Invalid force and potential computation

[Chelsea486MHz]

So here's me realising for months I've been rendering simulations using invalid computations. Behold the extent of my misery:

• The Lennard-Jones force (force_lennardjones)
• The Lennard-Jones potential (potential_lennardjones)
• The bond angle potential (potential_angle)
• The electrostatic potential (potential_electrostatic)

All provide incorrect results, with consequences ranging from invalid potential reduction to dead on system explosion.

I'll now have a break, have a beer, and proceeed to f@#! kill myself

[Chelsea486MHz]

The following functions are now fixed:

• The Lennard-Jones force (force_lennardjones)
• The Lennard-Jones potential (potential_lennardjones)

Further work will have to decide if the results of potential_electrostatic are satisfactory or not.

[Chelsea486MHz]

The following functions have been identified as faulty:

• force_lennardjones
• potential_lennardjones
• potential_angle

And disabled until a fix is developped.

[Chelsea486MHz]

The following functions are now fixed:

• The Lennard-Jones force (force_lennardjones)
• The Lennard-Jones potential (potential_lennardjones)

[Chelsea486MHz]

Wooooohooooooo

Okay so potential_angle and force_angle are fixed.

The issue was lying in the fact that I was trying to relate forces to angular displacements, while I in fact had to relate torques. Which makes sense.

Anyway, by applying a torque instead of a raw force, I had to scale down the torsion constant (C_AHO in util.h) down to 1E-18 Nm.rad-1. The bond angle potential is now correctly computed.

But something is now heavily getting on my nerves. What I thought was fixed isn't. The LJ potential is fucky.

[Chelsea486MHz]

LJ fixed.