d3q-discuss Mailing List for d3q

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Dear Dr. Paulatto,

I am wondering if D3Q can work along with the 'twochem' setting of QE-7.3.1
If possible, that would enhance the potential of both these features.

Regards,
Prasad Matukumilli

Dear all,

Once again I am writing regarding the D3Q code. When I run the d3q.x executable to compute the number of triplets on which the calculation of mat3R will be performed, I get:

  Looking for irreducible triplets out of       16 possibilities:
           1:   (  0.0000  0.0000  0.0000 )   (  0.0000  0.0000  0.0000 )   (  0.0000  0.0000  0.0000 )   -->_Q1.0_0_0_Q2.0_0_0_Q3.0_0_0
           2:   (  0.0000  0.0000  0.0000 )   (  0.0000 -0.4999 -0.0075 )   ( -0.0000  0.4999  0.0075 )   -->_Q1.0_0_0_Q2.0_-1o2_0_Q3.0_1o2_0
           3:   (  0.0000  0.0000  0.0000 )   ( -0.4999  0.0000  0.0168 )   (  0.4999 -0.0000 -0.0168 )   -->_Q1.0_0_0_Q2.-1o2_0_0_Q3.1o2_0_0
           4:   (  0.0000  0.0000  0.0000 )   ( -0.4999 -0.4999  0.0093 )   (  0.4999  0.4999 -0.0093 )   -->_Q1.0_0_0_Q2.-1o2_-1o2_0_Q3.1o2_1o2_0
           5:   (  0.0000 -0.4999 -0.0075 )   ( -0.4999  0.0000  0.0168 )   (  0.4999  0.4999 -0.0093 )   -->_Q1.0_-1o2_0_Q2.-1o2_0_0_Q3.1o2_1o2_0
     IRREDUCIBLE NUMBER OF TRIPLETS TO COMPUTE:   5
     Setting up GGA 2nd derivative

     ===================================================
     =             Starting D3 calculation             =
     ===================================================
     =     triplet    1 (    1 of    5,   0% done)     =
     ===================================================
     =    q1 = (    0.000000   0.000000   0.000000 )   =
     =    q2 = (    0.000000   0.000000   0.000000 )   =
     =    q3 = (    0.000000   0.000000   0.000000 )   =
     ===================================================
.
.
.
.
.
For the second point, the code performs:

Computing P_c^q2 |d^q2 psi_k>
         pert=    1     time taken/total=    169.6/  12079.4 secs     av.it.  35.9
         pert=    2     time taken/total=    168.6/  12248.0 secs     av.it.  36.0
         pert=    3     time taken/total=    168.5/  12416.4 secs     av.it.  35.9
         pert=    4     time taken/total=    144.9/  12561.3 secs     av.it.  33.9
         pert=    5     time taken/total=    144.5/  12705.8 secs     av.it.  33.9
         pert=    6     time taken/total=    142.9/  12848.7 secs     av.it.  33.8
         pert=    7     time taken/total=    147.6/  12996.3 secs     av.it.  35.1
         pert=    8     time taken/total=    149.9/  13146.2 secs     av.it.  35.6
         pert=    9     time taken/total=    147.9/  13294.1 secs     av.it.  34.9
         pert=   10     time taken/total=    150.3/  13444.4 secs     av.it.  35.6
         pert=   11     time taken/total=    150.3/  13594.7 secs     av.it.  35.1
         pert=   12     time taken/total=    149.6/  13744.4 secs     av.it.  35.0
         pert=   13     time taken/total=    163.9/  13908.2 secs     av.it.  35.1
         pert=   14     time taken/total=    226.9/  14136.0 secs     av.it.  35.6
         pert=   15     time taken/total=    225.8/  14362.6 secs     av.it.  34.9
         pert=   16     time taken/total=    232.8/  14595.5 secs     av.it.  35.6
         pert=   17     time taken/total=    230.7/  14826.6 secs     av.it.  35.1
         pert=   18     time taken/total=    228.5/  15055.7 secs     av.it.  34.9

     Computing P_c^G |d^q2 psi_k-q2>
         pert=    1     time taken/total=    240.2/  15296.7 secs     av.it.  35.9
         pert=    2     time taken/total=    242.8/  15540.3 secs     av.it.  36.0

Then crashes:

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine davcio (5023):
     error writing file "/scratch/90022a/eamouj78/system//D3_Q1.0_0_0_Q2.0_-1o2_0_Q3.0_1o2_0/system.dq2pq3.8"
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...

