SSCHA for free energy minimization

[anyi-mxh]

I'm encountering an issue while using SSCHA for free energy minimization. As I understand, the minimization process should involve multiple iterations, where new configurations are re-sampled after each structural update until convergence is achieved. However, in my case, the calculation stops right after generating the first set of 2000 configurations, without proceeding to further structural updates and re-sampling steps.

At the end of the run, the following warning message appears:
/home/apps/anaconda3/envs/sscha/lib/python3.11/site-packages/sscha/SchaMinimizer.py:636: ComplexWarning: Casting complex values to real discards the imaginary part
self.dyn.structure.coords[:,:] = new_struct

Does this mean that complex coordinates have appeared in the structure, causing the program to terminate prematurely? However, there is no imaginary frequency in the dynamic matrix before optimization. Or could there be other reasons why SSCHA does not proceed with the full free energy minimization workflow?
How should this problem be solved?

Your kind reply is appreciated.

Best,
mxh

[mesonepigreco]

Hi,
The warning is standard and not an issue at all (it comes from a Fourier Transform).
What does the output say? Is it converged in one iteration? Or is it out of statistical relevance, and should a new ensemble be generated? The output should explain why the code stopped.