Interface with MRCC

1

Hello,
I wanted to test the interface with MRCC. I attempted to run the example from the Psi4 manual (Interface to MRCC by M. Kállay), but I got an error:

‘str’ object is not callable; perhaps you missed a comma?
‘str’ object is not callable; perhaps you missed a comma?
‘str’ object is not callable; perhaps you missed a comma?
‘str’ object is not callable; perhaps you missed a comma?
“is” with a literal. Did you mean “==”?
Traceback (most recent call last):
File “/usr/local/psi4/bin/psi4”, line 331, in
exec(content)
File “”, line 25, in
File “/usr/local/psi4/lib/psi4/driver/driver.py”, line 570, in energy
wfn = procedures[‘energy’][lowername](lowername, molecule=molecule, **kwargs)
File “/usr/local/psi4/lib/psi4/driver/procrouting/proc.py”, line 4316, in run_mrcc
for line in open(‘iface’):
FileNotFoundError: [Errno 2] No such file or directory: ‘iface’
Printing out the relevant lines from the Psithon → Python processed input file:
H 1 1.0
H 1 1.0 2 104.5
“”",“h2o”)
core.IO.set_default_namespace(“h2o”)
core.set_global_option(“BASIS”, “cc-pVDZ”)
→ energy(‘mrccsdt’)

I also tried with the Python API , I got the same, albeit more detailed, error:

Error.txt (11.0 KB)

I have MRCC installed and working fine as a stand alone program. Any help would be greatly appreciated.

2

Bug reports always need to include a Psi4 version number.

Am I correct in assuming you’re using the new version of MRCC? I’m noticing the line Release date: February 22, 2020 in the MRCC section of your output file. MRCC may have broken the interface between MRCC and Psi in the newest version.

3

Dear jmisiewicz,
thanks for the answer, and I am sorry for the oversight. I am running Psi4: 1.4a2.dev439, and Git: Rev {master} 940eeec.

I noticed that Psi4 creates a folder in the scratch directory, where it writes fort.56 (input data for MRCC), fort.55 (integrals to pass to MRCC), MINP (input file for MRCC).
Upon inspection of these files, I found fort.56 includes information about MRCC keywords. It lacks geometry input, for example, although I am not sure if it is supposed to be here, whereas fort.55 has integral info inside (I cannot say if it is complete or not, though). MINP is empty.

When MRCC runs, it normally produces a file called ‘iface’. In this example, Psi4 calls MRCC to run on an empty MINP file, and hence produce no ‘iface’. Later, when Psi4 wants to retrieve output from MRCC, it generates a

FileNotFoundError: [Errno 2] No such file or directory: ‘iface’

and stops. Now it seems to me, at least for this example, that the problem is on Psi4 side because it generates an empty MINP file.

4

The Psi4 interface doesn’t run through the MINP file.

Please answer my question. Am I correct in assuming you’re using the new Feb. 22, 2020 version of MRCC? I want to be sure of that before I put more time into investigating this further.

5

Yes, you are correct

6

I’ve turned this into an issue on our GitHub page. I am fairly certain that the new MRCC is just refusing to allow the way that Psi used to send its information over. When I use old MRCC, it doesn’t even look for MINP. When you use the new MRCC, it does.

Unfortunately, I can’t be of further help here.

7

Copying a message from Roberto on the issue page:

Ran into this issue myself a week ago. It turns out that you need to patch a few file in the MRCC tarball and recompile to make it work (and this has to be done independently of the Psi4 interface) I couldn’t see this info advertised on the MRCC website, I happened to browse the “Patches” section in their download page and noticed that there were few patches more recent than the tarball I had just downloaded.

It’s an issue with your MRCC, and you’ll need to patch it. Getting help for the patching phase is a job for the MRCC forums.