hi Steve and John,
I confess I don't have a definitive answer. I've seen NVE and NVT in the literature. With NVE you don't have the influence of the thermostat which potentially has artificial effects on the flux, and with NVT your averages are canonical. I think the "right" way is to generate an ensemble of ics using NVT and then use these in NVE flux calculations but this is obviously more expensive.
I have no objections to changing the sample script.
Reese