[QE-users] mechanical stability under pressure from phonons
Thank you very much for your answer Prof. Marzari However, I still do not understand something. Elastic constants are equivalent to phonons in the long-wavelength limit. Thus, if the stability condition using elastic constants have explicitly the pressure, also the stability condition using phonons should have explicitly the pressure. Is the pressure cancelled out in some way? I do not understand it from Grimvall and I do not find any citation on this matter. Uri Argaman Ben-Gurion University ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] mechanical stability under pressure from phonons
Hi all In the literature the mechanical stability condition using elastic constants includes explicitly the pressure: In cubic symmetry, for example: C11 + 2*C12 +P>0; C44 +P>0; C11 + C12 + 2*P>0; from: Grimvall et al. REVIEWS OF MODERN PHYSICS, VOLUME 84, 2012 In this paper, they write: " There seems to be no rigorous discussion of this point in papers that have presented ab initio calculations of phonon dispersion curves under pressure,..." The stability condition from phonons: omega>0 for all modes. It seems that it is not equivalent, because the phonon condition is at constant volume. Is it known what is the condition at constant pressure? Than you very much Uri Argaman Ben-Gurion University of the Negev Israel ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] Quantum Espresso on Amazon Web Services
Dear QE users and devalopers I'm running QE on Amazon Web Services (AWS) and I see that the run times are very bad compared to other systems that I use and I also encountered crashes which I cannot understand the reason for them. Does anyone work on AWS and know something about the performance and special compilation issues? Thank you Uri Argaman Ben-Gurion University Israel ___ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[Pw_forum] Highest occupied level is higher than the lowest unoccupied level and the band-gap problem
Dear QE users and developers, I calculate the energy using pw.x of GeTe in non-equilibrium positions and unit-cell. I get the following results: "highest occupied, lowest unoccupied level (ev): 8.72567.626." I know that the material is a metal or semi-metal. The same thing happens with bismuth (in A7 structure) which is a semi-metal. I understand that this kind of calculations should be done with smearing and I read in previous messages by Paolo that it is known to happen in semi-metals and relates to the band-gap problem. I know the band-gap problem in DFT but I do not understand why a level which is lower in energy than the highest occupied level will not be occupied, and instead a higher level is occupied. I would expect that the calculation will not converge but I do not understand why it converges to the wrong state. Is it similar to the Janak theorem and fractional occupation in the Fe atom? The relevant levels for GeTe are: k = 0. 0.-0.8124 ( 1520 PWs) bands (ev): -17.0056 -17.0056 -16.9992 -16.9979 -16.9979 -3.7630 -1.5570 7.1574 7.1574 8.7256 8.7360 8.7360 9.2246 16.8372 16.8922 17.9505 17.9505 18.5818 18.5818 23.8031 occupation numbers 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. k = 0.1861-0.8594-0.2708 ( 1512 PWs) bands (ev): -17.0067 -17.0055 -16.9990 -16.9971 -16.9963 -3.3095 -1.0166 5.5826 5.8078 6.6930 7.6261 9.8147 9.9726 16.4111 16.5968 17.8171 17.8592 21.5515 21.7205 23.8798 occupation numbers 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] symmetry opertions
Dear QE users and developers What is the meaning of the f vector in the symmetry operations? For example (from the PW output): isym = 6 120 deg rotation - cart. axis [-1,1,1] cryst. s( 6) = ( 0 0 -1 )f =( 0.000 ) (-1 0 0 ) ( 0.500 ) ( 0 1 0 ) ( 0.500 ) cart.s( 6) = ( 0.000 -1.000 0.000 )f =( 0.000 ) ( 0.000 0.000 1.000 ) ( 0.500 ) ( -1.000 0.000 0.000 ) ( 0.500 ) Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] LO-TO splitting in matdyn and dynmat
Dear QE users and developers I a little bit confused about the LO-TO splitting implementation. As long as I understand, the LO-TO splitting in included in phonon calculation on a grid using matdyn as in the GRID_example in the phonon code (espresso version 5.4.0). But I see that a q direction also can be specified using dynmat but I do not found an example and when I try to use this option I get a very strange output that seems like a bug, shown below. Thank you very much Uri Argaman Ben-Gurion University Israel Program DYNMAT v.5.4.0 starts on 29Sep2017 at 11:25: 9 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 1 processors Reading Dynamical Matrix from file SnS.dyn ...Force constants read ...epsilon and Z* read ...Raman cross sections read Acoustic Sum Rule: || Z*(ASR) - Z*(orig)|| =8.750696E-03 Acoustic Sum Rule: ||dyn(ASR) - dyn(orig)||=1.408996E-04 Polarizability (A^3 units) multiply by 0.177561 for Clausius-Mossotti correction 208.514499 -0.0014310.000687 -0.001431 250.432087 -0.000245 0.000687 -0.000245 222.813312 IR activities are in (D/A)^2/amu units Raman activities are in A^4/amu units multiply Raman by 0.031528 for Clausius-Mossotti correction # mode [cm-1][THz] IR Raman depol.fact 1 -0.00 -0.0. 0.0.6282 2 -0.00 -0.0. 0.0.6732 3 0.000.0. 0.0.6056 4 30.900.92640. 166.30860.5986 5 33.681.00960. 180.04200.7500 6 37.241.11630.67.26290.7500 7 55.371.65990. 0.0.7066 8 59.541.78500. 391.98870.7500 9 67.972.03770.1981 0.0.6952 10 78.842.36360.16.02300.7495 11 80.232.40520. 2364.69610.6050 12 91.512.74340.7427 0.0.4744 13130.393.9089 54.0059 0.0.3529 14142.084.25950. 6899.51330.7500 15174.855.24190. 27086.71900.0153 16174.955.24490.0001 0.0.0155 17183.415.49830. 0.09050.6085 18187.945.6343 33.8187 0.0.0112 19195.105.84892.5529 0.0.7332 20198.895.96277.5259 0.0.0694 21200.095.99860. 7744.85590.0453 22204.426.12850. 466.84060.7492 23242.247.2621 36.7227 0.0.1018 24265.707.96550. 161.91040.7500 DYNMAT : 0.02s CPU 0.04s WALL This run was terminated on: 11:25: 9 29Sep2017 =--= JOB DONE. =--= 2503 0.00 0.000279 0.00 ) ( 0.000762 0.00-0.16 0.00 0.000363 0.00 ) ( 0.000352 0.00-0.372372 0.00 0.000355 0.00 ) ( -0.000591 0.00 0.372503 0.00-0.000279 0.00 ) ( -0.000352 0.00 0.372372 0.00-0.000355 0.00 ) ( 0.000161 0.00-0.333886 0.00 0.000141 0.00 ) ( -0.000762 0.00 0.16 0.00-0.000363 0.00 ) ( -0.000161 0.00 0.333886 0.00-0.000141 0.00 ) freq (6) = 1.116310 [THz] = 37.236100 [cm-1] ( 0.363523 0.00 0.000296 0.00 0.031110 0.00 ) ( 0.342584 0.00 0.000689 0.00 0.001621 0.00 ) ( 0.362422 0.00 0.000659 0.00-0.026327 0.00 ) ( -0.363523 0.00-0.000296 0.00-0.031110 0.00 ) ( -0.362422 0.00-0.000659 0.00 0.026327 0.00 ) ( 0.342738 0.00 0.000108 0.00 0.002015 0.00 ) ( -0.342584 0.00-0.000689 0.00-0.001621 0.00 ) ( -0.342738 0.00-0.000108 0.00-0.002015 0.00 ) freq (7) = 1.659934 [THz] = 55.369437 [cm-1] ( -0.000121 0.00 0.467450 0.00-0.62 0.00 ) ( -0.000248 0.00-0.177691 0.00 0.000166 0.00 ) ( 0.000107 0.00-0.467376 0.00 0.37
[Pw_forum] Problem with BURAI1.3 on windows
Hello I download BURAI1.3 for windows from: http://nisihara.wixsite.com/burai/resources and instal Java: Version 8 from: https://www.java.com/en/download/windows-64bit.jsp The graphics do not work. I also want to ask if there is a version for Linux? Thank you Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] symmetry notation in phonon calculations
Dear QE users and developers When I do a phonon calculations I see symmetry notations like this: Representation 1 1 modes -A_1 D_1 S_1 To be done What is the meaning of: A_1,D_1 and S_1 ? Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Nose-Hoover thermostat in MD in pw.x
