I am running simulations using the fix nve time integrations and either maintaining two constant temperature baths or fix energy added/subtracted at ends of the system with fixed position walls outside of them (nemd). I am attempting to force a temperature gradient to calculate thermal conductivity, but I am unable to obtain a temperature gradient and the system will eventially melt (initially solid). I am using Argon with LJ units to keep it as simple as possible. Should I try a different integration scheme? I have also tried the fix thermal/conductivity with a completely periodic system, also with no luck. If it helps also I compiled with the serial makefile.
Thank you for any insight,