I have installed PSI4 v1.7 in my Linux machine. When I give my psi4 command a message is given A NumPy version >=1.16.5 and <1.23.0 is required for this version of SciPy (detected verson 1.23.4)
So if I downgrade NumPy to 1.16.5 using conda, I guess the issue will be sorted. Am I right? But when downgrading NumPy, will it be a problem for other packages that depend on NumPy?
And currently I am facing a problem in creating my Psi4 output file. Though psi4 command generates an output file, the file size remains with no change for many hours after a certain kB file size. The program does not show any error or it does not terminate and shown as a running program in the terminal but output file size remain same. There seems to be no issue with memory so is this a normal thing in Psi4 calculations or can the above mentioned compatibility issue will be a reason for this?
As long as conda is able to solve the environment, it should be fine. If conda runs into trouble solving the environment, installing to a fresh conda environment will make sure you get the correct versions of dependencies.
As long as the process is still running, then likely everything is OK. For very large computations, there are portions that could take hours without writing to the output file. So, I wouldn’t consider this an issue unless a computation you expected to take 10 minutes is suddenly hanging for a few hours.
Thank you very much.
Then the reason for taking more time could be the optimising of a large molecule. So, to reduce computational cost can we do the optimisation focusing only on the region of our interest? And in that process can we neglect the continuity of bonds when selecting the residues of our interest for the optimisation? For example if I want to do the optimisation for the active site, then can I select only the active site residues which are distant from each other?