Using the graph as a reference, calculate the slope of the graph and plug the information into the equation. The units of the slope are angstroms squared over picoseconds.

Note: though the equation suggests that the slope near the end of the graph should be taken, ptraj's diffusion tool is most accurate at the beginning of the graph. This is somewhat apparent, as noise seems to increase as x increases.

Also note: The thermostat and barostat may disrupt the diffusion coefficient. To get more accurate measurements, run another simulation with the thermostat and barostat turned off.

In this case, the slope is 0.0635990 angstroms squared over picoseconds, or 63.5990 * 10 -11 m 2 over seconds (the units our reference paper uses). To find the diffusion coefficient, we need to divide this number by six, which gives us 10.5998 * 10 -11 m 2 over seconds. In comparison, the reference paper's diffusion constant is 14.9 * 10 -11 m 2 over seconds. A possible explanation for the difference is that our molecules use general Amber force fields, whereas the paper uses refined force fields.