Dear QE users and developers I run MD simulations using pw.x and the Andersen thermostat. I want to use the Nose-Hoover thermostat but noticed it only exists in cp.w. Is there a way to incorporate a Nose-Hoover thermostat in pw.x? Thank you very much Uri Argman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Raman intensity and problem in example 15 in the phonon examples
Dear QE users and developers I try to obtain Raman intensity using ph.x and dynmat.x. I run example 15 in the phonon examples (espresso 5.4.0) and I get only IR intensity. The plot of gnuplot give the Raman intensity but it is just the frequency in Thz and not really the intensity: plot 'plotdata_zno.dat' u (\$2):(\$3) title ' ZnO-RAMAN' w i lw 2 the file 'plotdata_zno.dat' contain: 1 -0.00 -0.0. 2 -0.00 -0.0. 3 -0.00 -0.0. 4 0.000.0. 5 0.000.0. 6 0.000.0. 7 50.571.51610. 8107.073.21000. 9158.384.74820. 10215.376.45650. 11276.678.29430. 12337.61 10.12120. (\$2):(\$3) This is the frequency in cm^-1 vs the frequency in THz. I do not understand where is the real intensity. After that I try to calculate the intensity with the following input: For ph.x: phonon for NaCl &inputph tr2_ph=1.0d-15, prefix='NaCl', fildvscf='aldv', amass(1)=22.9898, amass(2)=35.453, outdir='/home/argaman/tmp/tmp17116/', fildyn='NaCl.dyn', trans=.true., lraman=.true., epsil=.true., / 0 0 0 For dynmat.x: &input fildyn='NaCl.dyn', asr='simple'/ With norm-conserving PP and LDA. I still do not obtain the intensity for Raman in the file NaCl.dm.out, only IR: Reading Dynamical Matrix from file NaCl.dyn ...Force constants read ...epsilon and Z* read ...Raman cross sections read A direction for q was not specified:TO-LO splitting will be absent Polarizability (A^3 units) multiply by 0.655304 for Clausius-Mossotti correction 5.1472710.000.00 0.005.1472710.00 0.000.005.147271 IR activities are in (D/A)^2/amu units # mode [cm-1][THz] IR 1 0.000.0. 2 0.000.0. 3 0.000.0. 4183.555.50271.8492 5183.555.50271.8492 6183.555.50271.8492 Electronic dielectric permittivity tensor (F/m units) 2.5780300.000.00 0.002.5780300.00 0.000.002.578030 ... with zone-center polar mode contributions 5.4339260.000.00 0.00 5.4339260.00 0.000.005.433926 What did I do wrong? Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] compilation problem with intel parallel studio 2017
I compile QE with: ./configure compiler=intel MPIF90=mpiifort FC=ifort F77=ifort F90=ifort CXX=icpc CC=icc MKLROOT=/opt/intel/mkl LIBDIR=/opt/intel/mkl/ and it still do not find MKL libraries: checking for dummy main to link with Fortran 77 libraries... none checking for Fortran 77 name-mangling scheme... lower case, underscore, no extra underscore checking for library containing dgemm... no MKL not found in /opt/intel/composer*/mkl/lib/intel64: checking for library containing dgemm... no MKL not found in /opt/intel/Compiler/*/*/mkl/lib/em64t: checking for library containing dgemm... no MKL not found in /opt/intel/mkl/*/lib/em64t: checking for library containing dgemm... no MKL not found in /opt/intel/mkl*/lib/em64t: checking for library containing dgemm... no MKL not found in /opt/intel/mkl/lib: checking for library containing dgemm... no MKL not found in /opt/intel/impi/2017.1.132/intel64/lib: checking for library containing dgemm... no MKL not found in /opt/intel/mkl/lib/intel64: checking for library containing dgemm... -lmkl_intel_lp64 setting BLAS_LIBS... -L/opt/intel/mkl/lib/intel64 -lmkl_intel_lp64 -lmkl_sequential -lmkl_core checking for library containing dspev... none required checking FFT... in /opt/intel/mkl/lib/intel64: checking for DftiComputeForward in -lmkl_intel_lp64... yes checking /opt/intel/mkl/include/mkl_dfti.f90 usability... no checking /opt/intel/mkl/include/mkl_dfti.f90 presence... yes configure: WARNING: /opt/intel/mkl/include/mkl_dfti.f90: present but cannot be compiled configure: WARNING: /opt/intel/mkl/include/mkl_dfti.f90: check for missing prerequisite headers? configure: WARNING: /opt/intel/mkl/include/mkl_dfti.f90: see the Autoconf documentation configure: WARNING: /opt/intel/mkl/include/mkl_dfti.f90: section "Present But Cannot Be Compiled" configure: WARNING: /opt/intel/mkl/include/mkl_dfti.f90: proceeding with the preprocessor's result configure: WARNING: /opt/intel/mkl/include/mkl_dfti.f90: in the future, the compiler will take precedence checking for /opt/intel/mkl/include/mkl_dfti.f90... yes checking MASS... checking for library containing mpi_init... none required checking for library containing pdgemr2d... no checking for library containing pdgemr2d... no checking for library containing pdgemr2d... no checking for library containing pdgemr2d... no checking for library containing pdgemr2d... no checking for library containing pdgemr2d... no How to solve it? Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] ASR in an unstable structure
Dear QE users and developers I run a phonon calculation in SnSe (Roch-salt structure) and I get the frequencies in the Gamma point (dyn1 file): q = (0.0 0.0 0.0 ) ** freq (1) = -0.947936 [THz] = -31.619734 [cm-1] ( 0.463998 0.00 -0.152295 0.00 -0.248340 0.00 ) ( -0.708498 0.00 0.232546 0.00 0.379201 0.00 ) freq (2) = -0.947936 [THz] = -31.619734 [cm-1] ( -0.267553 0.00 -0.407542 0.00 -0.249970 0.00 ) ( 0.408539 0.00 0.622293 0.00 0.381690 0.00 ) freq (3) = -0.947936 [THz] = -31.619734 [cm-1] ( 0.115246 0.00 -0.332981 0.00 0.419526 0.00 ) ( -0.175974 0.00 0.508442 0.00 -0.640593 0.00 ) freq (4) = -0.098311 [THz] = -3.279313 [cm-1] ( -0.486385 0.00 0.205881 0.00 0.478423 0.00 ) ( -0.478812 0.00 0.202675 0.00 0.470973 0.00 ) freq (5) = -0.098311 [THz] = -3.279313 [cm-1] ( 0.492088 0.00 0.396133 0.00 0.329809 0.00 ) ( 0.484426 0.00 0.389965 0.00 0.324673 0.00 ) freq (6) = -0.098311 [THz] = -3.279313 [cm-1] ( 0.170660 0.00 -0.555462 0.00 0.412533 0.00 ) ( 0.168003 0.00 -0.546813 0.00 0.406109 0.00 ) ** After I use q2r.x and plotphonon to plot the spectrum (using matdyn.x with asr='crystal') I get the frequencies in the Gamma point: 0,0,0 (for acoustic, which is good) and ~ -1 -1 3 THz for the optics ,which is strange. From where the 3 THz frequency come from? Which frequencies are more reliable, the original ones or those of matdyn? I note that it is probably unstable structure but I still want to now what are the unstable modes. Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
Re: [Pw_forum] transverse acoustic, longitudinal acoustic, and out-of-plane transverse acoustic
> 1- Does the dimension of Supercell for phonon calculation determine > by nq1, nq2,and nq3 in the phonon input file? The parameters nq1, nq2,and nq3 is determine the Monkhorst-Pack grid for the DFPT calculation (phonon calculation) that done in q-space (you calculate the dynamical matrices explicitly and not the force constants). In DFPT there is no supercell but the convergence with respect to q-points is similar to the convergence with respect to the size of the supercell in supercell approaches. > 2- we know that the lowest three dispersion in the phonon spectra > show the three acoustic branches, but how can I find which one is > for TA, LA, and ZA? Phonons do not have just transverse or longitudinal polarization. There are phonons with polarization that is neither transverse nor longitudinal (although transverse and longitudinal polarizations are off-course possible). The most general polarization is elliptical. In general, the polarization vector is complex and you can know it from the calculation (after diagonalization of the dynamical matrices). You have this information in the dynamical matrices files which is output of the phonon calculation and also you have it in the output of matdyn.x in the file matdyn.modes. We recently publish a work in which we analyze the polarization of a complex acoustic phonon in the corner of a hexagonal Brillouin zone: http://journals.aps.org/prb/abstract/10.1103/PhysRevB.94.174305 You can maybe find it interesting. Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
