[AMBER-Developers] Zeroing forces of specific atoms

From: Parker de Waal <Parker.deWaal.vai.org>
Date: Tue, 31 Oct 2017 16:20:46 +0000

Hello everyone,

I'm currently trying to write a kernel that can modify the forces of specific atoms within GPU and am running into issues where the simulation

become numerically unstable right after the kernel is called (atom crds/frcs explode). I suspect this error stems from either a formatting problem or I'm not editing the correct locations within memory.

For now, my kernel can be found here https://gist.github.com/ParkerdeWaal/113b9c7460f21b111e4c83d5277cc856

Hopefully someone here can help me answer this (trivial?) problem.



AMBER-Developers mailing list
Received on Tue Oct 31 2017 - 09:30:04 PDT
Custom Search