A Bond-Rotation Approach to Molecular Simulation Theory

Conventional molecular simulation approaches are based on calculating the forces acting on individual atoms in a molecular system, followed by numerical integration of the corresponding Newtonian equations of motion. Algorithms such as the Velocity Verlet scheme are typically employed to propagate these dynamics and generate time trajectories. Although this atomistic framework has been highly successful, it also exhibits certain limitations. In particular, the equations of motion are defined on a per-atom basis, whereas in practice atomic displacements are often coordinated, with groups of atoms moving synchronously rather than independently. Furthermore, the dominant internal degrees of freedom in molecules correspond to rotations about single bonds, a feature that is not strongly represented in current models. Here, I introduce a theoretical framework for molecular simulation that represents motion in terms of collective rotations of groups of atoms about bonds, thereby emphasizing bond-rotational degrees of freedom. The method is physically motivated yet conceptually straightforward. Preliminary comparisons of structures generated by this approach with predictions from AlphaFold3 show decent agreement for small proteins, suggesting that this rotational framework may offer a complementary perspective to existing force-based models.