Installing VTST with VASP_6.4.3.

[graeme]

It should be good now.

[hajgato]

Maybe I am doing something wrong, then. I tried version 203 and from the svn (also version 203), and it still fails in the same way (gfortran, ifort)
Is there a newer version?

[graeme]

Thanks for reporting. Can you try some basic things and do a make clean, then make sure that you are getting the code from the vtstcode6.4.3 directory, and then if it still fails, report the error.

[hajgato]

[vsc43020@node3509 vasp.6.4.3]$ gcc --version
gcc (GCC) 13.2.0
Copyright (C) 2023 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

[vsc43020@node3509 vasp.6.4.3]$ mpirun --version
mpirun (Open MPI) 4.1.6

Report bugs to http://www.open-mpi.org/community/help/
[vsc43020@node3509 vasp.6.4.3]$ sha256sum ../vtstcode-203.tgz
471346810d5d7bf45834a26802aca69db38a52e37b15b9c1b8719dfaf4c314b3 ../vtstcode-203.tgz
[vsc43020@node3509 vasp.6.4.3]$ cat vtst.install
cp -v src/main.F src/main.F_vtst_backup
sed -z 's/ LATT_CUR%A,LATT_CUR%B,IO%IU6)/ TSIF,LATT_CUR%A,LATT_CUR%B,IO%IU6)/' -i src/main.F
sed -e 's/IF (LCHAIN) CALL chain_init( T_INFO, IO)/CALL chain_init( T_INFO, IO)/' -i src/main.F
mv -v src/chain.F src/chain.F_vtst_backup
cp -v ../vtstcode-*/vtstcode6.4.3/*.F src/.
cp -vr ../vtstcode-*/vtstcode6.4.3/pyamff_fortran/ src/.
sed -e 's# chain.o# bfgs.o dynmat.o instanton.o lbfgs.o sd.o cg.o dimer.o bbm.o fire.o lanczos.o neb.o qm.o ml_pyamff.o pyamff_fortran/fptype.o pyamff_fortran/atoms.o pyamff_fortran/neighborlist.o pyamff_fortran/behlerParrinello.o pyamff_fortran/nntype.o pyamff_fortran/trainType.o pyamff_fortran/normalization.o pyamff_fortran/mtfort90.o pyamff_fortran/fingerprints.o pyamff_fortran/neuralnetwork.o pyamff_fortran/nnmodule.o pyamff_fortran/pyamffType.o pyamff_fortran/lossgrad.o pyamff_fortran/adam.o pyamff_fortran/lbfgs_ml.o pyamff_fortran/rprop.o pyamff_fortran/opts.o pyamff_fortran/training.o pyamff_fortran/model.o opt.o chain.o#' -i src/.objects
sed -e 's/LIB=lib parser/LIB=lib parser pyamff_fortran/' -i src/makefile
[vsc43020@node3509 vasp.6.4.3]$ cat build/std/chain.F | grep -i chain.stress
SUBROUTINE chain_stress( stress )
END SUBROUTINE chain_stress
[vsc43020@node3509 vasp.6.4.3]$ nm build/std/chain.o | grep -i stress
0000000000000068 B __chain_MOD_pstress_local
[vsc43020@node3509 vasp.6.4.3]$ tail -9 make_std.out
/apps/gent/RHEL8/zen2-ib/software/binutils/2.40-GCCcore-13.2.0/bin/ld: main.o: in function `MAIN__':
main.f90:(.text+0x23cfa): undefined reference to `chain_stress_'
collect2: error: ld returned 1 exit status
make[2]: *** [makefile:134: vasp] Error 1
make[2]: Leaving directory '/tmp/tst/vasp.6.4.3/build/std'
cp: cannot stat 'vasp': No such file or directory
make[1]: *** [makefile:131: all] Error 1
make[1]: Leaving directory '/tmp/tst/vasp.6.4.3/build/std'
make: *** [makefile:17: std] Error 2

[graeme]

Ah, this is crazy making - your script looks perfect.

Ok, I hate to suggest this again: you did a 'make veryclean' before the build?

