I am doing an efficiency analylis on the different minimization options in Rosetta using FastRelax, MPRelax and lbfgs_armijo_nonmonotone minimizer on a membrane protein using the franklin2019 score function and I very very often get this error:
core.optimization.LineMinimizer: [ ERROR ] Inaccurate G!
It is clear to me that they do not perform as well when the error occurs but how worried should I be about it? I can get FastRelax and armijo to work without giving the error but MPRelax always has a bunch of those. I have read that it only worsen the accuracy but it is still useful, is that correct?