Josh Fass

Ahh, thank you for clarifying! Makes sense that it would be undefined instead of zero here. Since this situation can easily arise in cross-validation, some additional guidance for model selection...

Hi! I'm having trouble producing a minimal example to reproduce the problem, but this error arises when doing `MiniBatchKMedoids` clustering using this script, with `n_clusters`=`batch_size`=`10000`: https://github.com/maxentile/MSMs/blob/master/results/scripts/ha_rmsd.py Error message: `33221 Segmentation...

Hi Matthew! > What version of msmbuilder? Good call! This was using an earlier version: 3.3.1. Re-running this on our cluster using the current version of MSMBuilder now. Will report...

Thanks!

>It's probably a good idea to make sure the total potential energy matches the unmodified system pretty closely! This test is currently failing, but we're aiming to match within the...

Thanks so much!

I noticed the body of `GHMCRESPAIntegrator` is commented out: is that waiting on something else or can it just be un-commented and tested?

Reviewed the `GHMCIntegrator` and `XCGHMCIntegrator`, and they both look great to me! When we decide whether to include the RESPA integrators, I'll take a closer look at those. My only...

>Removing those properties SGTM. I agree they are incomplete characterizations that lack the autocorrelation contribution, although I wonder if they could still be useful for rough, back-of-the-envelope performance tuning. In...

Hi @ljmartin, The `CustomIntegrator`s implemented here are somewhat slower than the default `LangevinIntegrator`. We haven't yet determined why. We typically assume that the computational cost of integration is mostly determined...