Hi, I am wondering if there is a way using Gromacs tools to extract the cumulative heat exchanged between the simulation system and thermal bath during a constant temperature simulation starting from a nonequilibrium initial state?
For example, in a SMD simulation (steered MD) on pulling out an ligand from its binding pocket on a protein, I am interested to compute the heat supplied to released to the thermostat. But I couldn’t find anything related in the manual.
Thanks for your help!
Nanjie
When performing steered MD you should get the work from the integral over the applied force, not through the change in energy of the heat bath. The latter can contain all kind of other effects and artifacts and is affected by the whole system.
Note that you would need to average the obtained work over many replicates and use Jarzynski’s equality to obtain a (binding) free-energy. This work is usually a (large) overestimate due to frictional losses.
Thank you for getting back to me. Here we are not trying to compute the binding free energy using SMD; we are interested in studying the heat exchange between the simulation box with the environment (reservoir). In SMD, the work done by the pulling force causes both an energy change in the system and heat transfer with the thermostat, it is the latter quantity we would like to obtain. I am guessing that such real-time momentum transfer information could be hidden in the thermostat code?
Best,
Nanjie
If I understand you correctly you are actually after the enthalpic and entropic components of the work or free-energy. The best way to get the enthalpic contribution is to compute the (average) potential energy of the system in the states you want the difference between. But note that that requires orders of magnitude more sampling than that required for obtaining the free-energy, as this depends on the whole system.
If there were no integration errors, the change in heat bath energy would give you the entropic contribution. But the integration errors will swamp the change due to the work.
Thank you for your insights. Yes that’s why we try to avoid estimating the enthalpy from the total potential energy difference. But I think you’re right. The integration error will probably make estimating the entropic content from heat transfer unreliable.