This is most probably a memory issue. The grid I am using to do these calculations is the smallest for the case system I am using.

Is there a way to circumvent this problem?

Does the code print how much memory is needed? I could not find this information

Regards

Hello Elio,

I copy-pasted your input and it works for me.

The error must be elsewhere (are you sure you're not calli d3_r2q.x 
instead?)

kind regards

On 6/9/23 03:59, Elio Physics wrote:
> Dear  all,
>
> I am trying to test the d3q code by reproducing the results for 
> Silicon. Everything went fine except when I execute the d3_q2r.x flag 
> which applies to the 2^nd order dynamical matrices.  The dynamical 
> matrices for the 8 points in the q grid were produced as "matdyn0", 
> "matdyn1", "matdyn2",....,"matdyn8". My d3_q2r.x input is:
>
> &input
> fildyn  =  'matdyn'
> flfrc   =  'mat2R'
>  /
>
> However the code crashes with the error:
>
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      Error in routine q2r (1):
>      error reading input namelist
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
> I cannot figure out what is wrong with the input. I tried to rename 
> the files as "dyn0", "dyn1",.... and set fildyn='dyn', but I still get 
> the same error. Can anyone spot what the error could be?
>
> Thanks in advance..
>
> Elie Albert Moujaes
> Adjunct professor level 4
> Federal University of Rondonia
> Brazil
>
>
> _______________________________________________
> D3q-discuss mailing list
> D3q...@li...
> https://lists.sourceforge.net/lists/listinfo/d3q-discuss
-- 
Dr. Lorenzo Paulatto
IdR @ IMPMC - CNRS UMR 7590 & Sorbonne Université
phone: +33 (0)1 442 79822 / skype: paulatz
http://www.impmc.upmc.fr/~paulatto/ - https://anharmonic.github.io/
23-24/423 B115, 4 place Jussieu 75252 Paris CX 05
Dear  all,

I am trying to test the d3q code by reproducing the results for Silicon. Everything went fine except when I execute the d3_q2r.x flag which applies to the 2nd order dynamical matrices.  The dynamical matrices for the 8 points in the q grid were produced as "matdyn0", "matdyn1", "matdyn2",....,"matdyn8". My d3_q2r.x input is:

&input
fildyn  =  'matdyn'
flfrc   =  'mat2R'
 /

However the code crashes with the error:

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine q2r (1):
     error reading input namelist
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

I cannot figure out what is wrong with the input. I tried to rename the files as "dyn0", "dyn1",.... and set fildyn='dyn', but I still get the same error. Can anyone spot what the error could be?

Thanks in advance..

Elie Albert Moujaes
Adjunct professor level 4
Federal University of Rondonia
Brazil

Dear Dr. Paulatto,

I have been facing the below error everytime when I restart the run and
also noticed that the size of the current triplet in the TMP3 directory
crosses more than 1TB. Note that here is enough disk space available. Can
you please suggest how to continue d3q run without abrupt termination.

     =============== computing D3 start

     Symmetrizing and writing the tensor to disc

     CALLing d3matrix

       * symmetrizing with  2 sym.ops.

       point #  1  : star #  1  perm #  1 ( 1 2 3 )

      -->FILD3DYN/anh_Q1.0_0_-1o2_Q2.1o4_1o4_0_Q3.-1o4_-1o4_1o2

cannot open file FILD3DYN/anh_Q1.0_0_-1o2_Q2.1o4_1o4_0_Q3.-1o4_-1o4_1o2:
two xml files already opened

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     Error in routine write_d3dyn_xml (1):

     cannot open FILD3DYN/anh_Q1.0_0_-1o2_Q2.1o4_1o4_0_Q3.-1o4_-1o4_1o2

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...

