Activity for Roger Mason
-
Roger Mason posted a comment on a wiki page
Hello, I am trying to run xsd2pgschema on FreeBSD 13.3. I have the following setup: xsd2pgschema.jar@ -> xsd2pgschema-jdk17.jar I have created a script to start the application: !/usr/local/bin/zsh -f export JAVA_OPT_X2P="--add-opens java.base/java.lang=ALL-UNNAMED --add-opens java.base/java.math=ALL-UNNAMED --add-opens java.base/java.util=ALL-UNNAMED --add-opens java.base/java.util.concurrent=ALL-UNNAMED --add-opens java.base/java.net=ALL-UNNAMED --add-opens java.base/java.text=ALL-UNNAMED --add-opens...
-
It turned out that using gmake -jn was the culprit. Sorry for the noise. Thanks, Roger
-
Hello, I compiled elk-9.5.14 with the attached make.inc. When I run 'gmake test' all the tests crash like this: Running test in directory test_035... Abort(134862095): Fatal error in internal_Init: Other MPI error, error stack: internal_Init(66)....: MPI_Init(argc=0x0, argv=0x0) failed MPII_Init_thread(222): gpu_init failed I'm using: mpifort -v mpifort for MPICH version 4.1.3 Using built-in specs. COLLECT_GCC=gfortran13 COLLECT_LTO_WRAPPER=/usr/local/libexec/gcc13/gcc/x86_64-portbld-freebsd13.2/13.2.0/lto-wrapper...
-
Hello, I have added GlusterFS to my small home cluster so as to be able to run MPI jobs. I am encountering a crash when I run a job on the gluster volume: Elk code version 7.1.14 started Number of MPI processes : 1 Number of OpenMP threads per MPI process : 4 Total number of threads : 4 Maximum OpenMP nesting level : 4 Number of threads at first nesting level : 4 Number of MKL threads : 4 Info(elk): current task : 2 Info(checkmt): reduced muffin-tin radius of species 1 (Si) from 2.2000 to 1.6416...
-
Hello Anton, No I did not, on the grounds that the structure is well known. I will try that. Thanks, Roger
-
elk.in attached
-
Hello, I've been trying for some time to calculate the volume-energy EoS of alpha quartz. I thank all those who have helped me in the past. I have obtained a smooth volume-energy curve and fitted it using elk's eos utility with the result given below: alpha-Quartz Universal EOS Vinet P et al., J. Phys.: Condens. Matter 1, p1941 (1989) (Default units are atomic: Hartree, Bohr etc.) V0 = 812.4590686 E0 = -1322.618525 B0 = 0.6214049845E-02 B0' = 4.773593001 B0 (GPa) = 182.8236277 The bulk modulus (182.8...
-
Hi Kay, Roger Mason rmason@mun.ca writes: I'm now attempting to compile on a cluster to which I have access but getting a compilation failure. Ah. Forgot to copy over mkl_dfti.f90. Compilation continues. Thanks again, Roger
-
Hi Kay, "J. K. Dewhurst" jkdewhurst@users.sourceforge.net writes: Hi Roger, I've managed to converge the calculation (except for one point) using the following input file: (one point failed to converge) You will always get slight kinks in the energy-volume data if you look closely enough. This is because the number of augmented planewaves changes discontinuously as the unit cell is scaled. These kinks can be converged away to an acceptable degree for most calculations. I tried running the first elk.in...
-
Hello Kay, I assume you are using the most recent elk version. I am on 7.1.14, in which 'batch' does not seem to exist. I ran some jobs with volumes around the jump in the earlier plot and there is still a break in the plot. I tried increasing rgkmax further, to 8.5, but that results in linearisation energy and eigenvalue warnings: Warning(linengy): could not find 3 linearisation energies in s.c. loop 1 Warning(occupy): minimum eigenvalue less than minimum linearisation energy : -11781.01391 -67.35210000...
-
Hi again Kay, "J. K. Dewhurst" jkdewhurst@users.sourceforge.net writes: Could you post your input file? Here is an updated plot.
