nstexpanded was checking conditionally on ir<s><span style="text-align:right;">efep instead of ir</span></s>&gt;bExpanded. This might have led to incorrect energies being used for expanded ensemble sampling with free-energy=on and nstexpanded not a multiple of nstcalcenergy. I have not checked if results would actually be wrong in this case. *(from redmine: issue id 2714, created on 2018-10-28 by berkhess, closed on 2018-10-31)* * Changesets: * Revision e1e38bc34b826ed3b3cd9f9e06aa2b2eac168af0 by Berk Hess on 2018-10-31T06:25:55Z: ``` Correct grompp nstexpanded/nstcalcenergy check With expanded ensemble, but without free-energy perturbation, grompp would not check if nstexpanded was a multiple of nstcalcenergy. If the latter was not the case, results might have been incorrect. Also added a check in mdrun to catch old, incorrect tpr files. Fixes #2714 Change-Id: I554351e2c9873a71efbc6e803134eb744e229f1d ``` * Revision ef48ba374b1b1f9f4741f6a0a7502ede65670f86 by Berk Hess on 2018-10-31T07:20:44Z: ``` Correct VV integrator nstcalcenergy use With velocity Verlet integrators, mdrun would compute energies which contribute to averages in the energy output and might be used for expanded ensemble calculcations one step too late. Fixes #2718 Refs #2714 Change-Id: I67e4d00f7151ec9eb5dab6ca4e87b81ca12236e2 ``` * Revision 626e3818a325f07add061fed5287ced77697a58a by Berk Hess on 2018-10-31T12:33:40Z: ``` Work around expanded ensemble issues Two bugs could cause expanded ensemble sampling to use outdated or zero energies. In these cases mdrun now modifies nstcalcenergy to 1 to avoid these bugs. Note: This change should not be merged upstream, since there is a proper fix for release-2019. Refs #2714 Refs #2718 Change-Id: I79be9c5da55eaebb857bac6a98e6671720532e0e ```