FCI does not work in some cases

Dear Mihály,

I've tried to increase available memory but without success.
So, I'm trying to localize the source of the problem:

I've performed the following additional tests:
If I consider B-, i.e. add one more electron to the calculation then CISDTQP calculation starts successfully.
Moreover, I got positive results even for the CISDTQP calculation of B-- (7 electrons instead of 5).
So we have the following table for CISDTQP calculations:

5 electrons, configurations=190578024, optimal mem=???, crashes
6 electrons, configurations=985819352, optimal mem=72290, success
7 electrons, configurations=1605135960, optimal mem=89508, success

I'll proceed to search for further circumstances which leads to the crash.

Best,
Leonid.

Dear Leonid,
I have found a machine with much memory and run your calculation. I have not experienced any problem, see attached output file. So your problem might be related to the compiler, and you should try another version (I am using ifort 18.0.0.128).
Independently of this problem, I have found a bug which might cause problems for relativistic CC/CI calculations requiring much memory. You will find three files, mrcc.f, xmrcc.f, and goldstone.f, in the download area. It is recommended to use them.

Dear Mihály,
Thank you very much!
I'll try to use another compiler and updated source files.