-
Hello Kay, "J. K. Dewhurst" jkdewhurst@users.sourceforge.net writes: Thanks for your reply. Could you post your input file? There may be no need. I think the problem was an insufficient number of k-points. I'm currently running with ngridk set to '4 4 4'. I'll know if this is the fix within the next hour or so. Best wishes, Roger
-
Hello, I'm attempting to calculate the volume-energy EoS for alpha-quartz. I ran a series of calculations at different volumes, using default parameters for now, just to see what the volume-energy curve looks like. I'm attaching the plot, which shows a number of discontinuities. Does anyone have an idea what might be causing this? I'm running elk-7.1.14. Thanks, Roger
-
Replying to myself. "Roger Mason" rogermason@users.sourceforge.net writes: I have written some C++ code to handle the generation of elk input files, including those that create species files. ... I have resolved the problem, which was in my code, not in elk. Sorry for the noise. Roger
-
Hello, I have written some C++ code to handle the generation of elk input files, including those that create species files. The elk.in I generate looks like this: esccut -1.0 ecvcut -66 species 14 'Si' 'silicon' 28.0855 1.60 7 1 0 1 2 2 0 1 2 2 1 1 2 2 1 2 4 3 0 1 2 3 1 1 1 3 1 2 1 esccut -1.0 ecvcut -10 species 8 'O' 'oxygen' 15.9994 1.20 4 1 0 1 2 2 0 1 2 2 1 1 2 2 1 2 2 If I call elk from the command line it generates two species files. and stdout looks like: Elk code version 7.1.14 started Info(genspecies):...
-
Hello Kay, J. K. Dewhurst writes: I compiled Elk with GNU Fortran versions 10 and 11 on my Linux machine and (apart from the annoying argument mismatch problem) all is well and the tests pass. They pass for me too when run on a single machine. I sent a message to the users list about it on June 17th. The tests are only meant to be run on a single machine. Are you somehow running them across multiple nodes of a cluster? I assumed that test-mpi.sh in the tests subdirectory was designed to test MPI...
-
Hello Kay, J. K. Dewhurst writes: I'll fix it for the next release. This is the test output after making that change. Running test in directory test_017... Warning: Permanently added 'node004.cluster' (ECDSA) to the list of known hosts. Warning: Permanently added 'node001.cluster' (ECDSA) to the list of known hosts. Warning: Permanently added 'node002.cluster' (ECDSA) to the list of known hosts. Note: The following floating-point exceptions are signalling: IEEE_DIVIDE_BY_ZERO IEEE_UNDERFLOW_FLAG...
-
Hello, I'm still having problems getting elk-7.1.14 to pass all its tests when using MPI. I have the same problem with 7.2.42. I built a debug version of 7.1.14 and found that the crash happens here: Thread 1 received signal SIGFPE, Arithmetic exception. 0x00000000004abd30 in gndstate () at gndstate.f90:265 265 if (twrite.or.((nwrite.ge.1).and.(mod(iscl,nwrite).eq.0))) then This is the make.inc used to build it: MAKE = gmake F90 = mpif90 F90_OPTS = -g -ffast-math -funroll-loops -fopenmp -fallow-argument-mismatch...
-
Hello kay, J. K. Dewhurst writes: In the meantime you can use the option -fallow-argument-mismatch as a workaround and Elk should compile fine. Yes that fixed it, thanks. I thought there might be an otion for that but did not find it. Unfortunately I'm getting this runtime error: gmake test cd tests; ./test.sh Running test in directory test_001... PCI: Failed to initialize libpciaccess with pci_system_init(): 6 (Permission denied) I have libpciaccess-0.16 installed. Thanks, Roger
-
Hello, Building the most recent version of elk fails like this: cfftnd.f90:575:72: 575 | call cmf2kf ( lot, ido, l1, na, ch, 1, lot, c, jump, inc, wa(iw) ) | 1 Error: Type mismatch in argument 'ch' at (1); passed COMPLEX(8) to REAL(8) (lots of similar errors) when using gfortran10. These are merely warnings with gcc9. Cheers, Roger
-
Hello, I got the following test failure for version 7.1.14 (test 017). Elk code version 7.1.14 started Number of MPI processes : 4 Number of OpenMP threads per MPI process : 8 Total number of threads : 32 Maximum OpenMP nesting level : 4 Number of threads at first nesting level : 8 Number of MKL threads : 8 Info(elk): current task : 0 Info(checkmt): reduced muffin-tin radius of species 1 (Ga) from 2.4000 to 2.2888 Info(checkmt): reduced muffin-tin radius of species 2 (As) from 2.4000 to 2.2888 Info(elk):...