Re: [Pw_forum] charge density
Dear Alexandra Davila I also publish a message about this subject a few days ago. I also think it something with the pseudopotential but I check it also for norm-conserving pseudopotential and I have the same problem. In your message you write: "From each file I obtain different values 6.79 and 7.42 respectively." What are exactly these values? These values are the same of all the values of the charge differences which should be zero? Do you use plot_num=9? I note that I am not a developer but I am using QE for a long time. Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] problem with pp.x in the calculation of delta rho (plot_num=9)
Dear QE experts As I already mentioned in previous messages, I notice to a problem in the calculation of delta rho. I convinced that it is not an issue of numerical accuracy. When I perform a calculation of delta rho with norm-conserving pseudopotential in NaCl I get that the sum of delta rho in the xsf file is: 202.356763363 while the sum of the absolute value is: 786.819282749. It is not seems like a numerical noise. I also check that it is in the order of 10% from the density. When I perform the same calculation with ultrasoft pseudopotential I get that the sum of delta rho in the xsf file is: -4.988217364 while the sum of the absolute value is: 102.279406346. I also notice that the number of points in the grid is different with the different pseudopotential even though the parameters: e1(1) =1.0, e1(2)=0.0, e1(3) = 0.0, e2(1) =0.0, e2(2)=1.0, e2(3) = 0.0, e3(1) =0.0, e3(2)=0.0, e3(3) = 1.0, nx=50, ny=50, nz=50 are the same. I try to run this code with zero density to see what is the atomic densities but I do not succeed to understand how to this. I also have this problem in other materials with other pseudopotential. It is possible that some pseudopotential do not fit to the calculation of delta rho (plot_num=9)? The input files are (for norm converving): pw input: &control calculation='scf', restart_mode='from_scratch', pseudo_dir = '/home/uriargaman/espresso-5.4.0/pseudo', outdir = '/home/uriargaman/tmp15711/', prefix='NaCl', tstress = .true., tprnfor = .true., / &system ibrav =2, celldm(1) =10.7789934208, nat= 2, ntyp= 2, ecutwfc = 80 / &electrons diagonalization='david' conv_thr = 1.0e-8 mixing_beta = 0.7 / &ions ion_dynamics= 'bfgs' / &cell cell_dynamics= 'bfgs' / ATOMIC_SPECIES Na 22.9898 Na.blyp-sp-hgh.UPF Cl 35.453 Cl.blyp-hgh.UPF ATOMIC_POSITIONS (crystal) Na 0 0 0 Cl0.5 0.5 0.5 K_POINTS (automatic) 32 32 32 0 0 0 pp input: &inputpp prefix = 'NaCl' outdir = '/home/uriargaman/tmp15711/' filplot = 'delta_charge' plot_num= 9 / &plot nfile = 1 filepp(1) = 'delta_charge' weight(1) = 1.0 iflag = 3 output_format = 5 fileout = 'delta_rho.xsf' e1(1) =1.0, e1(2)=0.0, e1(3) = 0.0, e2(1) =0.0, e2(2)=1.0, e2(3) = 0.0, e3(1) =0.0, e3(2)=0.0, e3(3) = 1.0, nx=50, ny=50, nz=50 / Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] problem in post-processing: charge density minus superposition of atomic densities
Dear QE experts I do a PP calculation to get the charge density minus superposition of atomic densities with the following input: &inputpp prefix = 'ti' outdir = '/home/uriargaman/tmp8086/TMP_PLOTDELTARHO/' filplot = 'ti_delta_charge' plot_num= 9 / &plot nfile = 1 filepp(1) = 'ti_delta_charge' weight(1) = 1.0 iflag = 3 output_format = 5 fileout = 'ti.delta_rho.xsf' e1(1) =1.0, e1(2)=0.0, e1(3) = 0.0, e2(1) =0.0, e2(2)=1.0, e2(3) = 0.0, e3(1) =0.0, e3(2)=0.0, e3(3) = 1.0, nx=50, ny=50, nz=50 The most negative value is: -0.008 and the most positive value is: 0.05. In addition, the positive values do not balanced with the negative values (the integral over all space is positive). Why this integral do not vanish? Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] 2013 intel mkl library seems not to work with the 5.4 version of QE
Dear QE users and developers I think that the 5.4 version do not work with the intel 2013 mkl libraries while the 5.3 version works with it. I compile the 5.3 version on our cluster with 2013 intel libraries: Package ID: l_mkl_11.1.2.144 l_ccompxe_2013_sp1.2.144 l_fcompxe_2013_sp1.2.144 This package contains: Intel(R) Math Kernel Library 11.1 Update 2 for Linux* It is finished with no problems. When I compile with the exact same configure the 5.4 version I get the following error: fft_scalar.DFTI.f90(20): #error: can't find include file: mkl_dfti.f90 make[1]: *** [fft_scalar.o] Error 1 I note that this file exist in the /opt/intel directory as I write in the configure: LIBDIRS=/opt/intel. When I compile the 5.4 version on other machine with newer version of mkl (2015): Package ID: l_mkl_11.2.0.090 l_ccompxe_2015.0.090 l_fcompxe_2015.0.090 l_compxe_2015.0.090 This package contains: Intel(R) Math Kernel Library 11.2 for Linux* It works. Because of what I write, I believe that there is a problem of the 5.4 version with the 2013 intel mkl. I want to ask if there are any way to still working with the 2013 intel mkl or I must upgrade it? Thamk you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] problem with compilation the 5.4.0 version
Dear QE users and developers I compile the 5.4.0 version with the following configure: configure LIBDIRS=/opt/intel/mkl/lib/intel64 This is working for older versions. It is success. When I do make all I get: Warning: Nonexistent include directory "/include" In file included from fft_scalar.f90:27:0: fft_scalar.DFTI.f90:20:0: fatal error: mkl_dfti.f90: No such file or directory when I do the configure as: ./install/configure compiler=intel MPIF90=mpiifort FC=ifort F77=ifort F90=ifort CXX=icpc CC=icc I get the following error: /tmp/ifortZkzTF9.i90(5014): catastrophic error: Too many errors, exiting compilation aborted for fft_parallel.f90 (code 1) make[1]: *** [fft_parallel.o] Error 1 make[1]: Leaving directory `/home/argaman/espresso-5.4.0/FFTXlib' make: *** [libfft] Error 1 Thank you Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] charge density vizualization using xcrysden
Dear QE users and devalopers I visualize charge density in three dimensions but I do not succeed to show only a part of the unit cell using xcrysden (I do the calculation for large unit cell). I can do that using xcrysden or I need other visualization tool? In case it cannot be done using xcrysden, which visualization tool is recommended for 3d charge density? Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Acoustic phonon frequencies in Gamma
Dear QE users and developers How phonon calculations is done in the acoustic branches at the Gamma point? What is the perturbation? I know that in ph.x this frequencies can sometimes differs from zero but the displacements of the atoms is just rigid. I know that the deviation from zero is a matter of convergence. Thanks Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] how to obtain exchange energy and correlation energy separately?
