Re: [Wien] Parallel execution of SCF cycle

Laurence, in response to your query: "echo $WIEN_MPIRUN" came up blank, but I have added the necessary line to parallel_options, and as I say it That's ok. WIEN_MPIRUN is not known by the regular shell (would have to be added in .bashrc), but the *para_lapw scripts read it directly from

Re: [Wien] Parallel execution of SCF cycle

Laurence and Peter, Thanks very much for your help with this. It now seems to be working OK - I've now ran several SCF cycles to completion on 16 cores, which is very promising. I will need to do some more checks to ensure that the results are correct and all of lapw0/lapw1/lapw2 are behaving

Re: [Wien] Parallel execution of SCF cycle

Check by doing "echo $WIEN_MPIRUN". If this comes up blank or something different from what we said then it MUST be added to parallel_options. If it is not there then Wien2k was not installed correctly, from the sounds of it by someone else. If that is the case then there may be many other things

Re: [Wien] Parallel execution of SCF cycle

I have added the "setenv WIEN_MPIRUN ..." line to $WIENROOT/parallel_options. Is this the correct thing to do? Note I have not yet been able to test Peter's new lapw1para_lapw script because I am waiting for the cluster administrators to make the change (I do not have the right permissions).

Re: [Wien] Parallel execution of SCF cycle

> > In addition to what Peter said in his last post, does your > parallel_options now have (as he said in an earlier post): > setenv WIEN_MPIRUN "mpirun -np _NP_ -machinefile _HOSTS_ _EXEC_" > Or, did you define this somewhere else in your installation such as > .bashrc (using export)? > >

Re: [Wien] run_lapw

It is enough to replace the script. This is a shell-script, no fortran program. Thus no recompilation. Am 01.02.2023 um 14:00 schrieb pluto via Wien: Dear Prof. Blaha, I understand it is enough to simply replace the existing run_lapw file? Or does one need to recompile everything? Best,

Re: [Wien] run_lapw

Dear Prof. Blaha, I understand it is enough to simply replace the existing run_lapw file? Or does one need to recompile everything? Best, Lukasz On 2023-02-01 12:43, Peter Blaha wrote: Only 23. Am 01.02.2023 um 12:09 schrieb Laurence Marks: Which version(s) of Wien2k does this effect?

Re: [Wien] run_lapw

Only 23. Am 01.02.2023 um 12:09 schrieb Laurence Marks: Which version(s) of Wien2k does this effect? -- Professor Laurence Marks Department of Materials Science and Engineering, Northwestern University www.numis.northwestern.edu "Research is to see what

Re: [Wien] run_lapw

Which version(s) of Wien2k does this effect? -- Professor Laurence Marks Department of Materials Science and Engineering, Northwestern University www.numis.northwestern.edu "Research is to see what everybody else has seen, and to think what nobody else has thought" Albert Szent-Györgyi On Wed,

Re: [Wien] Parallel execution of SCF cycle

The problem is your .processes file, which is created in lapw1para_lapw (from your .machines file). it should look eg. like init:susi init:susi 1 : susi : 38 : 1 : 1 : 0 2 : susi : 38 : 1 : 2 : 1 So you are missing the hostnames in it. This has to do with the "." in your hostnames I

Re: [Wien] Parallel execution of SCF cycle

Thank you Peter - do you have any advice on how we can fix the parallel_options issue? I have attempted manually editing the file with your example below but this has resulted in more errors (probably as expected?). Is it likely that this is an installation/configuration issue?

[Wien] run_lapw

Dear wien2k users, There is a small bug in run_lapw for cases without inversion symmetry. Most likely not critical except when the default case.in1c file would produce a QTL-B error, which would not be fixed automatically. A new version is attached and the web-version has been corrected