Memory taken importing psi4 module (v1.4)

Continuing the discussion from Memory usage using jupyter (writing fchk):

Dear all,

Related to the above topic, I am dealing with some memory issues when running psi4 from jupyter. Since I am interested in using TDDFT, I am trying 1.4 version and found the module takes much more memory as compared with previous versions. Concretely, while with version 1.3.2, it only takes <100MB when importing psi4, it goes up to 600MB with version 1.4rc2.dev99. I was wondering if this is something to be expected for the stable 1.4 release or just something related to the pre-releases versions. If something inherent to 1.4 version, is there any way to deal with this in jupyter? As indicated in the previous topic, maybe using multithreading can help also here.




Hi Javier,

Thanks for reporting this. We’ll certainly take a look into this and see if there is anything that we can do t o reduce it.


I’m using 1.4 release (1.4+9485035) from conda and memory increases from 120MB to 190 MB after minimal usage (importing psi4, calculating energy, and writing out orbitals of Neon).

There is a discrepancy between the memory usage that Jupyter reports (326 MB) vs what docker reports the container is using (190 MB), but I don’t know enough about what they are counting to understand that difference.