hello, I wish to separate the "xc contribution" at the scf.out file into "exchange contribution" and "correlation contribution". What are you recommend? 1. to use the subroutine exxenergy2 from exx.f90 and substract the results from the "xc contribution"? 2. to do something similar to what is written in v_of_rho.f90 for each of the contributions? 3. maybe something smarter? I found similar question in the forum from 2014 with the answer that I have to modify the code. thanks in advance, Uri Argaman Materials engineering department Ben-Gurion university Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] exchange-correlation and pseudopotential
Dear QE users and developers I want to do calculation with MGGA_MS functional for titanium. I do not find pseudopotential with this XC (ultrasoft is preferred). How can I do such calculation? Is the best way to use pslibrary? Is it reasonable to work with different XC than used in the pseudopotential? Thank you very much Uri Argaman Materials Engineering Department Ben-Gurion University of the Negev Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Frozen phonon calculation using supercell in diamond structure
Dear QE users and developers I'm trying to understand the shape of the normal modes in the diamond structure and to calculate the frequency in the X symmetry point using supercell (in addition to the regular DFPT calculation of QE). The atoms in the unit cell in diamond structure are in (0,0,0) and (0.25,0.25,0.25). The wave-vector in X is (2 pi/a,0,0). Therefore, these two atoms have a phase difference of pi/2. Because of that I do not understand how to calculate the frequency using supercell because there are no times in which the kinetic energy are zero. Do I need to calculate the kinetic energy in these kind of modes? can I do it classically? Thank you in advance Uri Argaman Ben-Gurion University Material Engineering Department Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] normal modes visualization
There is a way to visualize phonon normal modes with QE not in the Gamma point (to produce files like dynmat.axsf for other points than Gamma)? Thank you Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Bug in ph.x sometimes occur in hexagonal phases
Thank you Andrea I think the phonon calculated with 13 dynamical matrices is wrong because when I calculate the anisotropic coefficients of thermal expansion in zirconium I do PW calculations as well as phonon calculations in several lattice parameters. In some of them (not in all of them), QE calculate 13 dynamical matrices and in the other 12. When I use this data to calculate the anisotropic coefficient of thermal expansion I got very strange results that do not agree with some publications which also calculate it from DFT. I do not know what is the problem. Maybe there is some problem in q2r.x. Using nr1=30, nr2=30, nr3=48 indeed solve the problem. In addition, different ecut and ecutrho also solve the problem. I now calculate the coefficient of anisotropic thermal expansion in zirconium again with ecut=50 and ecutrho=250. These calculations always produce 12 dynamical matrices. These calculations takes a few days on my machines. When I will get the results I will report if it makes any difference. Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Bug in ph.x sometimes occur in hexagonal phases
Thank you Andrea I change the atomic position to: ATOMIC_POSITIONS (crystal) Zr 0.660.330.25 0 0 0 Zr 0.330.660.75 And it does not fix the problem. dyn0 still contain 13 dynamical matrices. Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] Bug in ph.x sometimes occur in hexagonal phases
Dear QE users and developers When I do phonon calculations in hexagonal phases (titanium and zirconium) in a 4X4X4 q-points grid, QE calculate 12 dynamical matrices. But, in a specific geometries QE calculate 13 dynamical matrices, two of them are the same. dyn0 in that case is: 4 4 4 13 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.000E+00 0.156342786935722E+00 0.000E+00 0.000E+00 -0.312685573871444E+00 0.000E+00 0.288675134594792E+00 0.000E+00 0.000E+00 0.288675134594792E+00 0.156342786935722E+00 0.000E+00 0.288675134594792E+00 -0.312685573871444E+00 0.000E+00 0.288675134594792E+00 -0.156342786935722E+00 0.000E+00 -0.577350269189585E+00 0.000E+00 0.000E+00 -0.577350269189585E+00 0.156342786935722E+00 0.000E+00 -0.577350269189585E+00 -0.312685573871444E+00 0.250E+00 0.433012701892189E+00 0.000E+00 0.250E+00 0.433012701892189E+00 0.156342786935722E+00 0.250E+00 0.433012701892189E+00 -0.312685573871444E+00 You can see that the 5'th and the 7'th dynamical matrix are the same. I want to emphasize that this bug is only at a certain lattice parameter. For example, this is the input files: This is the SCF input: &control calculation='scf' restart_mode='from_scratch', prefix='zr', pseudo_dir = '/home/argaman/espresso_intel/pseudo/', outdir='/home/argaman/tmp/tmp5110/', tstress = .true. , tprnfor = .true. , verbosity='high', wf_collect=.true. / &system ibrav= 4, celldm(1) =5.98584, celldm(3) =1.59905042567 , nat= 2, ntyp= 1, ecutwfc =45,ecutrho=225, occupations='smearing', smearing="methfessel-paxton", degauss=0.01, / &electrons conv_thr = 1.0d-9, electron_maxstep=1000 mixing_beta = 0.7d0, / ATOMIC_SPECIES Zr 91.224 zr_pbe_v1.uspp.F.UPF ATOMIC_POSITIONS (crystal) Zr 0.60.30.250000 0 0 Zr 0.30.60.75000 K_POINTS {automatic} 18 18 18 0 0 0 This is the phonon input: phonon for Zr &inputph tr2_ph=1.0d-15, prefix='zr', fildvscf='aldv', amass(1)=91.224, outdir='/home/argaman/tmp/tmp5110/', fildyn='zr.dyn', trans=.true., ldisp=.true., nq1=4, nq2=4, nq3=4 / I got this bug on versions: 5.0.3, 5.1 and 5.2.0 Thank you Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] using QE with extremely high pressures
Dear QE users and developers I just want to understand a point. What one should consider in scf calculations (with QE) when going to extremely high pressures except (obviously) of being sure to have an adequate pseudopotential and reaching numerical convergence? Thanks, Uri Argaman Material Engineering Department Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] PlotPhonon for FCC structure
Dear QE developers This is the file Cubic_FCC in the Include directory in PlotPhon: Face Centered Cubic 1.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 1.00 # Cartesian 1 0.000 0.000 0.000 G 50 1.000 0.000 0.000 X 50 1.000 1.000 1.000 G 50 0.500 0.500 0.500 L 50 1.000 0.000 0.000 X 50 1.000 0.500 0.000 W 50 0.500 0.500 0.500 L There is two different point called G (gamma). How it is possible that 1.000 1.000 1.000 is the gamma point? Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] projected DOS
Dear QE users and developers I notice that there is no implementation for the tetrahedron method for projected DOS. Is there any way to calculate projected DOS (calculated with projwfc.x) to get comparable results with the total DOS (calculated with dos.x with the tetrahedron method)? Is it possible to extract the details of the DOS using the methods implemented in QE like features of van Hove singularities? In addition, is there any published works using projected DOS calculated with QE? Recently, new codes for DOS calculations published (optados.org). Is there a significant difference compared with the tetrahedron method as implemented in QE? Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] projected density of states
Dear QE users and devalopers In the website: https://wiki.alcf.anl.gov/wparker/index.php/Platinum_projected_density_of_states I see a very interesting comparison between Quantum Espresso, ABINIT and GPAW. It can be seen that Quantum Espresso produced a very smooth projected density of states. Why does it happens? and why there is a significant difference between the different codes? In addition, it is strang that quantum espresso produced not smooth total DOS that have sharp picks (like in this publication: http://iopscience.iop.org/2053-1591/2/1/016505/) but the PDOS is smooth. Thank you in advance Uri Argaman Materisl Engineering Depantment Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] S matrix not positive definite in Fermi surface calculation
Dear QE users and developers I do Fermi surface calculation and I got an error message: "Error in routine cdiaghg (126): S matrix not positive definite" I note that in some crystal geometries I got this message and in other geometries I didn't got this error and the Fermi surface looks fine and reasonable. This is the input file: &control calculation='bands' restart_mode='from_scratch', prefix='ti', pseudo_dir='/home/uriargaman/QE/espresso-5.1/pseudo/', outdir='/home/uriargaman/tmp31971/TMP_FS/', tstress = .true. , tprnfor = .true. , verbosity='high', / &system ibrav= 4, celldm(1) =5.48090015121, celldm(3) =1.6120416161 , nat= 2, ntyp= 1, ecutwfc =50,ecutrho=500, occupations='smearing', smearing="methfessel-paxton", degauss=0.01, nbnd=16 / &electrons conv_thr = 1.0d-9, electron_maxstep=1000 mixing_beta = 0.7d0, / ATOMIC_SPECIES Ti 47.9 Ti.bp-sp-van_ak.UPF ATOMIC_POSITIONS (crystal) Ti 0.60.30.250000 0 0 Ti 0.30.6 0.75000 K_POINTS 4913 What can be the problem and how can I solve it? Thank you in advance Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] "eigenvalues not converged" warnings appear in one compilation but does not appear in another
Dear QE users and developers I compile QE version 5.0.3 on cluster with Intel compilers and I got no warnings in some PW calculation. When I run the same input in another machine with another compilation I got the warning: "c_bands: 1 eigenvalues not converged" several times. I want to mention that this warning did not appear in the last SCF cycle and there is no difference in the energy. There is a very small difference in the stress. Do I have something to worry about? Uri Argaman Ben-Gurion University of the Negev Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
Re: [Pw_forum] How to improve the precision of phonon calculation?