**********************************
Prasad Matukumilli

Hello thank you for reporting!
The distinction of normal and umklapp contributions was quickly hacked in the code and is not implemented in the current version, but can be quickly hacked again.

The key point to keep in mind is that a process is umklapp if q1+q2+q3 =/= 0 after all three qX verctors have been wrapped inside the Brillouin zone. There are subroutines to do all this already in the code.
In linewidth.f90, subroutine linewidth_q at line 116 the contribution from q1, q2 and q3 is computed (they are stored in the array xq as xq(:,1), ...). You can add a condition like:
IF( ALL(ABS(refold_bz(xq(:,1),S%bg) &
+ refold_bz(xq(:,2),S%bg) &
+ refold_bz(xq(:,3), S%bg))<1.d-6) )THEN
print*, "This q1+q2+q3 combination is Normal"
--> skip this point if you only want Umklapp
ELSE
print*, "This q1+q2+q3 combination is Umklapp"
ENDIF

Then depending on your willingness to code, you can either add a new output variable and save both umklapp and normal, or just save one of the two, and run the code twice (higher CPU cost, lower human cost)
Please note that the graphene example is very under-converged. It takes a grid of 32x32x1 electronic k-points in pw.x to obtain a good phonon dispersion, ad ecutwfc of about 100Ry, a phonon grid of at least 8x8x1 q-points and a d3 double-grid of 4x4x1 q-points.
cheers

--
Lorenzo Paulatto - Paris
On Mar 16 2021, at 10:58 am, Prasad Matukumilli <pra...@gm...> wrote:
> Dear Dr. Paulatto,
> I need your help in decomposing broadening into normal and umklapp processes (or final state character) similar to Fig.4 (or Fig. 3) of PRB 87, 214303 (2013). I hope I can modify the code to extract such decomposed info.
>
> BTW, a small typo --'crytal' in Graphene example causes dispersion to look different.
> --
> **********************************
> Prasad Matukumilli
>
> _______________________________________________
> D3q-discuss mailing list
> D3q...@li...
> https://lists.sourceforge.net/lists/listinfo/d3q-discuss

Dear Dr. Paulatto,
I need your help in decomposing broadening into normal and umklapp
processes (or final state character) similar to  Fig.4 (or Fig. 3) of PRB
87, 214303 (2013). I hope I can modify the code to extract such decomposed
info.

BTW, a small typo --'crytal' in Graphene example causes dispersion to look
different.
-- 
**********************************
Prasad Matukumilli

> (i) I am able to reproduce all the features of figure 5, however the 
> range of y-axis is very different in the two plots. (see the attached 
> figure)

There is some normalization and unit of measure question about these 
plots, it may actually be that those in the paper are multiplied by 
Omega. If I remember correctly, they should be normalized to one when 
integrated from zero to infinity.

> 
> (ii) I am trying to plot the spectral function overlaid on phonon 
> dispersion - something similar to figure 4 of the adobe mentioned 
> article. the manual says that this figure can be produced by: (page 29 
> of thermal 2 manual)

Do you mean figure 3? It think it is just a typo in the manual: you can 
do a "plot 'spf.dat' with image" (i.e. not splot) if the lenght of the 
step along the path is always the same.
Or you can do
set view map
set pm3d
set log cb
splot 'spf.dat' u 1:2:(0):3 w pm3d

