Alex Izvorski

@godotalgorithm What type of primitives are needed for GPU MOZYME, is it mostly sparse matrix multiply? How sparse and how large is it? Can this be done with either PyTorch...

@godotalgorithm Thanks for explaining, that's very interesting. Givens rotations - basically they're matrix-matrix multiplications with a matrix that just has two off-diagonal elements? (and then truncation). I had a look...

@godotalgorithm Thanks Jonathan! That's amazing work. I read it through and the code is very readable. I'll need a little bit to understand the sequence of operations in carries out...

Update: the exact number of atoms which causes the crash is 834. The number of orbitals doesn't seem to matter, it really is atoms. Works: 833 helium atoms https://gist.github.com/aizvorski/a6616970339d8447a98989b4d0455db8#file-helium833-xyz Crashes:...

@haneug I can confirm this, the process stack in ulimit -s was the limiting factor. `ulimit -s unlimited` works. I think it's fair to say any SIGSEGV crash is a...

@awvwgk Thanks, that's a good collection of links! I don't know too much about Fortran specifically, but perhaps using some compiler feature to avoid large arrays on the stack (without...

@awvwgk About OMP_STACKSIZE: the compiler options to reduce stack use may also apply to OpenMP code, but if not, maybe we could default to OMP_STACKSIZE=physical memory/number of threads? That's only...

@RiesBen @peastman I just wanted to say this is an incredibly useful PR. Can we merge this in before the next release?

@peastman I made a pdb with every aminoacid variant, sequence is GLY-ASH-ASP-CYS-CYX-GLH-GLU-HID-HIE-HIP-HIN-LYN-LYS. Also, to check different caps, GLY-GLY and ACE-GLY-GLY-NME. These are made in pymol. Unfortunately these don't load into...

@peastman Once we do get variants.pdb loading, how can we check the bond orders and types? We could inspect those manually and set up a regression test. Alternatively, we could...