Dear Yu Hailin In questions like this you should show the SCF input that you use. The negative frequencies show that the crystal is unstable or your calculations are not converged. It can be the numerical parameters of the SCF calculation or the numerical parameters of the phonon calculation. As far as I know, the only numerical parameter used in the phonon calculation is tr2_ph. If your calculation is converged than the negative (imaginary) frequencies show that the structure is unstable. You must do relaxation (probably to zero pressure) before the SCF to ensure that the structure is on the minimum of the energy surface (at list local minimum). Best regards Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] accuracy of stress determination of vc-relax
Dear Prop. Giannozzi Thank you very much for your answer. There are two BFGS algorithms in the relaxation process: one for the ions positions and one for the cell parameters. From the name of the file "PW/src/move_ions.f90" I would guess that it is the BFGS algorithm for the ions positions. For getting converged stress I should check them both or it is enough to check just one of them? Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] accuracy of stress determination of vc-relax
Dear QE users and developers I do vc-relax with BFGS algorithm as follows: &CONTROL calculation = "vc-relax", pseudo_dir = '/home/uriargaman/espresso-5.0.3/pseudo/', outdir = '/home/uriargaman/tmp/tmp28898/', tstress=.true. verbosity='high', / &system ibrav= 4, celldm(1) =5.548454954360, celldm(3) =1.581979631777413 , nat= 2, ntyp= 1, ecutwfc =80, ecutrho=1200, occupations='smearing', smearing="methfessel-paxton", degauss=0.01, / &ELECTRONS conv_thr= 1.D-9, electron_maxstep=1000 mixing_beta = 0.7D0, / &IONS ion_dynamics="bfgs", / &cell cell_dynamics="bfgs", press=0.6, press_conv_thr=0.01, / ATOMIC_SPECIES Ti 47.9 Ti.pbe-sp-van_ak.UPF ATOMIC_POSITIONS (crystal) Ti 0.60.30.250000 0 0 Ti 0.30.60.75000 K_POINTS {automatic} 24 24 24 0 0 0 In the last iteration (after begin final coordinates) I got stress: total stress (Ry/bohr**3) (kbar) P= 0.71 0.0485 -0. -0. 0.71 -0.00 -0.00 -0. 0.0485 0. -0.00 0.71 0.00 -0. 0. 0.0481 -0.00 0.00 0.71 and the stress in the iteration before (last iteration before begin final coordinates): total stress (Ry/bohr**3) (kbar) P=0.73 0.0499 -0. 0. 0.73 -0.00 0.00 -0. 0.0499 0. -0.00 0.73 0.00 0. 0. 0.0495 0.00 0.00 0.73 I did'nt reach the pressure threshold although I have very good convergence with respect to all numerical parameters. The threshold is achieved only for hydrostatic condition but not for the value of the stress (P=0.73 kbar instead of 0.6 kbar). Is the parameter "press_conv_thr" is just for the hydrostatic condition and not for the absolute value of the stress? How can I get more accurate stress? Thank you very much Uri Argaman Ben-Gurion University Israel ___ Pw_forum mailing list Pw_forum@pwscf.org http://pwscf.org/mailman/listinfo/pw_forum
[Pw_forum] QE-GPU compilation problems
Dear Filippo Spiga Thank you very much for you response. 1. You write that there is no benefit in using GPU with small systems (less than 40-50 atoms). What about phonon calculations? 2. We want to do a much more complicated calculation but we do not succeed even with the very simple we write about before. 3. We try to work with a simple GPU for now, but we intend to work with a better one, but we need to try it first. 4. What can we do to run a more complicated calculations on the present GPU? Uri Argaman Ben-Gurion University Israel Dear Uri, > you are indeed right, the calculation is simple and it should work. There > are many think that could have done wrong (e.g FFT grid or the multi-plan > CUFFT driver that has not enough space on the card to initialize the FFT > plan). However your case is so simple that adding GPU will not bring you > any benefit. The QE-GPU is designed and has born to accelerate > time-to-solution for mid/large systems. If you have less than 40~50 atoms > (unless your case is particularly computational expensive) then QE-GPU will > not bring you any benefit. > Save you time, keep using the normal QE that works great. > Cheers, > F -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140928/3e0177b6/attachment.html
[Pw_forum] QE-GPU compilation problems
Dear Ivan, Thank you very much for your answer. It's possible that this problem is due to memory issues, but I have GPU with 1GB and from tests that I've done a simple PW calculation of one atom per unit cell doesn't crash with K points: 2 2 2 but does crash with K points: 3 3 3 Both calculations are preformed with ecutwfc=10. It doesn't seems reasonable that 1GB GPU couldn't handle such a simple calculation. Thanks in advance, *Uri Argaman **Ben-Gurion University **Israel* Dear Uri, > > from the error message I believe the memory of your GPU is simply not > capable enough to run such simulation. > Indeed, it fails during an operation of allocation on the GPU memory. > > Kind Regards, > > Ivan > > -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140923/61a0873c/attachment.html
[Pw_forum] QE-GPU compilation problems
Hi again Mr. Spiga, thanks for the reply. I was able to compile with your help. What I did was: 1. configure the main QE 5.1 with --enable-openmp and --disable-parallel. 2. make all the QE 5.1 3. recompile BLAS with -fPIC 4. configure the GPU (from the git master branch) with --enable-openmp --disable-parallel --enable-cuda --with-gpu-arch=sm_30 --with-cuda-dir=/usr/local/cuda-6.5 --with-phigemm --without-magma 5. and make -f Makefile.gpu all-gpu When i try to run pw-gpu, ph-gpu or neb-gpu I get the following error: [vloc_psi_k_cuda_v1.cu:171] Allocation of GPU resources for the plan failed! Program exits... Error condition encountered during test: exit status = 1 The cp-gpu is working. I checked during the run with nvidia-smi and it was using the GPU. Also the samples which come with CUDA work. Maybe you have an idea about this error. Thanks in advance and have a nice holiday, Uri Argaman Ben-Gurion University Israel Date: Mon, 22 Sep 2014 23:49:23 +0800 > From: Filippo Spiga > Subject: Re: [Pw_forum] QE-GPU compilation problems > To: PWSCF Forum > Message-ID: <4925EFBE-FF35-4671-9310-F74CABF15719 at gmail.com> > Content-Type: text/plain; charset=utf-8 > > Dear Uri, > > On Sep 22, 2014, at 10:17 PM, Uri Argaman wrote: > > I tried recompiling BLAS with -fPIC and after that it seems to be able > to compile pw-gpu, neb-gpu, cp-gpu, although it doesn't put pw.x, neb.x etc > in the bin directory, but trying to compile ph-gpu gives the following > error: > > Ok, that missing flag is a case the configure should handle. I will fix > this. I have no machine with GCC 4.8, I have to upgrade one and do some > testing. > > About "pw.x" or "neb.x", QE-GPU makefile does not create them. If you need > the CPU-only executable first run the QE configure in a "standard way" then > run the QE-GPU one. In this way you get both type of executable (but yes, > you need to do twice the procedure). > > Speaking about ph-gpu.x, this is a problem that has been already fixed in > the git master branch. Try to checkout the latest git version of QE-GPU or > please wait me to finalize a new QE-GPU version. As this point, without > pushing too much in release new functionalities, I can create just one new > minor version with a few important bug fixes related to the configure and > makefile. > > F > > -- > Mr. Filippo SPIGA, M.Sc. > http://filippospiga.info ~ skype: filippo.spiga > > ?Nobody will drive us out of Cantor's paradise.? ~ David Hilbert > -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140923/57dc6697/attachment.html
