Apply fixes for the CPC submission

[cffischer]

Correction of errors found after submission to CPC:

1. in RCI90_mpi : The rci90mpi.f90 routine opened the .clog file again after it had already been opened.
   The second open and associated write commands were commented out.

2. RMCDHF90_mpi: there were two corrections:
   i) getscd.f90 and getscd_I.f90 were deleted since they were not used.
   ii) The statement
   RNT = 2.D-6
   was replaced by
   !CFF ... RNT should be Z-dependent
   RNT = 2.0D-06/Z
   With this change, the grid for the hydrogen-like equation is a transformation of the hydrogen grid.
   With these changes, this version is equivalent to the CPC version of grasp2018.

[cffischer]

This repository is now equivalent to the latest CPC submission of the Grasp2018 and is ready for distribution as a public repository.

[mortenpi]

2. ii) The statement
RNT = 2.D-6
was replaced by
!CFF ... RNT should be Z-dependent
RNT = 2.0D-06/Z
With this change, the grid for the hydrogen-like equation is a transformation of the hydrogen grid.

I don't see this change. The PR currently only touches src/appl/rci90_mpi/rci90mpi.f90 and src/appl/rmcdhf90_mpi/getscd.f90

With these changes, this version is equivalent to the CPC version of grasp2018.

So the code has already been submitted to CPC? In any case, I will then make the repo public as soon as this is merged.

[cffischer]

The paper has been accepted but we need page numbers for the Manual. Manual +Program will be submitted to Belfast (not the publisher) once we have the page numbers. . Gediminas has been in touch with Intel about a problem and may have another small correction that needs to be made. The GitHub Program should not be released until I have modified it correctly. Best wait until I give you the go ahead. We could always delete what we have and reinstall the repository that we send to Belfast but I would like to try and learn how to do updates and this is a good test case. I will contact you in a few hours. Charlotte From: Morten Piibeleht [mailto:notifications@github.com] Sent: Sunday, October 14, 2018 5:43 PM To: compas/grasp2018 Cc: Charlotte Fischer; Author Subject: Re: [compas/grasp2018] cff/fix-error1 (#1) 2. ii) The statement RNT = 2.D-6 was replaced by !CFF ... RNT should be Z-dependent RNT = 2.0D-06/Z With this change, the grid for the hydrogen-like equation is a transformation of the hydrogen grid. I don't see this change. The PR currently only touches src/appl/rci90_mpi/rci90mpi.f90 and src/appl/rmcdhf90_mpi/getscd.f90 With these changes, this version is equivalent to the CPC version of grasp2018. So the code has already been submitted to CPC? In any case, I will then make the repo public as soon as this is merged. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1 (comment)> , or mute the thread <https://github.com/notifications/unsubscribe-auth/AMatyH1neOOVNw5LPBH9d_-MaRevvDlvks5uk9oogaJpZM4XbY2A> .Image removed by sender.

[mortenpi]

Is the removal of these comments intentional? I suppose the NELMNT part is no longer needed, but the other ones are useful.

[cffischer]

#1 I could not find those comments in the original tarball. The comments pertain to an old version of rci90 where the user specified how the eigenvalue is determined. Now this is done according to memory available and memory needed. It was not evident on my local machine. So why would it appear on a merge?

[mortenpi]

Here are the steps to clone this PR to your computer. As the branch being PRd lives on a fork, it is slightly non-trivial:

# Clone the CompAS repository, if you already haven't
git clone https://github.com/compas/grasp2018.git
# do this instead if you have SSH keys set up:
#git clone git@github.com:compas/grasp2018.git

# cd into your local repository
cd grasp2018

# Add the fork as a separate Git remote
git remote add cffischer https://github.com/cffischer/grasp2018.git
# do this instead if you have SSH keys set up:
#git remote add cffischer git@github.com:cffischer/grasp2018.git 

# Fetch the contents of the forked repository
git fetch cffischer

# Checkout the PR branch
git checkout cff/fix-error1

# After this your worktree (i.e. the files on the disk) should correspond
# to the code in the pull request. Check this by running
git status
git log
# git status should say "On branch cff/fix-error1" and the log should show
# the relevant commits.

[mortenpi]

The rmcdf_mpi and rci_mpi programs appear to compile and run. I did not check the correctness of the outputs though.

[cffischer]

Intel is extremely particular about adhering to a standard. Gediminas said rci90 and rci90_mpi both had “bugs”. Then he sent me a maneig.f90 for rci90_mpi and somehow I came to the conclusion that maneig.f90’s were the same. Actually, I believe he sent me the corrected version for rci90 because the parallel version needs to have the number of processor (his version had nprocs=1) and needs to limit the output so that the same thing is not printed 100 times when running a large job on 100 processors. So I need to sort this out with Gediminas.

There are lots of irrelevant comments in GRASP. Gediminas (and others) have the habit of commenting out lines but never removing any. So after a while the code is not very readable. In this case a procedure had changed but the comments remain.

Sorry. It would have been nice to get this finished. Besides, I forget things after a while.

[cffischer]

tool/lscomp.pl should now be executable

[cffischer]

Changing the permission of lscomp.pl to executable

[cffischer]

I think we have code that is "perfect" for CPC, although more confirmations would be helpful.