[Pw_forum] "Lanczos.x with k-point algorithm"

Iurii Timrov itimrov at sissa.it
Sun Jan 17 16:43:18 CET 2016


Dear Sajid Ali,

On 2016-01-17 00:40, Sajid Ali wrote:
> Dear,
>          Iurii Timrov
> I have tried using QE-5.5.0 for both SCF and TDDFPT

This version does not exist...

> But i recieived
> similar kind of error in lanczos.x output
> i.e. "mpirun noticed that process rank 1 with PID 1046 on node ermdc06
> exited on signal 11 (Segmentation fault)".
> Cant find a way out of it....?
> 

Do you perform a calculation for CH4 without any modifications? In your 
test do you use gamma_only case or general k points algorithm?

Do you run your calculation on a cluster using some script? How many 
cores did you use? How did you install QE? Could you ask your system 
administrator to help you? Try to run the code on a local workstation 
and see if it crashes.

Please provide more details (every step you do), because otherwise I 
cannot help you.

> SAJID ALI
> Cell#+61449764974
> Ph.D.  Scholar
> School of Mathematical and Physical Sciences
> University of Technology, Sydney
> Australia
> 
> 

Iurii Timrov
Postdoctoral Researcher
SISSA - International School for Advanced Studies
Condensed Matter Sector
Via Bonomea n. 265,
Trieste 34151, Italy