[Pw_forum] QE-GPU compilation problems
Dear Mr. Spiga, thanks for the fast reply. My gcc is 4.8.2 and the Quadro card is K600. On a fresh new QE 5.1 with QE-GPU 14.06 using internal libraries without MAGMA (--without-magma flag in configure) I get: cc -O3 -fopenmp -fPIC -z muldefs -shared -o libphigemm.so.2.0 phigemm_auxiliary.o phigemm_env.o phigemm_dgemm.o phigemm_zgemm.o phigemm_dgemm_specialK.o phigemm_zgemm_specialK.o phigemm_cgemm.o phigemm_sgemm.o -L/usr/local/cuda-6.5/lib64 -lcublas -lcufft -lcudart /home/iftahg/New-QE-GPU/espresso-5.1/phiGEMM/lib/libphigemm.a /home/iftahg/New-QE-GPU/espresso-5.1/BLAS/blas.a -L/usr/local/cuda-6.5/lib64 -lcublas -lcufft -lcudart /usr/bin/ld: /home/iftahg/New-QE-GPU/espresso-5.1/BLAS/blas.a(sgemm.o): relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC /home/iftahg/New-QE-GPU/espresso-5.1/BLAS/blas.a: error adding symbols: Bad value collect2: error: ld returned 1 exit status make[3]: *** [shared] Error 1 make[3]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/phiGEMM/src' make[2]: *** [phigemm] Error 2 make[2]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/phiGEMM' make[1]: *** [libphiGEMM] Error 2 make[1]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/GPU/install' make: *** [libphiGEMM] Error 2 I tried recompiling BLAS with -fPIC and after that it seems to be able to compile pw-gpu, neb-gpu, cp-gpu, although it doesn't put pw.x, neb.x etc in the bin directory, but trying to compile ph-gpu gives the following error: make[6]: Entering directory `/home/iftahg/New-QE-GPU/espresso-5.1/PW/src' if test -n "" ; then \ ( cd ../.. ; make || exit 1 ) ; fi gfortran -g -fopenmp -o pw.x \ pwscf.o libpw.a ../../Modules/libqemod.a ../../flib/ptools.a ../../flib/flib.a ../../clib/clib.a ../../iotk/src/libiotk.a /home/iftahg/New-QE-GPU/espresso-5.1/GPU/..//lapack-3.2/lapack.a /home/iftahg/New-QE-GPU/espresso-5.1/GPU/..//phiGEMM/lib/libphigemm.a /home/iftahg/New-QE-GPU/espresso-5.1/GPU/..//BLAS/blas.a -L/usr/local/cuda-6.5/lib64 -lcublas -lcufft -lcudart libpw.a(addusdens.o): In function `addusdens_': /home/iftahg/New-QE-GPU/espresso-5.1/PW/src/addusdens.f90:28: undefined reference to `addusdens_g_gpu_' libpw.a(newd.o): In function `__dfunct_MOD_newq': /home/iftahg/New-QE-GPU/espresso-5.1/PW/src/newd.f90:33: undefined reference to `newq_compute_gpu_' /home/iftahg/New-QE-GPU/espresso-5.1/PW/src/newd.f90:33: undefined reference to `newq_compute_gpu_' libpw.a(vloc_psi.o): In function `vloc_psi_gamma_': /home/iftahg/New-QE-GPU/espresso-5.1/PW/src/vloc_psi.f90:41: undefined reference to `vloc_psi_gamma_gpu_' libpw.a(vloc_psi.o): In function `vloc_psi_k_': /home/iftahg/New-QE-GPU/espresso-5.1/PW/src/vloc_psi.f90:230: undefined reference to `vloc_psi_k_gpu_' ../../Modules/libqemod.a(mp.o): In function `__mp_MOD_mp_end': /home/iftahg/New-QE-GPU/espresso-5.1/Modules/mp.f90:213: undefined reference to `closecudaenv_' ../../Modules/libqemod.a(mp.o): In function `__mp_MOD_mp_start': /home/iftahg/New-QE-GPU/espresso-5.1/Modules/mp.f90:184: undefined reference to `initcudaenv_' collect2: error: ld returned 1 exit status make[6]: *** [pw.x] Error 1 make[6]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/PW/src' make[5]: *** [pw] Error 1 make[5]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/PW' make[4]: *** [pw] Error 1 make[4]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1' make[3]: *** [tldeps] Error 1 make[3]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/PHonon/PH' make[2]: *** [phonon] Error 1 make[2]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/PHonon' make[1]: *** [phonon] Error 1 make[1]: Leaving directory `/home/iftahg/New-QE-GPU/espresso-5.1/install' make: *** [ph-gpu] Error 2 Which is similar to the error I got trying to compile with OpenBLAS (after making the appropriate changes to the make.sys and the make_phiGEMM.inc files). Thanks in advance, Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140922/9e14c164/attachment.html
[Pw_forum] QE-GPU compilation problems
Dear QE users and developers, I'm trying to compile GPU-accelerated Quantum Espresso and I'm running into troubles. I'm using QE-GPU 14.03 and I have attempted to compile it with QE 5.1 and QE 5.0.2. My system specifications are: intel xeon CPU, nvidia quadro graphic card with Compute Capability of 3.0, CUDA 6.5, Ubuntu 14.04 lts, OpenMPI 1.8.2, gfortran. If I understand correctly, there is no patch needed for the QE 5.1 version. On the QE 5.0.2 I used the espresso-5.0.2_GPU-14.03.patch. Since I don't have Intel's MKL, I started by trying to compile it with the internal libraries: cd GPU ./configure --enable-openmp --disable-parallel --enable-cuda --with-gpu-arch=sm_30 --with-cuda-dir=/usr/local/cuda-6.5 --with-phigemm --with-magma cd .. make -f Makefile.gpu pw-gpu I got Consistent warnings about nonexistent quark include directory and the error: gfortran -g -fopenmp -o pw-gpu.x \ pwscf.o ../../PW/src/libpw.a libpwgpu.a ../../Modules/libqemod.a ../Modules/libqemodgpu.a ../../flib/ptools.a ../../flib/flib.a ../../clib/clib.a ../../iotk/src/libiotk.a /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//qe-magma/lib/libmagma.a /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//lapack-3.2/lapack.a /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//phiGEMM/lib/libphigemm.a /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//BLAS/blas.a -L/usr/local/cuda-6.5/lib64 -lcublas -lcufft -lcudart /usr/bin/ld: /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//qe-magma/lib/libmagma.a(dtrevc3_mt.o): undefined reference to symbol '__cxa_pure_virtual@@CXXABI_1.3' //usr/lib/x86_64-linux-gnu/libstdc++.so.6: error adding symbols: DSO missing from command line collect2: error: ld returned 1 exit status make[1]: *** [pw-gpu.x] Error 1 I tried adding -lstdc++ flag which resulted in many undefined reference errors: /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//qe-magma/lib/libmagma.a(zgeqp3.o): In function `magma_zgeqp3': zgeqp3.cpp:(.text+0x33f): undefined reference to `cblas_dznrm2' /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//qe-magma/lib/libmagma.a(zlaqps.o): In function `magma_zlaqps': zlaqps.cpp:(.text+0x207): undefined reference to `cblas_idamax' zlaqps.cpp:(.text+0x97c): undefined reference to `cblas_dznrm2' zlaqps.cpp:(.text+0xa15): undefined reference to `cblas_dznrm2' /home/iftahg/QE-GPU_2/espresso-5.1/GPU/..