[Wien] Problem with force minimization
Dear all, I'm trying to do force minimization using WIEN2K. I've tried different schemes including PORT, NEW1 and NEWT, with different step sizes for ionic movements, but none of them has worked so far. Regardless of them method I use, the forces reach to certain values after few steps and then remain unchnaged (better to say that they tends to oscillate around some certain values). My system is rather simple and due to symmetry consideration, the force values need to be minimized only along the z axis. For minimization, I use the command: min_lapw -j run_lapw -I -fc 1. - ec 0.1 -i 40 Could somebody tell me what I need to add or change in my command line or in the input files, in order to optimize the atomic positions. Your help is greatly appreciated, Sincerely, S. Bahramy
[Wien] Compilation problem of lapw0_mpi(fftw)
Dear WIEN2k Users, I am trying to install WIEN2k_10.1 on a cluster computer. But I got an error in compiling lapw0_mpi. The error message is : - mpif90 -FR -w -mp1 -prec_div -pad -ip -DINTEL_VML -O3 -xP -DParallel -c W2kinit.F mpif90 -o lapw0_mpi cputim.o modules.o reallocate.o ainv.o am05_xscss.o b88.o blyp.o brj *---* zfft3d.o W2kutils.o W2kinit.o -L/opt/intel/fce/9.1.052/lib -i-static -Bstatic -lguide -lguide_stats -lsvml -Bdynamic -lpthread -L/opt/intel/mkl/10.0.2.018/lib/em64t -Bstatic -lmkl_scala pack -lmkl_blacs_lp64 -lmkl_blacs_ilp64 -lmkl_lapack -lmkl_em64t -lguide -Bdynamic -lpth read -L/usr/local/fftw/lib -lfftw_mpi -lfftw -L/usr/local/mpich-1.2.7/lib -lmpich fftw_para.o(.text+0x7c): In function `exec_fftw_para_': : undefined reference to `fftwnd_f77_mpi_' fftw_para.o(.text+0xba): In function `exec_fftw_para_': : undefined reference to `fftwnd_f77_mpi_' fftw_para.o(.text+0xf5): In function `init_fftw_para_': : undefined reference to `fftw3d_f77_mpi_create_plan_' fftw_para.o(.text+0x11d): In function `init_fftw_para_': : undefined reference to `fftw3d_f77_mpi_create_plan_' fftw_para.o(.text+0x143): In function `init_fftw_para_': : undefined reference to `fftwnd_f77_mpi_local_sizes_' make[1]: *** [lapw0_mpi] Error 1 make[1]: Leaving directory `/usr/local/WIEN2k_10/SRC_lapw0' make: *** [para] Error 2 (END) - os:Redhat Linux ifort:9.1.052 mkl:10.0.2.018 mpich1.2.7 fftw2.1.5 (with --enable-mpi switch) My Makefile option is : FC = ifort MPF = mpif90 CC = cc FOPT = -FR -w -mp1 -prec_div -pad -ip -DINTEL_VML -O3 -xP FPOPT = -FR -w -mp1 -prec_div -pad -ip -DINTEL_VML -O3 -xP DParallel = '-DParallel' FGEN = $(PARALLEL) LDFLAGS = -L/opt/intel/fce/9.1.052/lib -i-static -Bstatic -lguide -lguide_stats -lsvml -Bdynamic -lpthread R_LIBS = -L/otp/intel/mkl/10.0.2.018/lib/em64t -Bstatic -lmkl_lapack -lmkl_em64t -lguide -Bdynamic -lpthread RP_LIBS = -L/opt/intel/mkl/10.0.2.018/lib/em64t -Bstatic -lmkl_scalapack -lmkl_blacs_lp64 -lmkl_blacs_ilp64 -lmkl_lapack -lmkl_em64t -lguide -Bdynamic -lpthread -L/usr/local/fftw/lib -lfftw_mpi -lfftw -L/usr/local/mpich-1.2.7/lib -lmpich Though I refered to WIEN mailing list and FAQ on web for option/flags, I still got an compiling error of fftw. I am wondering if anyone could help me. Regards --- Hiroki Takahashi hiroki_takahashi at mitsui-kinzoku.co.jp Mitsui Mining Smelting Co., Ltd. ---
[Wien] Compilation problem of lapw0_mpi(fftw)
Dear Hiroki Takahashi, Have you performed configure/make/make install with ifort compiler? It seems that you don't have appropriate libraries after installation of the fftw suite. I used the following command to configure it: ./configure --prefix=$HOME/lib/fftw2 LDFLAGS=-L/share/mpi/mvapich/lib CC=icc FC=ifort F77=ifort --enable-mpi Try to reinstall fftw and recompile lapw0. I hope that will help you. Best regards, - Maxim Rakitin 26.10.2010 13:45, Hiroki Takahashi ?: Dear WIEN2k Users, I am trying to install WIEN2k_10.1 on a cluster computer. But I got an error in compiling lapw0_mpi. The error message is : - mpif90 -FR -w -mp1 -prec_div -pad -ip -DINTEL_VML -O3 -xP -DParallel -c W2kinit.F mpif90 -o lapw0_mpi cputim.o modules.o reallocate.o ainv.o am05_xscss.o b88.o blyp.o brj *---* zfft3d.o W2kutils.o W2kinit.o -L/opt/intel/fce/9.1.052/lib -i-static -Bstatic -lguide -lguide_stats -lsvml -Bdynamic -lpthread -L/opt/intel/mkl/10.0.2.018/lib/em64t -Bstatic -lmkl_scala pack -lmkl_blacs_lp64 -lmkl_blacs_ilp64 -lmkl_lapack -lmkl_em64t -lguide -Bdynamic -lpth read -L/usr/local/fftw/lib -lfftw_mpi -lfftw -L/usr/local/mpich-1.2.7/lib -lmpich fftw_para.o(.text+0x7c): In function `exec_fftw_para_': : undefined reference to `fftwnd_f77_mpi_' fftw_para.o(.text+0xba): In function `exec_fftw_para_': : undefined