Just to double check, do: "rm -rf build'

and "grep -i chain_stress" src/chain.F

and make sure that you see:

SUBROUTINE chain_stress( stress )
END SUBROUTINE chain_stress

and double-check your .objects file and that your sed command added the vtst objects properly

[hajgato]

You mention a vtst v2.04 (https://theory.cm.utexas.edu/forum/view ... 676#p26676) But what I use is VTST 203 https://theory.cm.utexas.edu/code/vtstcode-203.tgz Am I using the latest version?

I built VASP from scratch (I prefer not to make clean, and co, I prefer freshly untarred source codes). Without VTST it just builds and functional, passes all the tests (fast)
I have successfully build various versions of VASP/VTST with the same script (just changing the "cp -v ../vtstcode-*/vtstcode6.4.3/*.F src/.
&& cp -vr ../vtstcode-*/vtstcode6.4.3/pyamff_fortran/ src/." part to the right vtstcdoe.X.X)

[vsc43020@node3509 vasp.6.4.3]$ cat build/std/chain.F | grep -i chain.stress
SUBROUTINE chain_stress( stress )
END SUBROUTINE chain_stress
so the chain.F is the right version.

But, chain_stress is is not in the object file:
[vsc43020@node3509 vasp.6.4.3]$ nm build/std/chain.o | grep -i stress
0000000000000068 B __chain_MOD_pstress_local

meanwhile, fore example, chain force is in the object code:
[vsc43020@node3509 vasp.6.4.3]$ nm build/std/chain.o | grep -i force
0000000000003d20 T __chain_MOD_chain_force
00000000000001b8 B __chain_MOD_force_tol
0000000000000220 b force_all.0

[vsc43020@node3509 vasp.6.4.3]$ grep -C2 chain.o build/std/.objects
elf.o \
hamil_rot.o \
bfgs.o dynmat.o instanton.o lbfgs.o sd.o cg.o dimer.o bbm.o fire.o lanczos.o neb.o qm.o ml_pyamff.o pyamff_fortran/fptype.o pyamff_fortran/atoms.o pyamff_fortran/neighborlist.o pyamff_fortran/behlerParrinello.o pyamff_fortran/nntype.o pyamff_fortran/trainType.o pyamff_fortran/normalization.o pyamff_fortran/mtfort90.o pyamff_fortran/fingerprints.o pyamff_fortran/neuralnetwork.o pyamff_fortran/nnmodule.o pyamff_fortran/pyamffType.o pyamff_fortran/lossgrad.o pyamff_fortran/adam.o pyamff_fortran/lbfgs_ml.o pyamff_fortran/rprop.o pyamff_fortran/opts.o pyamff_fortran/training.o pyamff_fortran/model.o opt.o chain.o \
dyna.o \
fileio.o \

so the object file should be fine. (unless the order is changed between vasp.6.4.2 and vasp.6.4.3)

I have tried also to use the bfd linker instead of gold (-fuse-ld=bfd), and tried also -fno-dce (no dead code elimination), but none of them helped.

[graeme]

do you need continuation \ marks after the object file lines?

This is what I have:

VTST_OBJS = bfgs.o dynmat.o instanton.o lbfgs.o sd.o cg.o dimer.o bbm.o \
fire.o lanczos.o neb.o qm.o \
pyamff_fortran/*.o ml_pyamff.o \
opt.o

...
hamil.o \
$(VTST_OBJS) \
constrmag.o \
...

[hajgato]

of course in the object file is a one-liner but the forum uses automatic line-breaks, so you see as they would different lines.

relocating the position in the object file just changes the order of compilation (i compile with make -j1 to avoid problems).

Anyway, I used exactly the same object file what you did, nothing changed, the same error.

[graeme]

If you can send me your chain.F and .objects file, I can take a look. Maybe also main.F.

[hajgato]

sent a private message with the title forum/viewtopic.php?p=26691#p26691

[graeme]

Ok, I am a complete idiot and I apologize for wasting your time here. This is all to do with a failure in my code update.

Please check out version 204, which should have a chain.F with a version: VTST: version 4.2, (11/09/24)

Hopefully this all works now. I appreciate your help to fix this.

[hajgato]

I can confirm that it works with vtst 204.
At the end the important thing that the compilation succeeded.
Thanks for your effort!