Which is much slower but gives much more flexibility. This is an actual 
example from Bi2Se3 with also the phonon dispersion superimposed:
splot 'cmpSPF-bz_T300_s2.out' u 1:2:(0):3 w pm3d, \
    for [i=6:20] '../LW_noso_noeps/cmpPath-bz_T80_s2.out' \
    u 2:i:(1) w l lc 'white' not

cheers

> 
> set log cb
> splot 'spf_full_T300_s0.3.out' u 1:2:3 w image
> 
> however looking at the output file (spf_full_T300_s0.3.out'). I think 
> this file does not have the necessary information to produce something 
> similar to figure 4.
> 
> Not sure where I am going wrong. Any pointers would be appreciated.
> 
> 
> Best
> Alex
> 
> 
> _______________________________________________
> D3q-discuss mailing list
> D3q...@li...
> https://lists.sourceforge.net/lists/listinfo/d3q-discuss
> 

-- 
Lorenzo Paulatto - Paris

Dear Dr. Paulatto,

I am trying to reproduce the spectral functional of  PdH ( as discussed in
the reference below).

https://arxiv.org/pdf/1411.5628.pdf

The IFcs for this computation were adopted from example files of thermal 2
( under the tutorial 2016 folder). I have the following two question:

(i) I am able to reproduce all the features of figure 5, however the range
of y-axis is very different in the two plots. (see the attached figure)

(ii) I am trying to plot the spectral function overlaid on phonon
dispersion - something similar to figure 4 of the adobe mentioned article.
the manual says that this figure can be produced by: (page 29 of thermal 2
manual)

set log cb
splot 'spf_full_T300_s0.3.out' u 1:2:3 w image

however looking at the output file (spf_full_T300_s0.3.out'). I think this
file does not have the necessary information to produce something similar
to figure 4.

Not sure where I am going wrong. Any pointers would be appreciated.

Best
Alex

Dear Dr. Paulatto,

I have one small confusion about lw.sma_full.*  files when we set the
store_lw=.true. during thermal conductivity sma calculations.

The manual says these are fwhm values (say lw1) .. so the phonon lifetimes
can be simply obtained as 1/lw1. Is my understanding correct?

Also i was wondering why the thermal2 code does not invoke crystal
symmerty  for reducing the number  of q-points.  Is there any particular
reason for that?

Last one. Is it required the integrate grid to be commensurate with lattice
vector? E.g if i have a a hexagonal cell, with a large c axis should i need
to scale my qpoint accordingly?

Thanks a lot for your time.

Best,
David

> 
> One unrelated query (just out of curiosity): I see that we can convert 
> IFCs from the Finite difference method (from ShengBTE) into thermal2.  
> Is there a  way to achieve opposite: Using D3Q IFCs into ShengBTe?

I don't know. I was in touch with a shengBTE developer many years ago 
and we exchanged informations to do this kind of conversion but I have 
not heard back. It should be trivial to do, the main difference being 
that ShengBTE uses periodic 3-body force constants, which can be 
obtained with from my software using the "-f 0" option of d3_qq2rr.x and 
d3_recenter.x. All the rest is just a matter of converting the units of 
measures (from bohr/a0^2 to eV/Å^3) and paying attention to the order of 
the indexes.

cheers

-- 
Lorenzo Paulatto - Paris

Thanks so much. This is very helpful

One unrelated query (just out of curiosity): I see that we can convert IFCs
from the Finite difference method (from ShengBTE) into thermal2.  Is there
a  way to achieve opposite: Using D3Q IFCs into ShengBTe?

Thanks

Regards
David

On Tue, Jun 23, 2020 at 2:30 PM Lorenzo Paulatto <pa...@gm...> wrote:

