Currently, if the simulation fails, the trajectories are returned as zeroes. Initially, we'll check for this condition after the simulation runs and raise an exception to warn the user. Further down the line, it'd be nice to work out if the model will fit in GPU memory prior to attempted execution. That should probably be done by cupSODA itself rather than the interface.
Currently, if the simulation fails, the trajectories are returned as zeroes. Initially, we'll check for this condition after the simulation runs and raise an exception to warn the user. Further down the line, it'd be nice to work out if the model will fit in GPU memory prior to attempted execution. That should probably be done by cupSODA itself rather than the interface.