I have been using two different versions of ESPResSo (2.1.0 and 2.1.2j) lately. Somehow I ran a script using both versions and got different energy/configuration profiles.
To test a simpler case I ran lj_liquid.tcl with both versions. When the thermostat is set to langevin, I get same results from both runs. However, when I replace the langevin thermostat with dpd
"thermostat dpd 1.0 1.0 1.12246"
results seem to vary. I did the Galilei transform to stop the center of mass motion and just checked the release notes but didn't see any reported bugs on this.