> Hello David, unfortunately the unit of velocity is a bit of a mess. If
> you are reading it from a file produced by d3_tk.x using the option
> store_lw=.true. then the unit is actually Ry. But if you get it from
> d3_r2q.x then there is a RY_TO_CMM1 factor in it, which mean that you
> have to divide by ( 13.6058 * 8065.5 ). I'm going to ix this sooner or
> later... If you want to get rid of this factor you can just remove it
> from line 773 of of PROGRAM_r2q.f90 and recompile
>
> regards
>
> On 6/23/20 11:34 AM, david yang wrote:
> > Dr. Paulatto,
> >
> > I  have some doubt about the unit of group velocity in thermal code. The
> > manal says they in Rydberg (Rydberg bohr ? ) unit with the conversion of
> > (1.09*10^6) m/sec
> >
> > If I use this My velocity comes out 407.581209999999998672*1.09*10^6,
> > which is too large to be correct.
> >
> > Simply by taking the slope of the dispersion I get about ~3400 m/sec
> > which is in the right ballpark.
> >
> > I would appreciate if you could point out my error here.
> >
> > Thank you
> > David
> >
> >
> > _______________________________________________
> > D3q-discuss mailing list
> > D3q...@li...
> > https://lists.sourceforge.net/lists/listinfo/d3q-discuss
> >
>
> --
> Lorenzo Paulatto - Paris
>

Hello David, unfortunately the unit of velocity is a bit of a mess. If 
you are reading it from a file produced by d3_tk.x using the option 
store_lw=.true. then the unit is actually Ry. But if you get it from 
d3_r2q.x then there is a RY_TO_CMM1 factor in it, which mean that you 
have to divide by ( 13.6058 * 8065.5 ). I'm going to ix this sooner or 
later... If you want to get rid of this factor you can just remove it 
from line 773 of of PROGRAM_r2q.f90 and recompile

regards

On 6/23/20 11:34 AM, david yang wrote:
> Dr. Paulatto,
> 
> I  have some doubt about the unit of group velocity in thermal code. The 
> manal says they in Rydberg (Rydberg bohr ? ) unit with the conversion of 
> (1.09*10^6) m/sec
> 
> If I use this My velocity comes out 407.581209999999998672*1.09*10^6, 
> which is too large to be correct.
> 
> Simply by taking the slope of the dispersion I get about ~3400 m/sec 
> which is in the right ballpark.
> 
> I would appreciate if you could point out my error here.
> 
> Thank you
> David
> 
> 
> _______________________________________________
> D3q-discuss mailing list
> D3q...@li...
> https://lists.sourceforge.net/lists/listinfo/d3q-discuss
> 

-- 
Lorenzo Paulatto - Paris

Dr. Paulatto,

I  have some doubt about the unit of group velocity in thermal code. The
manal says they in Rydberg (Rydberg bohr ? ) unit with the conversion of
(1.09*10^6) m/sec

If I use this My velocity comes out 407.581209999999998672*1.09*10^6, which
is too large to be correct.

Simply by taking the slope of the dispersion I get about ~3400 m/sec which
is in the right ballpark.

I would appreciate if you could point out my error here.

Thank you
David

p.s. I guess you're using an awful lot of k-points, maybe you can do the 
d3 calculations with less k-points and a larger smearing to save some 
input/output and disk space

cheers

On 6/15/20 3:08 PM, david yang wrote:
> Dear D3Q, and thermal2 developers.
> 
> I am using D3Q, in combination with thermal2 to calculate thermal 
> conductivity of  bulk TMDs..
> 
> I am facing a slight problem while doing the D3Q step. While performing 
> the calculations the D3Q code temporarily stores huge data under 
> D3_Q1.0_0_0_Q2* directory.. Often exceeding 1TB..( e.g the directory 
> size below
> 
> 20G D3_Q1.0_1o4_0_Q2.1o4_0_-1o2_Q3.-1o4_-1o4_1o2
> 20G D3_Q1.0_1o4_0_Q2.1o4_1o4_0_Q3.-1o4_-1o2_0
> *405G D3_Q1.0_1o4_0_Q2.1o4_1o4_-1o2_Q3.-1o4_-1o2_1o2
> *
> *
> *
> *Since I have a limited scratch space it often kills my job (:*
> 
> I was wondering if storing this data is necessary for using D3Q with 
> thermal2. If not, how can we suppress the writing of this data.
> 
> Below i append my d3q.in <http://d3q.in> file:
> 
> D3Q of TiS2
>   &inputd3q
>     mode='full'
>     prefix='TiS2'
>     outdir='./'
>     fild3dyn='TiS2.dyn',
>     fildrho = 'auto:drho',
>     fildrho_dir = 'FILDRHO'
>     ethr_ph = 1.d-10
>     restart=.false.
> /
> 4 4 2
> 
> 
> 
> Thanks
> David
> 
> 
> _______________________________________________
> D3q-discuss mailing list
> D3q...@li...
> https://lists.sourceforge.net/lists/listinfo/d3q-discuss
> 

