Best practice

Force field

Force fields are usually parametrized based on thermodynamic quantities only. However, NMR relaxation quantities depend on both structural and dynamical quantities.

Simulation precision

NMR relaxation rate measurements are extremely sensitive to the precision of the simulation. The cut-off, for instance, was noted to have a slight impact on \(R_1\).

Box size

NMR relaxation measurements are not extremely sensitive to the box size, however, a small effect of the box size can be see, particularly when reaching extremely small boxes:

Looking at the correlation functions, a strong effect of the box size can be see on the inter-molecular contribution, while almost no effect is seen on the intra-molecular contribution:

Despite the strongly modified correlation functions obtained for small boxes, the relaxation rate is not so affected:

Simulation duration

For comparison with experimental value, the total duration of the simulation must either be larger than \(\tau_c\), where \(\tau_c\) is the longest characteristic motion in the system, or be low enough to match the actual Larmor frequency used in experiments.

Dumping frequency

Dumping period must be smaller than the smaller correlation time of the system, or a significative error on \(R_1\) will be connected.

For bulk system: check isotropy

Although generally true for bulk systems, as showed for glycerol [3], it can be worth ensuring that the relation

\[\frac{1}{6} G^{(0)} (t) = G^{(1)} (t) = \frac{1}{4} G^{(2)} (t)\]

actually stands. For a system of bulk water, the superimposition is clearly verified:

#todo : superimpose water, PEG-water, and slit silica on the same graph

If not, all three correlation functions must be calculated.