yade-dev team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> It is totally equivalent for me, but proportional displacement seems
> easier and more robust (no risk of resetting the speed in an engine
> before it is actually computed, no special cases in Newton's law,...).
> I would go for proportional instantaneous displacement.
>
> So, the scheme is :
>
> Define the strain rate e'_kk in each direction
>
> - e'_kk = constante
> or
> - e'_kk(t+dt) = e'_kk(t) + (stress_error)/mass
>
> Assign proportional displacement dr = e * r
>
> We are still discussing this here, I'm anticipating a little, but I
> think we can converge to something compatible with that.
I commited that code the day before yesterday, see
NewtonIntegrator::homotheticCellResize; it all happens inside
leapfrogTranslate. (Sega, thanks for the code cleanup in the integrator,
now it looks really beautiful!)

v