-- 
Lorenzo Paulatto - Paris

Hello David, all that data is actually necessary. It may be possible to use
less disk, but it woul required a complete rewrite of the core. You can use
a scratch space which is local to the computing node (e.g. /tmp) if it is
better, although modernization computing clusters tend to have very little
local disk.

Regards

-- 
Lorenzo Paulatto

On Mon, 15 Jun 2020, 15:13 david yang, <dya...@gm...> wrote:

> Dear D3Q, and thermal2 developers.
>
> I am using D3Q, in combination with thermal2 to calculate thermal
> conductivity of  bulk TMDs..
>
> I am facing a slight problem while doing the D3Q step. While performing
> the calculations the D3Q code temporarily stores huge data under
> D3_Q1.0_0_0_Q2* directory.. Often exceeding 1TB..( e.g the directory size
> below
>
> 20G D3_Q1.0_1o4_0_Q2.1o4_0_-1o2_Q3.-1o4_-1o4_1o2
> 20G D3_Q1.0_1o4_0_Q2.1o4_1o4_0_Q3.-1o4_-1o2_0
>
> *405G D3_Q1.0_1o4_0_Q2.1o4_1o4_-1o2_Q3.-1o4_-1o2_1o2*
>
> * Since I have a limited scratch space it often kills my job (:*
>
> I was wondering if storing this data is necessary for using D3Q with
> thermal2. If not, how can we suppress the writing of this data.
>
> Below i append my d3q.in file:
>
> D3Q of TiS2
>  &inputd3q
>    mode='full'
>    prefix='TiS2'
>    outdir='./'
>    fild3dyn='TiS2.dyn',
>    fildrho = 'auto:drho',
>    fildrho_dir = 'FILDRHO'
>    ethr_ph = 1.d-10
>    restart=.false.
> /
> 4 4 2
>
>
>
> Thanks
> David
> _______________________________________________
> D3q-discuss mailing list
> D3q...@li...
> https://lists.sourceforge.net/lists/listinfo/d3q-discuss
>

Dear D3Q, and thermal2 developers.

I am using D3Q, in combination with thermal2 to calculate thermal
conductivity of  bulk TMDs..

I am facing a slight problem while doing the D3Q step. While performing the
calculations the D3Q code temporarily stores huge data under
D3_Q1.0_0_0_Q2* directory.. Often exceeding 1TB..( e.g the directory size
below

20G D3_Q1.0_1o4_0_Q2.1o4_0_-1o2_Q3.-1o4_-1o4_1o2
20G D3_Q1.0_1o4_0_Q2.1o4_1o4_0_Q3.-1o4_-1o2_0

*405G D3_Q1.0_1o4_0_Q2.1o4_1o4_-1o2_Q3.-1o4_-1o2_1o2*

* Since I have a limited scratch space it often kills my job (:*

I was wondering if storing this data is necessary for using D3Q with
thermal2. If not, how can we suppress the writing of this data.

Below i append my d3q.in file:

D3Q of TiS2
 &inputd3q
   mode='full'
   prefix='TiS2'
   outdir='./'
   fild3dyn='TiS2.dyn',
   fildrho = 'auto:drho',
   fildrho_dir = 'FILDRHO'
   ethr_ph = 1.d-10
   restart=.false.
/
4 4 2

Thanks
David

Dear D3Q, and thermal2 developers.

I am using D3Q, in combination with thermal2 to calculate thermal
conductivity of  bulk TMDs..