reference to `fftwnd_f77_mpi_' fftw_para.o(.text+0xf5): In function `init_fftw_para_': : undefined reference to `fftw3d_f77_mpi_create_plan_' fftw_para.o(.text+0x11d): In function `init_fftw_para_': : undefined reference to `fftw3d_f77_mpi_create_plan_' fftw_para.o(.text+0x143): In function `init_fftw_para_': : undefined reference to `fftwnd_f77_mpi_local_sizes_' make[1]: *** [lapw0_mpi] Error 1 make[1]: Leaving directory `/usr/local/WIEN2k_10/SRC_lapw0' make: *** [para] Error 2 (END) - os:Redhat Linux ifort:9.1.052 mkl:10.0.2.018 mpich1.2.7 fftw2.1.5 (with --enable-mpi switch) My Makefile option is : FC = ifort MPF = mpif90 CC = cc FOPT = -FR -w -mp1 -prec_div -pad -ip -DINTEL_VML -O3 -xP FPOPT = -FR -w -mp1 -prec_div -pad -ip -DINTEL_VML -O3 -xP DParallel = '-DParallel' FGEN = $(PARALLEL) LDFLAGS = -L/opt/intel/fce/9.1.052/lib -i-static -Bstatic -lguide -lguide_stats -lsvml -Bdynamic -lpthread R_LIBS = -L/otp/intel/mkl/10.0.2.018/lib/em64t -Bstatic -lmkl_lapack -lmkl_em64t -lguide -Bdynamic -lpthread RP_LIBS = -L/opt/intel/mkl/10.0.2.018/lib/em64t -Bstatic -lmkl_scalapack -lmkl_blacs_lp64 -lmkl_blacs_ilp64 -lmkl_lapack -lmkl_em64t -lguide -Bdynamic -lpthread -L/usr/local/fftw/lib -lfftw_mpi -lfftw -L/usr/local/mpich-1.2.7/lib -lmpich Though I refered to WIEN mailing list and FAQ on web for option/flags, I still got an compiling error of fftw. I am wondering if anyone could help me. Regards --- Hiroki Takahashi hiroki_takahashi at mitsui-kinzoku.co.jp Mitsui Mining Smelting Co., Ltd. --- ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
[Wien] Problem with force minimization
Dear Saeed, What system do you investigate? Did you perform scf-calculation without minimization? Was it successful? Which errors did you get during minimization? In case you need to minimize forces along the z-axis only, you can freeze atom positions in case.inM file. There is a nice manual on structure optimization here: http://www.wien2k.at/reg_user/textbooks/Optimization-Notes.pdf. You can find all answers on your question in the document. Best regards, - Maxim Rakitin 26.10.2010 13:36, Saeed Bahramy ?: Dear all, I'm trying to do force minimization using WIEN2K. I've tried different schemes including PORT, NEW1 and NEWT, with different step sizes for ionic movements, but none of them has worked so far. Regardless of them method I use, the forces reach to certain values after few steps and then remain unchnaged (better to say that they tends to oscillate around some certain values). My system is rather simple and due to symmetry consideration, the force values need to be minimized only along the z axis. For minimization, I use the command: min_lapw -j run_lapw -I -fc 1. -ec 0.1 -i 40 Could somebody tell me what I need to add or change in my command line or in the input files, in order to optimize the atomic positions. Your help is greatly appreciated, Sincerely, S. Bahramy ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
[Wien] Problem with force minimization
Dear Saeed, What is the maximum value of forces in your calculations now? By default, minimization stops when the forces are less than 2.0 mRy/a.u. (case.inM file, the first line, the second field), if you didn't change it for less value. Best regards, - Maxim Rakitin 26.10.2010 14:17, Saeed Bahramy ?: Dear Maxim, Thanks for your prompt reply. Actually I get no error. My problem is that force values do not converge to what I want (1 mRy/a.u). I have a hexagonal unit cell containing 6 atoms. The x and y coordinates are at hight symmetry points (1/3 and 2/3) so basically the force values along x and y are zero. Due to the same reason, the program itself freezes the atomic x and y positions in case.inM. I have no problem with scf-calculation. I invoke the command run_lapw -I -ec 0.1 -fc 1.0 -i 40 and after few iterations, I get what I expect. So far so good. The problem starts when I want to optimize the atomic positions. As I mentioned it doesn' matter what scheme I choose in case.inM (PORT, NEW1 or NEWT), for the first few steps, The force values decrease, but then they remain unchanged. I have already read the tutorial which you linked it for me, but it just gives some general idea how to do force minimization using WIEN2K. Sincerely, Saeed On Oct 26, 2010, at 4:58 PM, Maxim Rakitin wrote: Dear Saeed, What system do you investigate? Did you perform scf-calculation without minimization? Was it successful? Which errors did you get during minimization? In case you need to minimize forces along the z-axis only, you can freeze atom positions in case.inM file. There is a nice manual