> ________________________________________
> From: Iurii Timrov <itimrov at sissa.it>
> Sent: 16 January 2016 14:01
> To: Sajid Ali
> Cc: PWSCF Forum
> Subject: Re: [Pw_forum] "Lanczos.x with k-point algorithm"
> 
> Dear Sajid Ali,
> 
> Why do you use now QE-5.2.1? Before you used QE-5.2.0. First of all, 
> you
> should stick to one version, otherwise it is not possible to solve the
> problem. I suggest to use the latest official release, i.e. QE-5.3.0.
> 
> On 2016-01-16 00:09, Sajid Ali wrote:
>> Dear,
>>       Iurii Timrov
>>      Thanks for your reply. I tried running default tests provided 
>> with
>> the
>> TDDFPT code by command 'make', but did not succeed. I was returned
>> with the error message
> 
> Try to run just one test:
> 
> cd TDDFPT/Examples/CH4
> make
> 
> Does this work for you?
> 
>>  "make[1]: Entering directory
>> `/panfs/panspermia/12108332/Programs/QE/espresso-5.2.1/TDDFPT/Examples/CH4'
>> Error reading bands from output file
> 
> Do this:
> 
> cd TDDFPT/Examples/CH4
> make clean
> make
> 
> Does it work now?
> 
>> Checking CH4.pw-out using CH4.pw-ref : make[1]: *** [small_test] Error
>> 5
>> make[1]: Leaving directory
>> `/panfs/panspermia/12108332/Programs/QE/espresso-5.2.1/TDDFPT/Examples/CH4'
>> make[1]: Entering directory
>> `/panfs/panspermia/12108332/Programs/QE/espresso-5.2.1/TDDFPT/Examples/SiH4'
>> mpirun -np 48   "../../../bin/turbo_lanczos.x" < SiH4.tddfpt-st-in >
>> SiH4.tddfpt-st-out
> 
> 48 cores are too many. Try to use just 1 or 2 cores for testing.
> 
>> /bin/sh: mpirun: command not found
>> make[1]: *** [SiH4.tddfpt-st-out] Error 127
>> make[1]: Leaving directory
>> `/panfs/panspermia/12108332/Programs/QE/espresso-5.2.1/TDDFPT/Examples/SiH4'
>> make[1]: Entering directory
>> `/panfs/panspermia/12108332/Programs/QE/espresso-5.2.1/TDDFPT/Examples/Benzene'
>> Checking Benzene.pw-out using Benzene.pw-ref :  [OK]
>> Checking Benzene.tddfpt-st-out using Benzene.tddfpt-st-ref : [NOT OK]
>> Different number of norms in files
>> make[1]: *** [small_test] Error 5
>> make[1]: Leaving directory
>> `/panfs/panspermia/12108332/Programs/QE/espresso-5.2.1/TDDFPT/Examples/Benzene'
>> make: *** [small_test] Error 2
>> -bash-4.1$ Different number of norms in files
>> -bash: Different: command not found
>> -bash-4.1$ make[1]: *** [small_test] Error 5
>> -bash: make[1]:: command not found"
>> 
>> As you suggested me to provide the input files for PWscf and TDDFPT
>> for my own calculations. so i am pasting them here respectively, I
>> will be thankful for any help you can provide to me as it is really
>> important for me....
>> For SCF  input file is
>> "&CONTROL
>> calculation = 'scf'
>> prefix='BN'
>> pseudo_dir = './'
>> outdir='./'
>> tstress=.true.
>> tprnfor = .true.
>> forc_conv_thr=1.0d-4
>> nstep=200
>> /
> 
> For 'scf' you don't need 'forc_conv_thr' and 'nstep'.
> 
>>     &SYSTEM
>> ibrav=0
>> celldm(1)=1.889726
>> nat=2
>> ntyp=2
>> ecutwfc=50.0
>> nosym = .true.,
>> nosym_evc= .true.,
>> /
> 
> Are you sure that you need to set 'nosym = .true.' and 'nosym_evc=
> .true.'? For HGH norm-conserving pseudopotentials the cutoff of 50 Ry 
> is
> probably too low (others probably can comment more on this issue). 
> Check
> the convergence wrt ecutwfc.
> 
>>     &ELECTRONS
>> mixing_beta=0.7
>> conv_thr =  1.0d-8
>> electron_maxstep=200
>> /
> 
> The default value for 'electron_maxstep' should be fine.
> 
>>     &IONS
>> trust_radius_max=0.2
>> /
>>     &CELL
>> cell_dynamics='bfgs'
>> /
> 
> The cards "ions" and "cell" are not needed for the SCF calculation.
> 
>> ATOMIC_SPECIES
>>    N  14.0067     N.pbe-hgh.UPF
>>    B  10.8110     B.pbe-hgh.UPF
>> 
>> CELL_PARAMETERS (alat=  1.88972600)
>>    2.509073557  -0.002024330   0.000000000
>>    1.252797201   2.173934109   0.000000000
>>   -0.000000000  -0.000000000  29.979498417
>> 
>> ATOMIC_POSITIONS (crystal)
>> B       -0.000266196  -0.000266358  -0.000000000
>> N        0.333599496   0.333599658   0.000000000
>> 
>> K_POINTS {automatic}
>> 4 4 1 0 0 0 "
>> For Lanczos the input file is
>> "&lr_input
>>    prefix = 'BN',
>>     outdir = './',
>>     restart=.false.,
>>     restart_step=500
>> /
>> &lr_control
>>   itermax = 500,
>>   ipol = 1,
>> /
>> [ &lr_post
>>  epsil = 0.0004d0
>>  ]
>> /"
>> 
> 
> The input for the turbo_lanczos.x code seems to be fine.
> 
>> SAJID ALI
>> Cell#+61449764974
>> Ph.D.  Scholar
>> School of Mathematical and Physical Sciences