I am facing a slight problem while doing the D3Q step. While performing the
calculations the D3Q code temporarily stores huge data under
D3_Q1.0_0_0_Q2* directory.. Often exceeding 1TB..( e.g the directory size
below

20G D3_Q1.0_1o4_0_Q2.1o4_0_-1o2_Q3.-1o4_-1o4_1o2
20G D3_Q1.0_1o4_0_Q2.1o4_1o4_0_Q3.-1o4_-1o2_0

*405G D3_Q1.0_1o4_0_Q2.1o4_1o4_-1o2_Q3.-1o4_-1o2_1o2*

* Since I have a limited scratch space it often kills my job (:*

I was wondering if storing this data is necessary for using D3Q with
thermal2. If not, how can we suppress the writing of this data.

Below i append my d3q.in file:

D3Q of TiS2
 &inputd3q
   mode='full'
   prefix='TiS2'
   outdir='./'
   fild3dyn='TiS2.dyn',
   fildrho = 'auto:drho',
   fildrho_dir = 'FILDRHO'
   ethr_ph = 1.d-10
   restart=.false.
/
4 4 2

Thanks
David

Hello,
if you include vdW with the semi-empiric ion-ion correction, it would 
actually be trivial to include it in D3, it is probably a few day of 
programming and testing. We can discuss this matter more in detail if 
you are interested.

You can include it or not in pw.x, it would make no difference as it has 
no *direct* effect on the electronic structure.

On the other hand, if you want to use a vdW-enabled functional, that is 
probably not going to work if you use it in the pw calculation

cheers

On 6/11/20 10:26 AM, Alex Kent wrote:
> Dear Dr. Paulatto,
> 
> I have one quick query about D3Q code. As per manual I am aware that D3Q 
> cannot handle vdw as of now.  My question is: while setting up a D3Q 
> calculation should we not include any vdw tags in the pw.x step.. Or if 
> included will those tags be simply ignored while doing d3q step.
> 
> I ask because I am interested in calculating phonons with vdw and 
> finally third order terms without vdw.
> 
> 
> Thanks !!
> 
> 
> Best.
> Alex

-- 
Lorenzo Paulatto - Paris

Dear Dr. Paulatto,

I have one quick query about D3Q code. As per manual I am aware that D3Q
cannot handle vdw as of now.  My question is: while setting up a D3Q
calculation should we not include any vdw tags in the pw.x step.. Or if
included will those tags be simply ignored while doing d3q step.

I ask because I am interested in calculating phonons with vdw and finally
third order terms without vdw.

Thanks !!

Best.
Alex

What we call "linewidth" or "gamma" in the code is the imaginary part of 
the self energy, or the half-width half-maximum of the phonon lines.

To answer you question, it is neither, tau is the inverse of the FULL 
width half maximum, hence:
tau = 1/(2*gamma)

hope this helps

regards
Lorenzo

On 5/26/20 6:03 PM, Alex Kent wrote:
> Dear Dr. Paulatto,
> 
> I have one quick query about thermal2 code. I am interested in 
> calculating phonon-life times along high symmetry path in the Brillouin 
> zone. I can calculated the line width along a high symmetry path using 
> thermal 2.
> 
> 
> Within the thermal2 implementation. what will be correct way to obtain 
> phonon lifetime (\tau_ph=1/linewidth, or \tau_ph=1/(2*pi*linewidth)  )
> 
> Thanks a lot for your help.
> 
> Best.
> Alex
> 

-- 
Lorenzo Paulatto - Paris

Dear Dr. Paulatto,

I have one quick query about thermal2 code. I am interested in calculating
phonon-life times along high symmetry path in the Brillouin zone. I can
calculated the line width along a high symmetry path using thermal 2.

Within the thermal2 implementation. what will be correct way to obtain
phonon lifetime (\tau_ph=1/linewidth, or \tau_ph=1/(2*pi*linewidth)  )

Thanks a lot for your help.

