Intro to Molecular Dynamics

Classical particle simulation using forces, time integration, and energy analysis.

Conceptual context

Molecular dynamics turns a microscopic description into a physically calculated movie. Each particle has $r$, $v$, and $m$, while interactions determine a $F$. The computer does not know the trajectory in advance; it reconstructs it through small $t$ updates. The key question is not only where particles move, but whether the calculation respects physics. This is tested by tracking $K$, $U$, and especially $E$ over $t$. Stable behavior supports interpretations of vibration, diffusion, equilibrium, and transport. Artificial drift may instead reveal an unsuitable $dt$ or an incorrectly implemented interaction model.

🟢 Essential level

A molecular dynamics simulation follows many particles as classical point-like objects. At each instant, their $r$ determines the interaction they feel. That interaction produces a $F$, which modifies the motion. The process repeats using a small $dt$, creating a digital trajectory. During motion, kinetic and potential energy usually exchange: one may increase while the other decreases. In an isolated system with conservative forces, $E$ should remain nearly constant. The curve need not be perfectly flat because the calculation is approximate, but it should not show continuous drift. Its evolution therefore acts as a numerical health trace, revealing whether the simulation preserves physical structure or quietly injects fictitious energy.

It also reveals whether microscopic motion is being resolved consistently across successive updates.

This combined energy reading is more informative than watching a single particle: one trajectory may look erratic while the total energy of the ensemble stays stable, showing that the apparent noise is genuine physical exchange rather than a numerical failure.

🔵 Formal level

Basic formulas

The evolution starts by obtaining $F$ from $U_config$ and $gradU$:

Force derived from the potential

The resulting force determines each particle's $a$ through:

Newton's second law for a particle

A $t$ integrator then updates $r$ and $v$. Velocity Verlet first predicts the new $r$, evaluates $a_new$ there, and finally obtains $v_new$:

Velocity Verlet scheme

Real student questions

Why is $E$ not exactly constant? Because a continuous trajectory is approximated by discrete updates. A good integrator keeps the error small and bounded.

Does negative energy mean failure? No. Its sign depends on the reference chosen for $U$.

Can I make $dt$ much larger to finish sooner? Only if the fastest dynamics remain resolved. Otherwise, false trajectories or instability appear.

Is instantaneous $T$ the real $T$? It is a fluctuating estimate that gains meaning through averaging and enough $f$.

Does energy conservation prove the model is correct? No. It supports numerical consistency, but $U_config$ may still represent the physics poorly.

Final synthesis

Molecular dynamics integrates forces to build microscopic trajectories. Its quality depends on the validity of $U_config$, the resolution of $dt$, and the evolution of $E$. A stable graph is not decoration; it is evidence that motion, interaction, and computation remain physically connected.