[gmx-users] Velocity Verlet integrator

[Mario Fernández Pendás]

Dear all,

In terms of computational efficiency, why the velocity Verlet integrator is
implemented in GROMACS in one full step for positions and two half steps
for velocities?

Would it be more efficient to merge the second halft step for velocities
with the first half step of the following scheme, ie, integrating in one
full step for velocities?

Thank you very much,
Mario Fernánez-Pendás