on structure optimization here: http://www.wien2k.at/reg_user/textbooks/Optimization-Notes.pdf. You can find all answers on your question in the document. Best regards, - Maxim Rakitin 26.10.2010 13:36, Saeed Bahramy ?: Dear all, I'm trying to do force minimization using WIEN2K. I've tried different schemes including PORT, NEW1 and NEWT, with different step sizes for ionic movements, but none of them has worked so far. Regardless of them method I use, the forces reach to certain values after few steps and then remain unchnaged (better to say that they tends to oscillate around some certain values). My system is rather simple and due to symmetry consideration, the force values need to be minimized only along the z axis. For minimization, I use the command: min_lapw -j run_lapw -I -fc 1. -ec 0.1 -i 40 Could somebody tell me what I need to add or change in my command line or in the input files, in order to optimize the atomic positions. Your help is greatly appreciated, Sincerely, S. Bahramy ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
[Wien] Problem with force minimization
Dear Maxim, It is around 15 mRy/a.u. on each atom. In my case.inM I've set the force criterion to 1.0 mRy/a.u.. I guess the atomic positions are very close to the optimal ones, but probably my trial steps are rather large Sincerely, Saeeed On Oct 26, 2010, at 5:27 PM, Maxim Rakitin wrote: Dear Saeed, What is the maximum value of forces in your calculations now? By default, minimization stops when the forces are less than 2.0 mRy/ a.u. (case.inM file, the first line, the second field), if you didn't change it for less value. Best regards, - Maxim Rakitin 26.10.2010 14:17, Saeed Bahramy ?: Dear Maxim, Thanks for your prompt reply. Actually I get no error. My problem is that force values do not converge to what I want (1 mRy/a.u). I have a hexagonal unit cell containing 6 atoms. The x and y coordinates are at hight symmetry points (1/3 and 2/3) so basically the force values along x and y are zero. Due to the same reason, the program itself freezes the atomic x and y positions in case.inM. I have no problem with scf-calculation. I invoke the command run_lapw -I -ec 0.1 -fc 1.0 -i 40 and after few iterations, I get what I expect. So far so good. The problem starts when I want to optimize the atomic positions. As I mentioned it doesn' matter what scheme I choose in case.inM (PORT, NEW1 or NEWT), for the first few steps, The force values decrease, but then they remain unchanged. I have already read the tutorial which you linked it for me, but it just gives some general idea how to do force minimization using WIEN2K. Sincerely, Saeed On Oct 26, 2010, at 4:58 PM, Maxim Rakitin wrote: Dear Saeed, What system do you investigate? Did you perform scf-calculation without minimization? Was it successful? Which errors did you get during minimization? In case you need to minimize forces along the z-axis only, you can freeze atom positions in case.inM file. There is a nice manual on structure optimization here: http://www.wien2k.at/reg_user/textbooks/Optimization-Notes.pdf . You can find all answers on your question in the document. Best regards, - Maxim Rakitin 26.10.2010 13:36, Saeed Bahramy ?: Dear all, I'm trying to do force minimization using WIEN2K. I've tried different schemes including PORT, NEW1 and NEWT, with different step sizes for ionic movements, but none of them has worked so far. Regardless of them method I use, the forces reach to certain values after few steps and then remain unchnaged (better to say that they tends to oscillate around some certain values). My system is rather simple and due to symmetry consideration, the force values need to be minimized only along the z axis. For minimization, I use the command: min_lapw -j run_lapw -I -fc 1. -ec 0.1 -i 40 Could somebody tell me what I need to add or change in my command line or in the input files, in order to optimize the atomic positions. Your help is greatly appreciated, Sincerely, S. Bahramy ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien ___ Wien mailing list Wien at zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
[Wien] Problem with force minimization
Verify whether your muffin tin spheres have not become almost touching (x nn, watch the screen output, or look in case.outputnn). If that does not help, please send us grep :FOR case.scf_mini grep :POS case.scf_mini Stefaan
[Wien] accuracy problem after running x sgroup
INDIA -- next part -- An HTML attachment was scrubbed... URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20101026/b9519772/attachment.htm