//qe-magma/lib/libmagma.a(zlatrd.o): In function `magma_zlatrd': zlatrd.cpp:(.text+0x716): undefined reference to `cblas_zdotc_sub' zlatrd.cpp:(.text+0xb25): undefined reference to `cblas_zdotc_sub' ... I also tried to compile with OpenBLAS and with the most recent FFTW instead of the internal libraries. ./configure --enable-openmp --disable-parallel --enable-cuda --with-gpu-arch=sm_30 --with-cuda-dir=/usr/local/cuda-6.5 --with-phigemm --with-magma FFT_LIBS=/usr/local/lib/libfftw3.a BLAS_LIBS=/home/iftahg/espresso-5.1/OpenBLAS/libopenblas_haswellp-r0.2.11.a and I got: gfortran -O3 -g -fopenmp -x f95-cpp-input -fopenmp -D__GFORTRAN -D__STD_F95 -D__FFTW -D__CUDA -D__PHIGEMM -D__MAGMA -DHAVE_CUBLAS -D__OPENMP -I../include -I/home/iftahg/QE-GPU_2/espresso-5.0.2/GPU/..//qe-magma/include -I/home/iftahg/QE-GPU_2/espresso-5.0.2/GPU/..//quark/include -I/home/iftahg/QE-GPU_2/espresso-5.0.2/GPU/..//phiGEMM/include -I/usr/local/cuda-6.5/include -I../../iotk/src -I../../PW/src -I../../Modules -I../Modules -I. -c newq_compute_gpu.f90 Warning: Nonexistent include directory "/home/iftahg/QE-GPU_2/espresso-5.0.2/GPU/..//quark/include" newq_compute_gpu.f90:30.6: USE mp_bands, ONLY : intra_bgrp_comm 1 Fatal Error: Can't open module file 'mp_bands.mod' for reading at (1): No such file or directory make[1]: *** [newq_compute_gpu.o] Error 1 make[1]: Leaving directory `/home/iftahg/QE-GPU_2/espresso-5.0.2/GPU/PW' Does any of these errors ring a bell? Is it plausible that getting Intel's MKL would solve my problems? Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140921/12157029/attachment.html
[Pw_forum] wrong total_weight in matdyn.x
Dear QE users and devalopers I do a phonon calculation of pure titanium in orthorhombic phase with 4 atoms in the unit cell: ibrav= 8 Ti -0.0 0.11000 0.250000 0 0 Ti -0.00011 -0.103719384 0.74888 Ti 0.50007 0.610015023 0.25018 Ti 0.49990 0.396287113 0.74901 The calculation succeed when I run it with: ecut=50 and ecutrho=500 (I use ultrasoft PP). When I increase the cutoff to ecut=50 and ecutrho=500 and don't change anything else I got this error in the from matdyn: Error in routine frc_blk (1): wrong total_weight Relaxation to 1000 kbar work good and the stress is: total stress (Ry/bohr**3) (kbar) P= 1000.04 0.00679839 0. 0. 1000.08 0.00 0.00 0. 0.00679814 0. 0.00 1000.04 0.00 0. 0. 0.00679787 0.00 0.00 1000.00 What is the problem? Thank you very much Uri Argaman Ben-Gurion university Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140914/c6479eda/attachment.html
[Pw_forum] PAW pseudopotentials
Dear QE users and developers Did anyone knows where I can find PAW pseudopotentials with GGA exchange-correlation for titanium? Thank you in advance Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140729/f5273190/attachment.html
[Pw_forum] matdyn.modes
dear QE users and developers I understand from previous messages that matdyn.modes contain the modes as displacement of the atoms in the unit cell and I understand that there is a real part and imaginary part but I don't understand what is the meaning of the imaginary part of the displacement of the atoms. In addition, in hexagonal lattice, the branches in the spectrum are represents transverse and longitudinal modes, or it is mixed? Thank you very much Uri Argaman Ben-Gurion University, Beer-Sheva, Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140717/67dfbf8a/attachment.html
[Pw_forum] smearing contrib. (-TS)
After carful check of the input file, the Fermi-Dirac smearing generates positive entropy. Sorry about the confusion. Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140614/32cc9441/attachment.html
[Pw_forum] smearing contrib. (-TS)
Dear QE users and developers I am doing calculations with QE of omega phase of titanium (hexagonal phase with 3 atoms per primitive unit cell). I am doing this calculations with both Methfessel-Paxton and Fermi-Dirac smearing. In both calculations I have got negative entropy (from PW output): with Methfessel-Paxton Smearing: smearing contrib. (-TS) = 0.7877 Ry and with Fermi-Dirac smearing: smearing contrib. (-TS) = 0.0254 Ry why does it happen? best regards Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140612/d778b35c/attachment.html
[Pw_forum] Density of States
Dear QE users, I do density of states calculations using QE and I try to calculate the electronic heat capacity of metals with the electronic DOS: 1.What is the energy in the DOS file? Is it the eigenvalues? How can I calculate the total energy or the one electron contribution from the DOS? (I try to do the integral: DOS(E)*E*dE from minus Infinity to E Fermi and I think it should be equal to the one electron contribution, but it does not). 2.What do you know about the excited states? DFT is a good approximation for excited states? Thank you very much Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140608/1e2ea8e2/attachment.html
[Pw_forum] phonon parallelism
Dear QE users I'm doing a phonon calculation with 4 q-points in each direction in hexagonal lattice. As a result, the program ph.x calculate 12 dynamical matrices, each matrix have a different run time (most of the time there is a dig difference). What is the best parallelization in cases like this? Thank you Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140416/4a7509fd/attachment.html
[Pw_forum] Gibbs Free energy
Dear Pratik, Within the QHA you must do calculations in constant volume. Therefore, it is straightforward to calculate the Helmholtz free energy. To calculate the Gibbs free energy you must know the pressure. To calculate the pressure you must calculate several volumes. You can calculate the Helmholtz free energy at several volumes in some temperature and use the fact that the pressure is minus the partial derivative of the Helmholtz free energy with respect to the volume (to do so, you can fit the Helmholtz free energy to some analytical function, e.g. parabola of Birch-Murnaghan equation of state). In addition, I want to mention a few things: 1.Recently, a bug in the QHA has been corrected in several structures. It is corrected in the SVN version. 2.In this approach, you must work with hydrostatic pressure. To do so, you need to do relaxation. 3.The pressure you see in the SCF output file is the pressure at zero Kelvin without phonon zero point energy contribution. It is not the pressure at non zero temperatures. 4.There is also a thermal electronic contribution to both Helmholtz free energy and Gibbs free energy. In most cases, it is very small relative to the phonon contribution, but you cannot know for sure until you calculate it. To calculate it, you need to calculate the electron DOS with dos.x. I hope I helped. Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140405/e52e3958/attachment.html
[Pw_forum] QHA: There is still a bug in the DOS calculation
Dear Paolo Giannozzi I reported a bug in the QHA DOS calculation a few weeks ago in hexagonal phases. The reason of this bug is that the ratio c/a in reciprocal lattice is a/c. I see that you corrected this bug in the hexagonal case but not in other cases, for example ibrav=6 (tetragonal). Thank you Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140328/1d2e555c/attachment.html
[Pw_forum] Bug in the QHA code of hexagonal phases
In the QHA package in the QE 5.0.3 there is a bug in calculating the q-points grid to the matdyn.x in the DOS calculation using tetra.x . The ratio c/a reading from the .fc file suppose to be 1/(c/a). To correct this bug you should change the line c2a=`echo "scale=8;$c2a1/2" | bc -l` in the [espresso directory]/QHA/Include/Tetrahedra To c2a=`echo "scale=8;1/$c2a1/2" | bc -l` Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140319/1e79071a/attachment.html
