You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
When calculating the forces within one cell (or any other quantity) should always be doable with Newton3.
However, when mixing functor calls with Newton3 == false in a Newton3 == true iteration the total energy is off by a factor of 0.5.
This is because of the following:
When calculating the global values, if Newton3 is activated, all values are divided by two:
Is your feature request related to a problem? Please describe.
When calculating the forces within one cell (or any other quantity) should always be doable with
Newton3
.However, when mixing functor calls with
Newton3 == false
in aNewton3 == true
iteration the total energy is off by a factor of 0.5.This is because of the following:
When calculating the global values, if Newton3 is activated, all values are divided by two:
AutoPas/src/autopas/molecularDynamics/LJFunctor.h
Lines 173 to 176 in 0255376
and later if newton 3 is deactivated again:
AutoPas/src/autopas/molecularDynamics/LJFunctor.h
Lines 749 to 754 in 0255376
Describe the solution you'd like
For deciding whether or not to divide everything by 0.5 do not (ab)use the
Newton3
flag but introduce another.Describe alternatives you've considered
🤷♂️
Additional context
At the moment we call the functor twice with
Newton3 == false
.The text was updated successfully, but these errors were encountered: