >> >>> >  >> >> >>> > ********************************** >> >> >>> > Prasad Matukumilli >> >> >>> > Research Scholar, >> >> >>> > Advanced Technology Development Centre, >> >> >>> > Indian Institute of Technology (IIT) >> >> >>> > Kharagpur  721302, India >> >> >>> > ********************************** >> >> >>> > >> >> >>> > >> >> >>> > >> >> >>> > >> >> >>> >  >> >> >>> > Introducing Performance Central, a new site from SourceForge and >> >> >>> > AppDynamics. Performance Central is your source for news, >> >> >>> > insights, >> >> >>> > analysis and resources for efficient Application Performance >> >> >>> > Management. >> >> >>> > Visit us today! >> >> >>> > >> >> >>> > >> >> >>> > >> >> >>> > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk >> >> >>> > _______________________________________________ >> >> >>> > Phonopyusers mailing list >> >> >>> > Phonopyusers@... >> >> >>> > https://lists.sourceforge.net/lists/listinfo/phonopyusers >> >> >>> > >> >> >>> >> >> >>> >> >> >>> >> >> >>>  >> >> >>> Atsushi Togo >> >> >>> http://atztogo.github.com/ >> >> >>> atz.togo@... >> >> >> >> >> >> >> >> >> >> >> >> >> >> >>  >> >> >> ********************************** >> >> >> Prasad Matukumilli >> >> >> Research Scholar, >> >> >> Advanced Technology Development Centre, >> >> >> Indian Institute of Technology (IIT) >> >> >> Kharagpur  721302, India >> >> >> ********************************** >> >> > >> >> > >> >> > >> >> > >> >> >  >> >> > ********************************** >> >> > Prasad Matukumilli >> >> > Research Scholar, >> >> > Advanced Technology Development Centre, >> >> > Indian Institute of Technology (IIT) >> >> > Kharagpur  721302, India >> >> > ********************************** >> >> >> >> >> >> >> >>  >> >> Atsushi Togo >> >> http://atztogo.github.com/ >> >> atz.togo@... >> > >> > >> > >> > >> >  >> > ********************************** >> > Prasad Matukumilli >> > Research Scholar, >> > Advanced Technology Development Centre, >> > Indian Institute of Technology (IIT) >> > Kharagpur  721302, India >> > ********************************** >> >> >> >>  >> Atsushi Togo >> http://atztogo.github.com/ >> atz.togo@... > > > > >  > ********************************** > Prasad Matukumilli > Research Scholar, > Advanced Technology Development Centre, > Indian Institute of Technology (IIT) > Kharagpur  721302, India > **********************************  Atsushi Togo http://atztogo.github.com/ atz.togo@... 
From: Prasad Matukumilli <prasadchoudary@gm...>  20130829 21:00:24

Dear Dr.Togo, My system is CNT. For 6,6 cnt, dot product scheme working fine but it's failing for 10,10 cnt to sort the phonon modes. Combination of schemes might work well. Which method would you suggest to use along with dot product scheme? Using characters, how can I proceed to compute trace of the representation(which I dont know how to get that representation square matrix out of eigen vector) Is there a kind transformation needed? Prasad On Thu, Aug 29, 2013 at 7:57 PM, Atsushi Togo <atz.togo@...> wrote: > Hi, > > Without concrete example, I can't say anything on this problem. > Probably for complicate cases, a mixture of techniques would be > helpful, but I need to write code and test it to know what is most > robust using examples. Especially to handle degenerate bands would be > problematic. > On characters, phonon modes on a band path have the same set of > characters of the irreducible representation. > > Togo > > On Thu, Aug 29, 2013 at 11:08 PM, Prasad Matukumilli > <prasadchoudary@...> wrote: > > Dear Dr.Togo, > > I am facing trouble in sorting of phonons while plotting phonon > dispersion. > > My present scheme similar to the one in phonopy (sorting according to > best > > match by comparing dot product of current eigenvector with all next > qpoints' > > eigenvectors) working almost fine when the system with small number of > > eigenvectors(due to less number of close matchings) but this scheme > failing > > when eigenvectors number is little more(where many pairs of degenerate > > eigenvectors and contains more confusion for sorting). Also if at one > > qpoint there is a wrong matching, it leads to many wrong predictions. I > am > > looking for some solutions address this. > > > > Can you please elaborate a little regarding usage of characters of phonon > > modes for this. > > > > Thanks, > > Prasad > > > > > > On Thu, Aug 29, 2013 at 6:26 PM, Atsushi Togo <atz.togo@...> > wrote: > >> > >> Hi Prasad, > >> > >> If you really expect more information, you should show your problem in > >> details. > >> > >> Togo > >> > >> On Wed, Aug 28, 2013 at 9:29 PM, Prasad Matukumilli > >> <prasadchoudary@...> wrote: > >> > Dear Sebastian, > >> > > >> > Thank you for quick response. > >> > Yes, it's just taking dot product. I am also using the same in my > code, > >> > but > >> > at the band connections this scheme is failing and resulting partly > >> > missing > >> > of some important branches from the dispersion. I would appreciate if > >> > you > >> > can provide some pointers for other schemes apart this dot product > >> > scheme. > >> > I also tried taking comparison of slopes along with dot product > scheme, > >> > but > >> > of no use and resulted in much incomplete dispersion. > >> > > >> > Thanks, > >> > Prasad > >> > > >> > > >> > On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen > >> > <sebastian@...> wrote: > >> >> > >> >> Dear Prasad > >> >> > >> >> > >> >> I think the function you are looking for is > "estimate_band_connection" > >> >> located in phonon/band_structure.py > >> >> > >> >> But basically it is just taking the dot product of all eigenvectors > of > >> >> two > >> >> adjacent points ("metric" array in code below). The eigenvectors with > >> >> largest dotproduct are most similar/"parallel" so they are assumed to > >> >> belong > >> >> to same phonon branch. > >> >> > >> >> The code is: > >> >> def estimate_band_connection(prev_eigvecs, eigvecs, prev_band_order): > >> >> metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) > >> >> connection_order = [] > >> >> indices = range(len(metric)) > >> >> indices.reverse() > >> >> for overlaps in metric: > >> >> maxval = 0 > >> >> for i in indices: > >> >> val = overlaps[i] > >> >> if i in connection_order: > >> >> continue > >> >> if val > maxval: > >> >> maxval = val > >> >> maxindex = i > >> >> connection_order.append(maxindex) > >> >> > >> >> band_order = [connection_order[x] for x in prev_band_order] > >> >> > >> >> return band_order > >> >> > >> >> > >> >> Cheers > >> >> Sebastian > >> >> > >> >> Sebastian Christensen > >> >> PhD student > >> >> Center for Materials Crystallography > >> >> Department of Chemistry & iNANO, Aarhus University > >> >> Office: 1511318 > >> >> > >> >> Langelandsgade 140, DK8000 Aarhus C, Denmark > >> >> Mail: sebastian@... > >> >> > >> >> ________________________________ > >> >> From: Prasad Matukumilli [prasadchoudary@...] > >> >> Sent: 27 August 2013 23:16 > >> >> To: Atsushi Togo > >> >> Cc: phonopyusers > >> >> Subject: Re: [Phonopyusers] method of sorting of phonon branches > >> >> > >> >> Dear Togo, > >> >> I have gone through the code, I failed to locate the lines for > >> >> comparison > >> >> of eigenvectors. As I am new to python, I request your help and time > to > >> >> direct me to those lines. > >> >> > >> >> Thanks > >> >> Prasad. > >> >> > >> >> > >> >> On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> > >> >> wrote: > >> >>> > >> >>> Hi, > >> >>> > >> >>> In the band connection in phonopy, eigenvectors at neighboring > >> >>> qpoints are compared and most similar ones are connected. It's not > >> >>> implemented in phonopy as a band connection method, but you may be > >> >>> able to use characters of phonon modes > >> >>> > >> >>> > >> >>> > http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes > >> >>> . > >> >>> In group velocity, a perturbation approach same as Hellmann–Feynman > >> >>> theorem is used, so sorting is usually unnecessary. > >> >>> http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . > >> >>> > >> >>> For more details, please read the code. > >> >>> > >> >>> Togo > >> >>> > >> >>> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli > >> >>> <prasadchoudary@...> wrote: > >> >>> > Dear Dr.Togo, > >> >>> > > >> >>> > I am working on to code 'sorting of phonon branches' resulted out > >> >>> > from > >> >>> > the > >> >>> > eigen solver. > >> >>> > I request you to explain the outline of method behind the sorting > of > >> >>> > phonon > >> >>> > branches which has been implemented in Phonopy for the features of > >> >>> > band_connection or group velocity. > >> >>> > > >> >>> > Thank you. > >> >>> >  > >> >>> > ********************************** > >> >>> > Prasad Matukumilli > >> >>> > Research Scholar, > >> >>> > Advanced Technology Development Centre, > >> >>> > Indian Institute of Technology (IIT) > >> >>> > Kharagpur  721302, India > >> >>> > ********************************** > >> >>> > > >> >>> > > >> >>> > > >> >>> > >  > >> >>> > Introducing Performance Central, a new site from SourceForge and > >> >>> > AppDynamics. Performance Central is your source for news, > insights, > >> >>> > analysis and resources for efficient Application Performance > >> >>> > Management. > >> >>> > Visit us today! > >> >>> > > >> >>> > > >> >>> > > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > >> >>> > _______________________________________________ > >> >>> > Phonopyusers mailing list > >> >>> > Phonopyusers@... > >> >>> > https://lists.sourceforge.net/lists/listinfo/phonopyusers > >> >>> > > >> >>> > >> >>> > >> >>> > >> >>>  > >> >>> Atsushi Togo > >> >>> http://atztogo.github.com/ > >> >>> atz.togo@... > >> >> > >> >> > >> >> > >> >> > >> >>  > >> >> ********************************** > >> >> Prasad Matukumilli > >> >> Research Scholar, > >> >> Advanced Technology Development Centre, > >> >> Indian Institute of Technology (IIT) > >> >> Kharagpur  721302, India > >> >> ********************************** > >> > > >> > > >> > > >> > > >> >  > >> > ********************************** > >> > Prasad Matukumilli > >> > Research Scholar, > >> > Advanced Technology Development Centre, > >> > Indian Institute of Technology (IIT) > >> > Kharagpur  721302, India > >> > ********************************** > >> > >> > >> > >>  > >> Atsushi Togo > >> http://atztogo.github.com/ > >> atz.togo@... > > > > > > > > > >  > > ********************************** > > Prasad Matukumilli > > Research Scholar, > > Advanced Technology Development Centre, > > Indian Institute of Technology (IIT) > > Kharagpur  721302, India > > ********************************** > > > >  > Atsushi Togo > http://atztogo.github.com/ > atz.togo@... >  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Atsushi Togo <atz.togo@gm...>  20130829 14:27:42

Hi, Without concrete example, I can't say anything on this problem. Probably for complicate cases, a mixture of techniques would be helpful, but I need to write code and test it to know what is most robust using examples. Especially to handle degenerate bands would be problematic. On characters, phonon modes on a band path have the same set of characters of the irreducible representation. Togo On Thu, Aug 29, 2013 at 11:08 PM, Prasad Matukumilli <prasadchoudary@...> wrote: > Dear Dr.Togo, > I am facing trouble in sorting of phonons while plotting phonon dispersion. > My present scheme similar to the one in phonopy (sorting according to best > match by comparing dot product of current eigenvector with all next qpoints' > eigenvectors) working almost fine when the system with small number of > eigenvectors(due to less number of close matchings) but this scheme failing > when eigenvectors number is little more(where many pairs of degenerate > eigenvectors and contains more confusion for sorting). Also if at one > qpoint there is a wrong matching, it leads to many wrong predictions. I am > looking for some solutions address this. > > Can you please elaborate a little regarding usage of characters of phonon > modes for this. > > Thanks, > Prasad > > > On Thu, Aug 29, 2013 at 6:26 PM, Atsushi Togo <atz.togo@...> wrote: >> >> Hi Prasad, >> >> If you really expect more information, you should show your problem in >> details. >> >> Togo >> >> On Wed, Aug 28, 2013 at 9:29 PM, Prasad Matukumilli >> <prasadchoudary@...> wrote: >> > Dear Sebastian, >> > >> > Thank you for quick response. >> > Yes, it's just taking dot product. I am also using the same in my code, >> > but >> > at the band connections this scheme is failing and resulting partly >> > missing >> > of some important branches from the dispersion. I would appreciate if >> > you >> > can provide some pointers for other schemes apart this dot product >> > scheme. >> > I also tried taking comparison of slopes along with dot product scheme, >> > but >> > of no use and resulted in much incomplete dispersion. >> > >> > Thanks, >> > Prasad >> > >> > >> > On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen >> > <sebastian@...> wrote: >> >> >> >> Dear Prasad >> >> >> >> >> >> I think the function you are looking for is "estimate_band_connection" >> >> located in phonon/band_structure.py >> >> >> >> But basically it is just taking the dot product of all eigenvectors of >> >> two >> >> adjacent points ("metric" array in code below). The eigenvectors with >> >> largest dotproduct are most similar/"parallel" so they are assumed to >> >> belong >> >> to same phonon branch. >> >> >> >> The code is: >> >> def estimate_band_connection(prev_eigvecs, eigvecs, prev_band_order): >> >> metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) >> >> connection_order = [] >> >> indices = range(len(metric)) >> >> indices.reverse() >> >> for overlaps in metric: >> >> maxval = 0 >> >> for i in indices: >> >> val = overlaps[i] >> >> if i in connection_order: >> >> continue >> >> if val > maxval: >> >> maxval = val >> >> maxindex = i >> >> connection_order.append(maxindex) >> >> >> >> band_order = [connection_order[x] for x in prev_band_order] >> >> >> >> return band_order >> >> >> >> >> >> Cheers >> >> Sebastian >> >> >> >> Sebastian Christensen >> >> PhD student >> >> Center for Materials Crystallography >> >> Department of Chemistry & iNANO, Aarhus University >> >> Office: 1511318 >> >> >> >> Langelandsgade 140, DK8000 Aarhus C, Denmark >> >> Mail: sebastian@... >> >> >> >> ________________________________ >> >> From: Prasad Matukumilli [prasadchoudary@...] >> >> Sent: 27 August 2013 23:16 >> >> To: Atsushi Togo >> >> Cc: phonopyusers >> >> Subject: Re: [Phonopyusers] method of sorting of phonon branches >> >> >> >> Dear Togo, >> >> I have gone through the code, I failed to locate the lines for >> >> comparison >> >> of eigenvectors. As I am new to python, I request your help and time to >> >> direct me to those lines. >> >> >> >> Thanks >> >> Prasad. >> >> >> >> >> >> On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> >> >> wrote: >> >>> >> >>> Hi, >> >>> >> >>> In the band connection in phonopy, eigenvectors at neighboring >> >>> qpoints are compared and most similar ones are connected. It's not >> >>> implemented in phonopy as a band connection method, but you may be >> >>> able to use characters of phonon modes >> >>> >> >>> >> >>> http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes >> >>> . >> >>> In group velocity, a perturbation approach same as Hellmann–Feynman >> >>> theorem is used, so sorting is usually unnecessary. >> >>> http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . >> >>> >> >>> For more details, please read the code. >> >>> >> >>> Togo >> >>> >> >>> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli >> >>> <prasadchoudary@...> wrote: >> >>> > Dear Dr.Togo, >> >>> > >> >>> > I am working on to code 'sorting of phonon branches' resulted out >> >>> > from >> >>> > the >> >>> > eigen solver. >> >>> > I request you to explain the outline of method behind the sorting of >> >>> > phonon >> >>> > branches which has been implemented in Phonopy for the features of >> >>> > band_connection or group velocity. >> >>> > >> >>> > Thank you. >> >>> >  >> >>> > ********************************** >> >>> > Prasad Matukumilli >> >>> > Research Scholar, >> >>> > Advanced Technology Development Centre, >> >>> > Indian Institute of Technology (IIT) >> >>> > Kharagpur  721302, India >> >>> > ********************************** >> >>> > >> >>> > >> >>> > >> >>> >  >> >>> > Introducing Performance Central, a new site from SourceForge and >> >>> > AppDynamics. Performance Central is your source for news, insights, >> >>> > analysis and resources for efficient Application Performance >> >>> > Management. >> >>> > Visit us today! >> >>> > >> >>> > >> >>> > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk >> >>> > _______________________________________________ >> >>> > Phonopyusers mailing list >> >>> > Phonopyusers@... >> >>> > https://lists.sourceforge.net/lists/listinfo/phonopyusers >> >>> > >> >>> >> >>> >> >>> >> >>>  >> >>> Atsushi Togo >> >>> http://atztogo.github.com/ >> >>> atz.togo@... >> >> >> >> >> >> >> >> >> >>  >> >> ********************************** >> >> Prasad Matukumilli >> >> Research Scholar, >> >> Advanced Technology Development Centre, >> >> Indian Institute of Technology (IIT) >> >> Kharagpur  721302, India >> >> ********************************** >> > >> > >> > >> > >> >  >> > ********************************** >> > Prasad Matukumilli >> > Research Scholar, >> > Advanced Technology Development Centre, >> > Indian Institute of Technology (IIT) >> > Kharagpur  721302, India >> > ********************************** >> >> >> >>  >> Atsushi Togo >> http://atztogo.github.com/ >> atz.togo@... > > > > >  > ********************************** > Prasad Matukumilli > Research Scholar, > Advanced Technology Development Centre, > Indian Institute of Technology (IIT) > Kharagpur  721302, India > **********************************  Atsushi Togo http://atztogo.github.com/ atz.togo@... 
From: Prasad Matukumilli <prasadchoudary@gm...>  20130829 14:09:15

Dear Dr.Togo, I am facing trouble in sorting of phonons while plotting phonon dispersion. My present scheme similar to the one in phonopy (sorting according to best match by comparing dot product of current eigenvector with all next qpoints' eigenvectors) working almost fine when the system with small number of eigenvectors(due to less number of close matchings) but this scheme failing when eigenvectors number is little more(where many pairs of degenerate eigenvectors and contains more confusion for sorting). Also if at one qpoint there is a wrong matching, it leads to many wrong predictions. I am looking for some solutions address this. Can you please elaborate a little regarding usage of characters of phonon modes for this. Thanks, Prasad On Thu, Aug 29, 2013 at 6:26 PM, Atsushi Togo <atz.togo@...> wrote: > Hi Prasad, > > If you really expect more information, you should show your problem in > details. > > Togo > > On Wed, Aug 28, 2013 at 9:29 PM, Prasad Matukumilli > <prasadchoudary@...> wrote: > > Dear Sebastian, > > > > Thank you for quick response. > > Yes, it's just taking dot product. I am also using the same in my code, > but > > at the band connections this scheme is failing and resulting partly > missing > > of some important branches from the dispersion. I would appreciate if you > > can provide some pointers for other schemes apart this dot product > scheme. > > I also tried taking comparison of slopes along with dot product scheme, > but > > of no use and resulted in much incomplete dispersion. > > > > Thanks, > > Prasad > > > > > > On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen > > <sebastian@...> wrote: > >> > >> Dear Prasad > >> > >> > >> I think the function you are looking for is "estimate_band_connection" > >> located in phonon/band_structure.py > >> > >> But basically it is just taking the dot product of all eigenvectors of > two > >> adjacent points ("metric" array in code below). The eigenvectors with > >> largest dotproduct are most similar/"parallel" so they are assumed to > belong > >> to same phonon branch. > >> > >> The code is: > >> def estimate_band_connection(prev_eigvecs, eigvecs, prev_band_order): > >> metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) > >> connection_order = [] > >> indices = range(len(metric)) > >> indices.reverse() > >> for overlaps in metric: > >> maxval = 0 > >> for i in indices: > >> val = overlaps[i] > >> if i in connection_order: > >> continue > >> if val > maxval: > >> maxval = val > >> maxindex = i > >> connection_order.append(maxindex) > >> > >> band_order = [connection_order[x] for x in prev_band_order] > >> > >> return band_order > >> > >> > >> Cheers > >> Sebastian > >> > >> Sebastian Christensen > >> PhD student > >> Center for Materials Crystallography > >> Department of Chemistry & iNANO, Aarhus University > >> Office: 1511318 > >> > >> Langelandsgade 140, DK8000 Aarhus C, Denmark > >> Mail: sebastian@... > >> > >> ________________________________ > >> From: Prasad Matukumilli [prasadchoudary@...] > >> Sent: 27 August 2013 23:16 > >> To: Atsushi Togo > >> Cc: phonopyusers > >> Subject: Re: [Phonopyusers] method of sorting of phonon branches > >> > >> Dear Togo, > >> I have gone through the code, I failed to locate the lines for > comparison > >> of eigenvectors. As I am new to python, I request your help and time to > >> direct me to those lines. > >> > >> Thanks > >> Prasad. > >> > >> > >> On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> > wrote: > >>> > >>> Hi, > >>> > >>> In the band connection in phonopy, eigenvectors at neighboring > >>> qpoints are compared and most similar ones are connected. It's not > >>> implemented in phonopy as a band connection method, but you may be > >>> able to use characters of phonon modes > >>> > >>> > http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes > >>> . > >>> In group velocity, a perturbation approach same as Hellmann–Feynman > >>> theorem is used, so sorting is usually unnecessary. > >>> http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . > >>> > >>> For more details, please read the code. > >>> > >>> Togo > >>> > >>> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli > >>> <prasadchoudary@...> wrote: > >>> > Dear Dr.Togo, > >>> > > >>> > I am working on to code 'sorting of phonon branches' resulted out > from > >>> > the > >>> > eigen solver. > >>> > I request you to explain the outline of method behind the sorting of > >>> > phonon > >>> > branches which has been implemented in Phonopy for the features of > >>> > band_connection or group velocity. > >>> > > >>> > Thank you. > >>> >  > >>> > ********************************** > >>> > Prasad Matukumilli > >>> > Research Scholar, > >>> > Advanced Technology Development Centre, > >>> > Indian Institute of Technology (IIT) > >>> > Kharagpur  721302, India > >>> > ********************************** > >>> > > >>> > > >>> > >  > >>> > Introducing Performance Central, a new site from SourceForge and > >>> > AppDynamics. Performance Central is your source for news, insights, > >>> > analysis and resources for efficient Application Performance > >>> > Management. > >>> > Visit us today! > >>> > > >>> > > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > >>> > _______________________________________________ > >>> > Phonopyusers mailing list > >>> > Phonopyusers@... > >>> > https://lists.sourceforge.net/lists/listinfo/phonopyusers > >>> > > >>> > >>> > >>> > >>>  > >>> Atsushi Togo > >>> http://atztogo.github.com/ > >>> atz.togo@... > >> > >> > >> > >> > >>  > >> ********************************** > >> Prasad Matukumilli > >> Research Scholar, > >> Advanced Technology Development Centre, > >> Indian Institute of Technology (IIT) > >> Kharagpur  721302, India > >> ********************************** > > > > > > > > > >  > > ********************************** > > Prasad Matukumilli > > Research Scholar, > > Advanced Technology Development Centre, > > Indian Institute of Technology (IIT) > > Kharagpur  721302, India > > ********************************** > > > >  > Atsushi Togo > http://atztogo.github.com/ > atz.togo@... >  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Atsushi Togo <atz.togo@gm...>  20130829 12:56:09

Hi Prasad, If you really expect more information, you should show your problem in details. Togo On Wed, Aug 28, 2013 at 9:29 PM, Prasad Matukumilli <prasadchoudary@...> wrote: > Dear Sebastian, > > Thank you for quick response. > Yes, it's just taking dot product. I am also using the same in my code, but > at the band connections this scheme is failing and resulting partly missing > of some important branches from the dispersion. I would appreciate if you > can provide some pointers for other schemes apart this dot product scheme. > I also tried taking comparison of slopes along with dot product scheme, but > of no use and resulted in much incomplete dispersion. > > Thanks, > Prasad > > > On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen > <sebastian@...> wrote: >> >> Dear Prasad >> >> >> I think the function you are looking for is "estimate_band_connection" >> located in phonon/band_structure.py >> >> But basically it is just taking the dot product of all eigenvectors of two >> adjacent points ("metric" array in code below). The eigenvectors with >> largest dotproduct are most similar/"parallel" so they are assumed to belong >> to same phonon branch. >> >> The code is: >> def estimate_band_connection(prev_eigvecs, eigvecs, prev_band_order): >> metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) >> connection_order = [] >> indices = range(len(metric)) >> indices.reverse() >> for overlaps in metric: >> maxval = 0 >> for i in indices: >> val = overlaps[i] >> if i in connection_order: >> continue >> if val > maxval: >> maxval = val >> maxindex = i >> connection_order.append(maxindex) >> >> band_order = [connection_order[x] for x in prev_band_order] >> >> return band_order >> >> >> Cheers >> Sebastian >> >> Sebastian Christensen >> PhD student >> Center for Materials Crystallography >> Department of Chemistry & iNANO, Aarhus University >> Office: 1511318 >> >> Langelandsgade 140, DK8000 Aarhus C, Denmark >> Mail: sebastian@... >> >> ________________________________ >> From: Prasad Matukumilli [prasadchoudary@...] >> Sent: 27 August 2013 23:16 >> To: Atsushi Togo >> Cc: phonopyusers >> Subject: Re: [Phonopyusers] method of sorting of phonon branches >> >> Dear Togo, >> I have gone through the code, I failed to locate the lines for comparison >> of eigenvectors. As I am new to python, I request your help and time to >> direct me to those lines. >> >> Thanks >> Prasad. >> >> >> On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> wrote: >>> >>> Hi, >>> >>> In the band connection in phonopy, eigenvectors at neighboring >>> qpoints are compared and most similar ones are connected. It's not >>> implemented in phonopy as a band connection method, but you may be >>> able to use characters of phonon modes >>> >>> http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes >>> . >>> In group velocity, a perturbation approach same as Hellmann–Feynman >>> theorem is used, so sorting is usually unnecessary. >>> http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . >>> >>> For more details, please read the code. >>> >>> Togo >>> >>> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli >>> <prasadchoudary@...> wrote: >>> > Dear Dr.Togo, >>> > >>> > I am working on to code 'sorting of phonon branches' resulted out from >>> > the >>> > eigen solver. >>> > I request you to explain the outline of method behind the sorting of >>> > phonon >>> > branches which has been implemented in Phonopy for the features of >>> > band_connection or group velocity. >>> > >>> > Thank you. >>> >  >>> > ********************************** >>> > Prasad Matukumilli >>> > Research Scholar, >>> > Advanced Technology Development Centre, >>> > Indian Institute of Technology (IIT) >>> > Kharagpur  721302, India >>> > ********************************** >>> > >>> > >>> >  >>> > Introducing Performance Central, a new site from SourceForge and >>> > AppDynamics. Performance Central is your source for news, insights, >>> > analysis and resources for efficient Application Performance >>> > Management. >>> > Visit us today! >>> > >>> > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk >>> > _______________________________________________ >>> > Phonopyusers mailing list >>> > Phonopyusers@... >>> > https://lists.sourceforge.net/lists/listinfo/phonopyusers >>> > >>> >>> >>> >>>  >>> Atsushi Togo >>> http://atztogo.github.com/ >>> atz.togo@... >> >> >> >> >>  >> ********************************** >> Prasad Matukumilli >> Research Scholar, >> Advanced Technology Development Centre, >> Indian Institute of Technology (IIT) >> Kharagpur  721302, India >> ********************************** > > > > >  > ********************************** > Prasad Matukumilli > Research Scholar, > Advanced Technology Development Centre, > Indian Institute of Technology (IIT) > Kharagpur  721302, India > **********************************  Atsushi Togo http://atztogo.github.com/ atz.togo@... 
From: Prasad Matukumilli <prasadchoudary@gm...>  20130828 12:30:07

Dear Sebastian, Thank you for quick response. Yes, it's just taking dot product. I am also using the same in my code, but at the band connections this scheme is failing and resulting partly missing of some important branches from the dispersion. I would appreciate if you can provide some pointers for other schemes apart this dot product scheme. I also tried taking comparison of slopes along with dot product scheme, but of no use and resulted in much incomplete dispersion. Thanks, Prasad On Wed, Aug 28, 2013 at 3:58 PM, Sebastian Christensen <sebastian@... > wrote: > Dear Prasad > > > I think the function you are looking for is "estimate_band_connection" > located in phonon/band_structure.py > > But basically it is just taking the dot product of all eigenvectors of > two adjacent points ("metric" array in code below). The eigenvectors with > largest dotproduct are most similar/"parallel" so they are assumed to > belong to same phonon branch. > > The code is: > def estimate_band_connection(prev_eigvecs, eigvecs, prev_band_order): > metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) > connection_order = [] > indices = range(len(metric)) > indices.reverse() > for overlaps in metric: > maxval = 0 > for i in indices: > val = overlaps[i] > if i in connection_order: > continue > if val > maxval: > maxval = val > maxindex = i > connection_order.append(maxindex) > > band_order = [connection_order[x] for x in prev_band_order] > > return band_order > > > Cheers > Sebastian > > Sebastian Christensen > PhD student > Center for Materials Crystallography > Department of Chemistry & iNANO, Aarhus University > Office: 1511318 > > Langelandsgade 140, DK8000 Aarhus C, Denmark > Mail: sebastian@... >  > *From:* Prasad Matukumilli [prasadchoudary@...] > *Sent:* 27 August 2013 23:16 > *To:* Atsushi Togo > *Cc:* phonopyusers > *Subject:* Re: [Phonopyusers] method of sorting of phonon branches > > Dear Togo, > I have gone through the code, I failed to locate the lines for comparison > of eigenvectors. As I am new to python, I request your help and time to > direct me to those lines. > > Thanks > Prasad. > > > On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> wrote: > >> Hi, >> >> In the band connection in phonopy, eigenvectors at neighboring >> qpoints are compared and most similar ones are connected. It's not >> implemented in phonopy as a band connection method, but you may be >> able to use characters of phonon modes >> >> http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes >> . >> In group velocity, a perturbation approach same as Hellmann–Feynman >> theorem is used, so sorting is usually unnecessary. >> http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . >> >> For more details, please read the code. >> >> Togo >> >> On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli >> <prasadchoudary@...> wrote: >> > Dear Dr.Togo, >> > >> > I am working on to code 'sorting of phonon branches' resulted out from >> the >> > eigen solver. >> > I request you to explain the outline of method behind the sorting of >> phonon >> > branches which has been implemented in Phonopy for the features of >> > band_connection or group velocity. >> > >> > Thank you. >> >  >> > ********************************** >> > Prasad Matukumilli >> > Research Scholar, >> > Advanced Technology Development Centre, >> > Indian Institute of Technology (IIT) >> > Kharagpur  721302, India >> > ********************************** >> > >> > >>  >> > Introducing Performance Central, a new site from SourceForge and >> > AppDynamics. Performance Central is your source for news, insights, >> > analysis and resources for efficient Application Performance Management. >> > Visit us today! >> > >> http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk >> > _______________________________________________ >> > Phonopyusers mailing list >> > Phonopyusers@... >> > https://lists.sourceforge.net/lists/listinfo/phonopyusers >> > >> >> >> >>  >> Atsushi Togo >> http://atztogo.github.com/ >> atz.togo@... >> > > > >  > ********************************** > Prasad Matukumilli > Research Scholar, > Advanced Technology Development Centre, > Indian Institute of Technology (IIT) > Kharagpur  721302, India > ********************************** >  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Sebastian Christensen <sebastian@ch...>  20130828 10:41:48

Dear Prasad I think the function you are looking for is "estimate_band_connection" located in phonon/band_structure.py But basically it is just taking the dot product of all eigenvectors of two adjacent points ("metric" array in code below). The eigenvectors with largest dotproduct are most similar/"parallel" so they are assumed to belong to same phonon branch. The code is: def estimate_band_connection(prev_eigvecs, eigvecs, prev_band_order): metric = np.abs(np.dot(prev_eigvecs.conjugate().T, eigvecs)) connection_order = [] indices = range(len(metric)) indices.reverse() for overlaps in metric: maxval = 0 for i in indices: val = overlaps[i] if i in connection_order: continue if val > maxval: maxval = val maxindex = i connection_order.append(maxindex) band_order = [connection_order[x] for x in prev_band_order] return band_order Cheers Sebastian Sebastian Christensen PhD student Center for Materials Crystallography Department of Chemistry & iNANO, Aarhus University Office: 1511318 Langelandsgade 140, DK8000 Aarhus C, Denmark Mail: sebastian@...<mailto:sebastian@...> ________________________________ From: Prasad Matukumilli [prasadchoudary@...] Sent: 27 August 2013 23:16 To: Atsushi Togo Cc: phonopyusers Subject: Re: [Phonopyusers] method of sorting of phonon branches Dear Togo, I have gone through the code, I failed to locate the lines for comparison of eigenvectors. As I am new to python, I request your help and time to direct me to those lines. Thanks Prasad. On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...<mailto:atz.togo@...>> wrote: Hi, In the band connection in phonopy, eigenvectors at neighboring qpoints are compared and most similar ones are connected. It's not implemented in phonopy as a band connection method, but you may be able to use characters of phonon modes http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes . In group velocity, a perturbation approach same as Hellmann–Feynman theorem is used, so sorting is usually unnecessary. http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . For more details, please read the code. Togo On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli <prasadchoudary@...<mailto:prasadchoudary@...>> wrote: > Dear Dr.Togo, > > I am working on to code 'sorting of phonon branches' resulted out from the > eigen solver. > I request you to explain the outline of method behind the sorting of phonon > branches which has been implemented in Phonopy for the features of > band_connection or group velocity. > > Thank you. >  > ********************************** > Prasad Matukumilli > Research Scholar, > Advanced Technology Development Centre, > Indian Institute of Technology (IIT) > Kharagpur  721302, India > ********************************** > >  > Introducing Performance Central, a new site from SourceForge and > AppDynamics. Performance Central is your source for news, insights, > analysis and resources for efficient Application Performance Management. > Visit us today! > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > _______________________________________________ > Phonopyusers mailing list > Phonopyusers@...<mailto:Phonopyusers@...> > https://lists.sourceforge.net/lists/listinfo/phonopyusers >  Atsushi Togo http://atztogo.github.com/ atz.togo@...<mailto:atz.togo@...>  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Prasad Matukumilli <prasadchoudary@gm...>  20130827 21:17:21

Dear Togo, I have gone through the code, I failed to locate the lines for comparison of eigenvectors. As I am new to python, I request your help and time to direct me to those lines. Thanks Prasad. On Sun, Aug 25, 2013 at 6:51 PM, Atsushi Togo <atz.togo@...> wrote: > Hi, > > In the band connection in phonopy, eigenvectors at neighboring > qpoints are compared and most similar ones are connected. It's not > implemented in phonopy as a band connection method, but you may be > able to use characters of phonon modes > > http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes > . > In group velocity, a perturbation approach same as Hellmann–Feynman > theorem is used, so sorting is usually unnecessary. > http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . > > For more details, please read the code. > > Togo > > On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli > <prasadchoudary@...> wrote: > > Dear Dr.Togo, > > > > I am working on to code 'sorting of phonon branches' resulted out from > the > > eigen solver. > > I request you to explain the outline of method behind the sorting of > phonon > > branches which has been implemented in Phonopy for the features of > > band_connection or group velocity. > > > > Thank you. > >  > > ********************************** > > Prasad Matukumilli > > Research Scholar, > > Advanced Technology Development Centre, > > Indian Institute of Technology (IIT) > > Kharagpur  721302, India > > ********************************** > > > > >  > > Introducing Performance Central, a new site from SourceForge and > > AppDynamics. Performance Central is your source for news, insights, > > analysis and resources for efficient Application Performance Management. > > Visit us today! > > > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > > _______________________________________________ > > Phonopyusers mailing list > > Phonopyusers@... > > https://lists.sourceforge.net/lists/listinfo/phonopyusers > > > > > >  > Atsushi Togo > http://atztogo.github.com/ > atz.togo@... >  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Atsushi Togo <atz.togo@gm...>  20130825 13:21:29

Hi, In the band connection in phonopy, eigenvectors at neighboring qpoints are compared and most similar ones are connected. It's not implemented in phonopy as a band connection method, but you may be able to use characters of phonon modes http://phonopy.sourceforge.net/settingtags.html#charactersofirreduciblerepresentationsofphononmodes . In group velocity, a perturbation approach same as Hellmann–Feynman theorem is used, so sorting is usually unnecessary. http://phonopy.sourceforge.net/groupvelocity.html#groupvelocity . For more details, please read the code. Togo On Sun, Aug 25, 2013 at 9:59 PM, Prasad Matukumilli <prasadchoudary@...> wrote: > Dear Dr.Togo, > > I am working on to code 'sorting of phonon branches' resulted out from the > eigen solver. > I request you to explain the outline of method behind the sorting of phonon > branches which has been implemented in Phonopy for the features of > band_connection or group velocity. > > Thank you. >  > ********************************** > Prasad Matukumilli > Research Scholar, > Advanced Technology Development Centre, > Indian Institute of Technology (IIT) > Kharagpur  721302, India > ********************************** > >  > Introducing Performance Central, a new site from SourceForge and > AppDynamics. Performance Central is your source for news, insights, > analysis and resources for efficient Application Performance Management. > Visit us today! > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > _______________________________________________ > Phonopyusers mailing list > Phonopyusers@... > https://lists.sourceforge.net/lists/listinfo/phonopyusers >  Atsushi Togo http://atztogo.github.com/ atz.togo@... 
From: Atsushi Togo <atz.togo@gm...>  20130825 13:11:06

Hi, Alejandro explains all. In the case of supercell approach, finite displacement of an atom appears periodicity of supercell, so it has to be considered though. It's considered as an interpolation scheme, i.e., except for commensurate points (e.g. q=(0.5, 0, 0) for 2x1x1 supercell), they are not exact. This approximation works because usually atomic interaction is short range. If the unit cell is huge, the interaction from an atom to an atom outside of unit cell is almost zero. In this case, even using unit cell only, this approach works well. But electrostatic polarization is an exception and would be treated differently anyway. Togo On Fri, Aug 23, 2013 at 4:16 PM, Santu Baidya <santubaidya2009@...>wrote: > > >  Forwarded message  > From: Santu Baidya <santubaidya2009@...> > Date: 23 August 2013 12:44 > Subject: Re: [Phonopyusers] How from a single unit cell (not supercell) > phonon dispersion appears? > To: Alejandro Rébola <arebol2@...> > > > Dear phonopy users, > > I have got an explanation for the single unit cell phonon calculation > using phonopy and I need to know if that is correct or not. > > From phonopy manual I found that > A force and a secondorder force constant [image: \Phi_{\alpha \beta}]are given by > > [image: F_\alpha(jl) = \frac{\partial V }{\partial r_\alpha(jl)}] > where l is an unit cell and > > [image: \Phi_{\alpha\beta}(jl, j'l') = \frac{\partial^2 V}{\partial > r_\alpha(jl) \partial r_\beta(j'l')} = \frac{\partial > F_\beta(j'l')}{\partial r_\alpha(jl)},] > For single unit cell it will be l = l' > > VASP gives Force on atoms although at single qpoint (by DFPT method) also > mentioned in phonopy manual. > > phonopy calculates Force constant matrix from the Forces. > > Then Dynamical Matrix form will be > > D (jj', q) = 1/√(mj mj' ) [image: \Phi_{\alpha \beta}] (j0, j'0) exp( > iq . [r(j'0)  r(j0)]) > > > > And then from here we get dynamical matrix for pair interactions and then > solve the dynamical eqn of motion. > > We get then phonon frequencies at other qpoints also. > > > Could any one please tell me if this is correct explanation or not. > > > Thanks. > > Santu Baidya > > > > > > > > > > > On 23 August 2013 11:19, Santu Baidya <santubaidya2009@...> wrote: > >> Dear Alejandro, >> >> Thanks for your reply . Ofcourse it helps me to understand. >> >> Can I ask you then if I use original unit cell (not supercell) whose >> lattice parameters ~ 8 Angstrom (large enough) and do not make any >> supercell and do a VASP calculation (GAMMA point ) to get forces within >> that range. >> >> And then I use phonopy I shall get frequencies even at other Kponts. >> >> Is that right? >> >> Could you please explain it little more. >> >> Thanks again. >> >> Santu Baidya >> >> >> > > >  > *"The happiest people do not always have the best of all,* * they simply > appreciate what they find on their way!!!" SANTU > * > > > >  > *"The happiest people do not always have the best of all,* * they simply > appreciate what they find on their way!!!" SANTU > * > > >  > Introducing Performance Central, a new site from SourceForge and > AppDynamics. Performance Central is your source for news, insights, > analysis and resources for efficient Application Performance Management. > Visit us today! > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > _______________________________________________ > Phonopyusers mailing list > Phonopyusers@... > https://lists.sourceforge.net/lists/listinfo/phonopyusers > >  Atsushi Togo http://atztogo.github.com/ atz.togo@... 
From: Prasad Matukumilli <prasadchoudary@gm...>  20130825 13:00:03

Dear Dr.Togo, I am working on to code 'sorting of phonon branches' resulted out from the eigen solver. I request you to explain the outline of method behind the sorting of phonon branches which has been implemented in Phonopy for the features of band_connection or group velocity. Thank you.  ********************************** Prasad Matukumilli Research Scholar, Advanced Technology Development Centre, Indian Institute of Technology (IIT) Kharagpur  721302, India ********************************** 
From: Santu Baidya <santubaidya2009@gm...>  20130823 07:16:15

 Forwarded message  From: Santu Baidya <santubaidya2009@...> Date: 23 August 2013 12:44 Subject: Re: [Phonopyusers] How from a single unit cell (not supercell) phonon dispersion appears? To: Alejandro Rébola <arebol2@...> Dear phonopy users, I have got an explanation for the single unit cell phonon calculation using phonopy and I need to know if that is correct or not. >From phonopy manual I found that A force and a secondorder force constant [image: \Phi_{\alpha \beta}] are given by [image: F_\alpha(jl) = \frac{\partial V }{\partial r_\alpha(jl)}] where l is an unit cell and [image: \Phi_{\alpha\beta}(jl, j'l') = \frac{\partial^2 V}{\partial r_\alpha(jl) \partial r_\beta(j'l')} = \frac{\partial F_\beta(j'l')}{\partial r_\alpha(jl)},] For single unit cell it will be l = l' VASP gives Force on atoms although at single qpoint (by DFPT method) also mentioned in phonopy manual. phonopy calculates Force constant matrix from the Forces. Then Dynamical Matrix form will be D (jj', q) = 1/√(mj mj' ) [image: \Phi_{\alpha \beta}] (j0, j'0) exp( iq . [r(j'0)  r(j0)]) And then from here we get dynamical matrix for pair interactions and then solve the dynamical eqn of motion. We get then phonon frequencies at other qpoints also. Could any one please tell me if this is correct explanation or not. Thanks. Santu Baidya On 23 August 2013 11:19, Santu Baidya <santubaidya2009@...> wrote: > Dear Alejandro, > > Thanks for your reply . Ofcourse it helps me to understand. > > Can I ask you then if I use original unit cell (not supercell) whose > lattice parameters ~ 8 Angstrom (large enough) and do not make any > supercell and do a VASP calculation (GAMMA point ) to get forces within > that range. > > And then I use phonopy I shall get frequencies even at other Kponts. > > Is that right? > > Could you please explain it little more. > > Thanks again. > > Santu Baidya > > >  *"The happiest people do not always have the best of all,* * they simply appreciate what they find on their way!!!" SANTU *  *"The happiest people do not always have the best of all,* * they simply appreciate what they find on their way!!!" SANTU * 
From: Alejandro Rébola <arebol2@ui...>  20130823 05:56:02

Dear Santu, The way phonopy works is by creating the dynamical matrix for each q [let's call it D(q)] and then diagonalizing it. To create D(q) in theory you'd need all the force constants for each lattice vector R (let's assume no basis). They are all the possible pair of interactions in the lattice. However since the number of R's is very large, this would be impossible to do in practice, so the summation has to be cut off at some point. What I understand phonopy is doing is to consider only interactions within the range of the supercell you are using. The larger the supercell you use the more accurate your calculation, also because in this way you reduce interactions between periodic images. Hope this helps. Best, Alejandro On Thu, August 22, 2013 11:50 pm, Santu Baidya wrote: > Dear phonopy users, > > I did a phonon band dispersion calculation without making a supercell > using > phonopy. > > First I used VASP, IBRION=8 (DFPT) to calculate forces. Then I used > phonopy > to get band dispersion. > > What I know for single unit cell I should always get GAMMA point phonons > only. But phonopy is capable of giving frequencies at other kpoints also > ! > > Could anyone please please explain it how is it possible. > > Thanks. > > Santu Baidya > > > > > *"The happiest people do not always have the best of all,* * they simply > appreciate what they find on their way!!!" SANTU > * >  > Introducing Performance Central, a new site from SourceForge and > AppDynamics. Performance Central is your source for news, insights, > analysis and resources for efficient Application Performance Management. > Visit us today! > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk_______________________________________________ > Phonopyusers mailing list > Phonopyusers@... > https://lists.sourceforge.net/lists/listinfo/phonopyusers > 
From: Santu Baidya <santubaidya2009@gm...>  20130823 04:50:47

Dear phonopy users, I did a phonon band dispersion calculation without making a supercell using phonopy. First I used VASP, IBRION=8 (DFPT) to calculate forces. Then I used phonopy to get band dispersion. What I know for single unit cell I should always get GAMMA point phonons only. But phonopy is capable of giving frequencies at other kpoints also ! Could anyone please please explain it how is it possible. Thanks. Santu Baidya *"The happiest people do not always have the best of all,* * they simply appreciate what they find on their way!!!" SANTU * 
From: Santu Baidya <santubaidya2009@gm...>  20130823 02:25:36

dear phonopy users, i have found that even if i do not make any supercell and use original unit cell to do phonon dispersion calculation phonopy gives band dispersion. My question is with single unit cell i should get only gamma point phonon how i am getting phonon frequencies at other k points?  *"The happiest people do not always have the best of all,* * they simply appreciate what they find on their way!!!" SANTU * 
From: Atsushi Togo <atz.togo@gm...>  20130808 05:33:56

Hi, Mean square displacement is calculated from the number of bosons excited (See http://phonopy.sourceforge.net/thermaldisplacement.html#thermaldisplacement) But in MODULATION tag, no such thing is considered at all, and you can not compare different phonon modes. Togo On Wed, Aug 7, 2013 at 9:17 PM, Aldo Humberto Romero <aromero@...> wrote: > Well... if these are two points in the Brillouin zone, probably they will > have > differente energies to excite the same energy mode (that is why we have > dispersion), therefore, it is expected that the same phonon branch changes > with q, E(q) after all!. > > > > Dear All, > > > > I would like to know how the amplitude of atom movement is determined > > for a modulation. I am using the "MODULATION" keyword to generate > > modulated structures due to modes at qvectors (1/2,0,0) and > > (1/2,1/2,0). For example my input file for the first of these is: > > ATOM_NAME = Mg Sn O > > DIM = 2 2 2 > > MODULATION = 1/2 0 0 2 2 2, 1 20 > > > > I find that the dominant atom movement ( which is due to Sn atoms in > > my system ) is much larger for q=(1/2, 1/2, 0) compared to q=(1/2,0,0) > > when I use the same amplitude. Can someone explain to me why this > > occurs? > > > > Many thanks, > > David. > >  > > > > Dr. David A. Tompsett > > Materials Chemistry Group > > Department of Chemistry > > University of Bath > > Bath BA2 7AY > > U.K. > > Tel: +44 7907 566351 (mobile) > > Fax: +44 1225 386231 > > http://people.bath.ac.uk/msi20/ > > > > > > >  > > Get 100% visibility into Java/.NET code with AppDynamics Lite! > > It's a free troubleshooting tool designed for production. > > Get down to codelevel detail for bottlenecks, with <2% overhead. > > Download for free and get started troubleshooting in minutes. > > > http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk > > _______________________________________________ > > Phonopyusers mailing list > > Phonopyusers@... > > https://lists.sourceforge.net/lists/listinfo/phonopyusers > > > > >  > > Prof. Aldo Humberto Romero > CINVESTAVUnidad Queretaro > Libramiento Norponiente 2000 > CP 76230, Queretaro, QRO, Mexico > tel: 442 211 9909 > fax: 442 211 9938 > > email: aromero@... > aldorome@... > www: qro.cinvestav.mx/~aromero > > >  > Get 100% visibility into Java/.NET code with AppDynamics Lite! > It's a free troubleshooting tool designed for production. > Get down to codelevel detail for bottlenecks, with <2% overhead. > Download for free and get started troubleshooting in minutes. > http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk > _______________________________________________ > Phonopyusers mailing list > Phonopyusers@... > https://lists.sourceforge.net/lists/listinfo/phonopyusers  Atsushi Togo http://atztogo.github.com/ atz.togo@... 
From: Aldo Humberto Romero <aromero@qr...>  20130807 12:38:37

Well... if these are two points in the Brillouin zone, probably they will have differente energies to excite the same energy mode (that is why we have dispersion), therefore, it is expected that the same phonon branch changes with q, E(q) after all!. > Dear All, > > I would like to know how the amplitude of atom movement is determined > for a modulation. I am using the "MODULATION" keyword to generate > modulated structures due to modes at qvectors (1/2,0,0) and > (1/2,1/2,0). For example my input file for the first of these is: > ATOM_NAME = Mg Sn O > DIM = 2 2 2 > MODULATION = 1/2 0 0 2 2 2, 1 20 > > I find that the dominant atom movement ( which is due to Sn atoms in > my system ) is much larger for q=(1/2, 1/2, 0) compared to q=(1/2,0,0) > when I use the same amplitude. Can someone explain to me why this > occurs? > > Many thanks, > David. >  > > Dr. David A. Tompsett > Materials Chemistry Group > Department of Chemistry > University of Bath > Bath BA2 7AY > U.K. > Tel: +44 7907 566351 (mobile) > Fax: +44 1225 386231 > http://people.bath.ac.uk/msi20/ > > >  > Get 100% visibility into Java/.NET code with AppDynamics Lite! > It's a free troubleshooting tool designed for production. > Get down to codelevel detail for bottlenecks, with <2% overhead. > Download for free and get started troubleshooting in minutes. > http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk > _______________________________________________ > Phonopyusers mailing list > Phonopyusers@... > https://lists.sourceforge.net/lists/listinfo/phonopyusers >  Prof. Aldo Humberto Romero CINVESTAVUnidad Queretaro Libramiento Norponiente 2000 CP 76230, Queretaro, QRO, Mexico tel: 442 211 9909 fax: 442 211 9938 email: aromero@... aldorome@... www: qro.cinvestav.mx/~aromero 
From: <dt331@ba...>  20130807 11:25:31

Dear All, I would like to know how the amplitude of atom movement is determined for a modulation. I am using the "MODULATION" keyword to generate modulated structures due to modes at qvectors (1/2,0,0) and (1/2,1/2,0). For example my input file for the first of these is: ATOM_NAME = Mg Sn O DIM = 2 2 2 MODULATION = 1/2 0 0 2 2 2, 1 20 I find that the dominant atom movement ( which is due to Sn atoms in my system ) is much larger for q=(1/2, 1/2, 0) compared to q=(1/2,0,0) when I use the same amplitude. Can someone explain to me why this occurs? Many thanks, David.  Dr. David A. Tompsett Materials Chemistry Group Department of Chemistry University of Bath Bath BA2 7AY U.K. Tel: +44 7907 566351 (mobile) Fax: +44 1225 386231 http://people.bath.ac.uk/msi20/ 