[Pw_forum] QHA for hexagonal phases
Dear QE users I want to calculate thermodynamic properties of hexagonal phases (not HCP, c/a not equal to (8/3)^0.5) using QHA package of the QE. I need the tetrahedra file for hexagonal Brave lattice with a arbitrary c/a. Where can I found this file or where can I find instruction how to write these files. Thank you Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140317/1b1a983f/attachment.html
[Pw_forum] problem with relaxation and difference between SCF Fermi level and DOS Fermi level
Dear QE users 1. When I do vc-relax calculation to determine the pressure I get the desired pressure but in the last iteration I get a large jump in the pressure (after "begin final coordinate") sometimes in the order of 1000 kbar. Why is it happening and how can I fix it? 2. I notice that the Fermi level in the SCF file is not the same as the Fermi level in the DOS file. Why is that? Thank you very much Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140313/a3f5ca88/attachment.html
[Pw_forum] pseudopotential with PAW for titanium
Where I can found pseudopotential with PAW for titanium? Thank you Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140228/83417e34/attachment.html
[Pw_forum] QHA
Where can I found documentation or some instructions about QHA calculations with QE, especially free energy calculations? Thank you Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140119/5a97f428/attachment.html
[Pw_forum] phonon calculations of transition metals in hexagonal phases
does anyone knows a work on phonon calculations of transition metals in hexagonal phases with Quantum-Espresso? Uri Argaman Ben-Gurion University Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140111/9929b00c/attachment.html
[Pw_forum] phonon calculations in titanium
I have done phonon calculation in Titanium in both Alpha (non-ideal HCP) and Omega (hexagonal with 3 atoms per unit cell) phases after relaxation. When I done the phonon calculation with 2 q-points in each direction (Monkhorst-Pack) I have got a reasonable phonon spectrum with q2r.x and PlotPhon, but when I have done a phonon calculation with 3 q-points or 4 q-points in each direction I have got very large negative frequencies. I am working with degauss=0.01 Ry or degauss=0.005 Ry because this is the smearing in which the energy converge to 10^(-5) Ry. The energy is converge to 10^(-6) Ry in respect to all the numerical parameters (ecutwfc,ecutrho,k-points). When I increase the smearing to degauss=0.2 Ry I have got reasonable phonon spectrum with wiggling. From calculations I have done in zero Kelvin (only SCF without phonons) I found the phase transition in zero Kelvin in reasonable pressure only with degauss<0.01 Ry. To sum up, the cold curve looks reasonable with degauss<0.01 Ry but the phonon spectrum looks reasonable only with degauss approximately 0.2 Ry. What is your recommendation? In addition I would like to ask if the q2r.x and/or PlotPhon apply any ASR and what is the difference from matdyn.x in the sence that plotphon does not use *.freq but only *.fc file? I can put homogenous grid in the input file of matdyn.x. Is it causes any problems with hexagonal lattice? -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20140105/e33e8e0b/attachment.html
[Pw_forum] unstable bcc phase in titanium
Thank you *Jiayi Yan.* *Can I get somehow the code that works with QE?* -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20131107/7894b314/attachment.html
[Pw_forum] unstable bcc phase in titanium
I do a phonon calculation for titanium (with the ph.x). For the beta phase (bcc), I get negative frequencies in all pressures. It is known from the literature that the beta phase in titanium is stable only in high temperatures. What can I do to calculating the phonon spectrum in high temperatures? -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20131031/c297c3bb/attachment.html
[Pw_forum] negative frequencies....
I know this subject has already been discussed a lot here. But I did not understand yet the hole meaning of the negative frequency. 1. unstable structure should lead to imaginary frequency not a negative one. so how come it become negative in the phonon dispersion? is it related to the square of it? 2. how can we distinguish between unstable structure and problems with an convergence ones? 3. we have done a calculation of Ti hcp with q-point up to 2 2 2 and got no negative frequencies and good agreements with older published calculations. After change it to 4 4 4 q-points grid we got wiggles and negative frequencies. 4. Is there a simple way of being aware of Kohn anomalies or we must use el-ph coupling for that? thanks very much, Uri. -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20131010/5a58fff0/attachment.html
[Pw_forum] Phonon calculation
Thank you for your reply. >* It's worth to set nimage equal to the number of q-points *>* in the phonon >calculation? * not necessarily so. Different wave-vectors may have different workloads: not by order of magnitudes, but a low-symmetry wave-vector may easily take 5 times more CPU than a high-symmetry one. But still as far as I understand is there any cost of doing it? Am I loosing somthing by puting nimage=number of q-points in the dyn0 file? I understand that some of the parallel images will be finished before the other, But there is nothing I can do to speed it even more. >* In addition I want to ask how the complexity of the phonon calculation*>* >depends on the number of the k-points (in the PW calculation) * roughly linearly, but the number of actual k-points in phonon calculations depends upon the symmetry and the wave-vector >* and how the number of k-points affects the accuracy of the phonon*>* >calculation. * more or less (more rather than less) in the same way as it affects the calculation of other structural properties; quantities related to the electric field (effective charges and dielectric tensors) tend to converge more slowly with the number of k-points P. -- Paolo Giannozzi, Dept. Chemistry&Physics&Environment, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222 On 7 October 2013 09:48, Uri Argaman wrote: > It's worth to set nimage equal to the number of q-points in the phonon > calculation? In addition I want to ask how the complexity of the phonon > calculation depends on the number of the k-points (in the PW calculation) > and how the number of k-points affects the accuracy of the phonon > calculation. > > thank you > > Uri Argaman > > Ben-Gurion University, Israel > -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20131010/1aa891b1/attachment.html
[Pw_forum] Phonon calculation
It's worth to set nimage equal to the number of q-points in the phonon calculation? In addition I want to ask how the complexity of the phonon calculation depends on the number of the k-points (in the PW calculation) and how the number of k-points affects the accuracy of the phonon calculation. thank you Uri Argaman Ben-Gurion University, Israel -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20131007/ef528953/attachment.html