Best.
Alex

> Thank you for this wonderful code. Does the recent update (September 
> 2019) include the method described in
> 
> https://www.nature.com/articles/s41567-019-0520-x
> 

Dear Alex, it isn't at the moment. You can contact prof Nicola Marzari 
for more information.

kind regards

> Thank you for your time.
> 
> Best Regards
> 
> Alex

-- 
Lorenzo Paulatto - Paris

Dear D3Q Developers,

Thank you for this wonderful code. Does the recent update (September 2019)
include the method described in

https://www.nature.com/articles/s41567-019-0520-x

Thank you for your time.

Best Regards

Alex

Dear Sir,

I am trying to install d3q code (version:d3q-1.1.5-qe6.4) with qe6.4. I
could successfully install it on my personal desktop with gfortran
compiler, but when I am trying to install it on a cluster with ifort
compiler, I am getting the below errors. Could you please help me out to
rectify this?

f3_bwfft.f90(225): error #6463: This is not a derived type name.   [GRID]
    TYPE(grid),INTENT(inout) :: fc3
---------^
f3_bwfft.f90(547): error #6463: This is not a derived type name.   [GRID]
    TYPE(grid),INTENT(in)    :: fc3
---------^
f3_bwfft.f90(615): error #6463: This is not a derived type name.   [GRID]
    TYPE(grid),INTENT(in)    :: fc3
---------^
f3_bwfft.f90(420): error #6404: This name does not have a type, and must
have an explicit type.   [FC3]
    fc3%n_R = nxr_list
----^
f3_bwfft.f90(420): error #6460: This is not a field name that is defined in
the encompassing structure.   [N_R]
    fc3%n_R = nxr_list
--------^
f3_bwfft.f90(422): error #6460: This is not a field name that is defined in
the encompassing structure.   [FC]
    ALLOCATE(fc3%fc(3*nat,3*nat,3*nat, nxr_list))
-----------------^
f3_bwfft.f90(423): error #6460: This is not a field name that is defined in
the encompassing structure.   [IFC]
    ALLOCATE(fc3%ifc(3*nat,3*nat,3*nat, nxr_list))
-----------------^
f3_bwfft.f90(424): error #6460: This is not a field name that is defined in
the encompassing structure.   [XR2]
    ALLOCATE(fc3%xR2(3,nxr_list), fc3%xR3(3,nxr_list))
-----------------^
f3_bwfft.f90(424): error #6460: This is not a field name that is defined in
the encompassing structure.   [XR3]
    ALLOCATE(fc3%xR2(3,nxr_list), fc3%xR3(3,nxr_list))
--------------------------------------^
f3_bwfft.f90(429): error #6911: The syntax of this substring is invalid.
[FC]
      fc3%fc(:,:,:,irx)  = DBLE(fcx)
----------^
compilation aborted for f3_bwfft.f90 (code 1)
make[1]: *** [f3_bwfft.o] Error 1
make[1]: Leaving directory
`/home/pramod/codes/q-e-qe-6.4/d3q-1.1.5-qe6.4/thermal2'
make: *** [thermal2] Error 2

With Best Regards,

Pramod Kumar Verma
Research Scholar
CCMT Group,
Department of Physics
IISc, Bangalore-560012

2017	Jan	Feb	Mar	Apr (3)	May	Jun	Jul	Aug	Sep	Oct	Nov (2)	Dec (1)
2018	Jan (2)	Feb	Mar (2)	Apr (8)	May (4)	Jun (8)	Jul (6)	Aug	Sep	Oct (1)	Nov	Dec (1)
2019	Jan (1)	Feb	Mar (1)	Apr (2)	May	Jun (2)	Jul	Aug (1)	Sep	Oct (2)	Nov	Dec
2020	Jan	Feb	Mar	Apr	May (2)	Jun (10)	Jul (1)	Aug (2)	Sep	Oct	Nov	Dec
2021	Jan	Feb	Mar (3)	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2023	Jan	Feb	Mar	Apr	May	Jun (3)	Jul	Aug	Sep	Oct	Nov	Dec
2024	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct (1)	Nov	Dec