MPI-compiled version adds "PE0" to output file names

[chelsell]

I'm using APBS 1.5 on Ubuntu 18.04. In a variety of circumstances, APBS is adding the string "PE0" the end of filenames for outputs like dielectric maps. Here's an excerpt from an output file:

  X-shifted dielectric map to be written to dielx_1.dx
  Y-shifted dielectric map to be written to diely_1.dx
  Z-shifted dielectric map to be written to dielz_1.dx
  Kappa map to be written to kappa_1.dx
  Charge distribution to be written to charge_1.dx
  Skipping solve for mg-dummy run; zeroing solution array
  Total electrostatic energy = 0.000000000000E+00 kJ/mol
  Calculating forces...
  Writing x-shifted dielectric map to dielx_1-PE0.dx
  Writing y-shifted dielectric map to diely_1-PE0.dx
  Writing z-shifted dielectric map to dielz_1-PE0.dx
  Writing kappa map to kappa_1-PE0.dx
  Writing charge distribution to charge_1-PE0.dx

I don't appear to have any control over whether this string is added to the filenames--the input file certainly makes no mention of "-PE0"

This seems to be a known bug in older versions compiled with MPI: https://sourceforge.net/p/apbs/mailman/message/24456374/

I can confirm that recompiling APBS without MPI solves this problem, but it seems like a silly reason not to use MPI. Also it seems like compilation with MPI is the default if it can be detected--I could only get this bug to go away by specifying DENABLE_MPI=OFF

[sobolevnrm]

Hi Cole -

This is the expected behavior. Is it causing problems with other software you're using in conjunction with APBS?

Thanks,

Nathan

On Mon, Dec 9, 2019 at 1:27 PM Cole Helsell notifications@github.com wrote:

I'm using APBS 1.5 on Ubuntu 18.04. In a variety of circumstances, APBS is adding the string "PE0" the end of filenames for outputs like dielectric maps. Here's an excerpt from an output file:

X-shifted dielectric map to be written to dielx_1.dx Y-shifted dielectric map to be written to diely_1.dx Z-shifted dielectric map to be written to dielz_1.dx Kappa map to be written to kappa_1.dx Charge distribution to be written to charge_1.dx Skipping solve for mg-dummy run; zeroing solution array Total electrostatic energy = 0.000000000000E+00 kJ/mol Calculating forces... Writing x-shifted dielectric map to dielx_1-PE0.dx Writing y-shifted dielectric map to diely_1-PE0.dx Writing z-shifted dielectric map to dielz_1-PE0.dx Writing kappa map to kappa_1-PE0.dx Writing charge distribution to charge_1-PE0.dx

I don't appear to have any control over whether this string is added to the filenames--the input file certainly makes no mention of "-PE0"

This seems to be a known bug in older versions compiled with MPI: https://sourceforge.net/p/apbs/mailman/message/24456374/

I can confirm that recompiling APBS without MPI solves this problem, but it seems like a silly reason not to use MPI.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Electrostatics/apbs-pdb2pqr/issues/533?email_source=notifications&email_token=AAOX7WE2Y5MTXRHDRA76EQ3QX2Z3VA5CNFSM4JYSTAB2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4H7IYSLQ, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAOX7WHZ5XKUOLCUAHSDRL3QX2Z3VANCNFSM4JYSTABQ .

[chelsell]

Hi Nathan,

Yeah, I was having trouble with this while running software that depends on APBS. I can definitely patch that software to not be so literal about the names of APBS outputs, but I was worried that other things might be systematically different between APBS with MPI and APBS without. Is there documentation that explains this difference and any others?

[sobolevnrm]

Hello -

There are some differences in output formatting and naming of the files but these should be the only differences you encounter.

Thank you,

Nathan

On Mon, Dec 16, 2019 at 2:41 PM Cole Helsell notifications@github.com wrote:

Hi Nathan,

Yeah, I was having trouble with this while running software that depends on APBS. I can definitely patch that software to not be so literal about the names of APBS outputs, but I was worried that other things might be systematically different between APBS with MPI and APBS without. Is there documentation that explains this difference and any others?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/Electrostatics/apbs-pdb2pqr/issues/533?email_source=notifications&email_token=AAOX7WCJZGKDYLCDJWQTXQTQY772RA5CNFSM4JYSTAB2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHAMADQ#issuecomment-566280206, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAOX7WEN2JKCY7DQSMITXJLQY772RANCNFSM4JYSTABQ .

[sobolevnrm]

I'm going to close this issue for now. If there are specific software packages affected by this behavior, please let us know and we'll address those separately.