-
Hello Andrew, Andrew Shyichuk andrewshyichuk@users.sourceforge.net writes: Dear Roger, Task 2 is geometry optimization. To me, EOS must be calculated with task 0, i.e. without coordinate optimization, and only changing the scale. It makes sense to optimize the geometry once, but after that, I'd only change the scale. Moreover, doing latvopt=1 (i.e. optimize cell vectors) makes sense if and only if you have a basis that gives you smooth EOS. Without that, expect troubles: scatter will be way worse...
-
Hello, I'm attempting to calculate the volume-energy EoS for SiO2 in the alpha-quartz structure. The experimental value of Bo is around 37 GPa. The closest I have got is about 86 GPa. The latter value was a fit to a data set collected using task 2, and a set of scales from 0.96 to 1.04. There is quite a lot of scatter. There is much less scatter with task = 0 and only the scale varied (same range) from one calculation to the next. I'm using species files for Si and O that I set up following a thread...
-
Hello, I have been working on this on and off since May. I have made some, but not much, progress. The attached file shows how the scale parameter affects the calulated energy. The data points are colour coded according to the problems encountered. The problems are: linearisation energy not found, charge density incorrect, both problems or none. Michael, in his kind reply to my original question said: "To fix this it is either useful to increase the search radii demaxbnd or already increase the default...
-
Thank you Michael. I will need to look into this over the next few days. The material (alpha-quartz with earth surface cell parameters) is not very compressed, so perhaps something is incorrect in my input file. Thanks again, Roger
-
Hello, I am running models in the SiO2-TiO2 system. I am seeing this message in the console output: Info(checkmt): reduced muffin-tin radius of species 1 (O) from 1.8000 to 1.0959 Info(checkmt): reduced muffin-tin radius of species 2 (Si) from 2.2000 to 1.8864 Info(checkmt): reduced muffin-tin radius of species 3 (Ti) from 2.4000 to 1.7042 then: Warning(linengy): could not find 6 linearisation energies in s.c. loop 13 I assume from the fact that six linearisation energies cannot be found, and that...
-
Hello Kay, Indeed, that fixed the problem. For the benefit of others searching this in future, the file I modified is zfftifc.f90 Many thanks, Roger
-
Hello, I have fftw3-3.3.8. When I link elk-5.2.14 to fftw3 I get core dumps with 'gmake test'. Here is the make.inc I'm using: MAKE = gmake F90 = mpifort F90_OPTS = -O3 -ffast-math -funroll-loops -fopenmp F77 = mpifort F77_OPTS = -O3 -ffast-math -funroll-loops -fopenmp AR = ar LIB_SYS = LIB_LPK = lapack.a blas.a SRC_FFT = zfftifc_fftw.f90 LIB_FFT = libfftw3.a libfftw3_omp.a libfftw3_mpi.a # core dump SRC_OMP = The compiler is gcc8 with OpenMPI 2.1.5 on OS FreeBSD 11.2-RELEASE-p8 i386. Test output:...
-
Replying to my own post: I forgot to define swidth. Now all is well. Thanks, Rog...
-
Hello, I'm working on a solid solution between periclase (MgO) and wustite (FeO)....
-
Dear Michael, Thank you for taking a look at this. I can confirm convergence in 46...
-
Hello, I'm trying to calculate the energy of wustite (FeO) but cannot get the calculation...
-
Dear Michael, I allways comment 'SRC_MPI = mpi_stub.f90', er, except when I don't....
-
Hello Michael, INFO.OUT: http://filebin.ca/2haTNE9QMoMy/INFO.OUT stdout: http://filebin.ca/2haU1VGrMCbY/parallel.sh.o8372157...
-
Hello, I'm getting the following error: Warning(rhonorm): total charge density incorrect...
-
Dear Michael, Thank you for taking the time and trouble to diagnose the problem,...
-
Hello, I wish to use elk to explore the energetics of solid solution. As a starting...
1