>> University of Technology, Sydney
>> Australia
>> 
> 
> HTH
> 
> Iurii Timrov
> Postdoctoral Researcher
> SISSA - International School for Advanced Studies
> Condensed Matter Sector
> Via Bonomea n. 265,
> Trieste, Italy
> 
> 
>> 
>> ________________________________________
>> From: Iurii Timrov <itimrov at sissa.it>
>> Sent: 15 January 2016 17:10
>> To: PWSCF Forum
>> Cc: Sajid Ali
>> Subject: Re: [Pw_forum] "Lanczos.x with k-point algorithm"
>> 
>> Dear Sajid Ali,
>> 
>> I made a test using QE-5.3.0 (the latest release) and QE-5.2.0 (the 
>> one
>> which you use apparently) for the example TDDFPT/Examples/CH4 using 
>> the
>> general k-points implementation of TDDFPT. My calculations didn't 
>> crash
>> (neither in serial nor in parallel). Could you repeat these tests too?
>> 
>> In order to understand what is the problem with your calculations, you
>> should provide here the input data for PWscf and TDDFPT. Otherwise we
>> cannot help you. But before that try the default tests provided with
>> the
>> TDDFPT code, which I mentioned above.
>> 
>> Please note, even if you solve your problem, remember that the general
>> k-points implementation of TDDFPT was not benchmarked, and hence it 
>> can
>> give completely wrong results. This is why it is disabled in the code.
>> 
>> HTH
>> 
>> Iurii Timrov
>> 
>> 
>> On 2016-01-15 00:52, Sajid Ali wrote:
>>> Dear,
>>> 
>>>  All,
>>> 
>>> i am trying to do some TDDFPT calculations with K-point algorithm. I
>>> did comment the line ''IF ( .NOT. gamma_only ) && CALL errore('
>>> iosys', 'k-point algorithm is not tested yet',1)" in
>>> espresso/TDDFPT/src/lr_readin.f90 as was suggested by "Iurii Timrov".
>>> K point algorithm is implemented. But still i am returned the
>>> following error message in Lanczos.x output.[i am pasting the entire
>>> output file]..Can somebody help me out of this? It will be greatly
>>> appreciated.
>>> 
>>>  Program turboTDDFT v.5.2.0 starts on 15Jan2016 at 10:32:10
>>>  This program is part of the open-source Quantum ESPRESSO suite
>>>  for quantum simulation of materials; please cite
>>>  "P. Giannozzi et al., J. Phys.:Condens. Matter 21 395502 (2009);
>>>  URL http://www.quantum-espresso.org",
>>>  in publications or presentations arising from this work. More 
>>> details
>>> at
>>>  http://www.quantum-espresso.org/quote
>>> 
>>>  Parallel version (MPI), running on 24 processors
>>>  R & G space division: proc/nbgrp/npool/nimage = 24
>>>  Reading data from directory:
>>>  ./BN.save
>>>  Info: using nr1, nr2, nr3 values from input
>>>  Info: using nr1, nr2, nr3 values from input
>>>  IMPORTANT: XC functional enforced from input :
>>>  Exchange-correlation = SLA-PW-PBX-PBC ( 1 4 3 4 0 0)
>>>  Any further DFT definition will be discarded
>>>  Please, verify this is what you really want
>>>  Parallelization info
>>>  --------------------
>>>  sticks: dense smooth PW G-vecs: dense smooth PW
>>>  Min 13 13 3 2189 2189 339
>>>  Max 14 14 4 2219 2219 394
>>>  Sum 313 313 91 52939 52939 8919
>>>  Subspace diagonalization in iterative solution of the eigenvalue
>>> problem:
>>>  a serial algorithm will be used
>>>  Normal read
>>>  WARNING: Generalised k-points algorithm
>>>  LANCZOS LINEAR-RESPONSE SPECTRUM CALCULATION
>>>  Number of Lanczos iterations = 1500
>>>  Starting Lanczos loop 2
>>>  Lanczos iteration: 1 Pol:1
>>>  lr_apply_liouvillian: not applying interaction
>>>  alpha(00000001)= 0.000000
>>>  beta (00000001)= 2.374383
>>>  gamma(00000001)= 2.374383
>>>  z1= 1 0.000000000000000E+00 0.000000000000000E+00
>>>  Lanczos iteration: 2 Pol:1
>>> --------------------------------------------------------------------------
>>> 
>>> mpirun noticed that process rank 20 with PID 62359 on node c3node01
>>> exited on signal 11 (Segmentation fault).
>>> --------------------------------------------------------------------------
>>> 
>>> 
>>> SAJID ALI
>>> 
>>> Cell#+61449764974
>>> 
>>> Ph.D. Scholar
>>> 
>>> School of Mathematical and Physical Sciences
>>> 
>>> University of Technology, Sydney
>>> 
>>> Australia
>>> 
>>> 
>>> _______________________________________________
>>> Pw_forum mailing list
>>> Pw_forum at pwscf.org
>>> http://pwscf.org/mailman/listinfo/pw_forum
>> 
>> --
>> Dr. Iurii TIMROV
>> Postdoctoral Researcher
>> SISSA - International School for Advanced Studies
>> Condensed Matter Sector
>> Via Bonomea n. 265,
>> Trieste 34151, Italy




More information about the users mailing list