Molecular Dynamics and Free Energy from Sampling
Integrating the equations of motion for a system of atoms produces a trajectory, and averages along that trajectory stand in for thermodynamic quantities. The result is only as good as the sampling, so the honest questions are about convergence and the force field.
What a learner can do afterwards
- Explains what a thermostat does and why a raw trajectory needs one to represent a chosen ensemble
- Judges whether a simulation is converged from the behaviour of a monitored quantity
- Describes how biasing along a coordinate produces a free energy profile
- States which errors come from the force field and which from too little sampling
1 · Read
You move every atom step by step under forces from the force field. A raw run drifts in energy, so you attach a thermostat that holds your chosen temperature. The thermostat keeps the trajectory inside the ensemble that your averages assume.
Suppose you watch the average energy as the run grows. Early on the line slopes downward, so you extend the run. Later it wiggles around a flat line, and only then do you read the average.
Some events are too rare to wait for. You add a bias along one coordinate, such as a distance, to push the system over the barrier. You then strip the bias out in analysis. The corrected histogram becomes the free energy profile.
Two kinds of error look alike, so you test them apart. Wrong bonded terms or charges mislead every run, however long, and that is force field error. A short run misleads even with a good model, and that is sampling error. You extend time to test sampling and you swap models to test the force field.
You hold temperature with a thermostat, grow the run until averages flatten, bias rare events to map free energy, and test sampling and force field apart.
2 · Watch
Take it off screen
Where it sits
8 questions wait behind this lesson, each with its answer explained. Every answer feeds the sky: stars light as they are learned, and dim when it is time to come back.