Code review comment for lp:~maddevelopers/mg5amcnlo/2.7.3_tflipping

Thanks Olivier,
how did you achieve the speed improvement for the matrix elements?
Cheers,

Marco

On Sun, Dec 6, 2020 at 10:08 PM Olivier Mattelaer <
<email address hidden>> wrote:

> Olivier Mattelaer has proposed merging
> lp:~maddevelopers/mg5amcnlo/2.7.3_tflipping into lp:mg5amcnlo.
>
> Commit message:
> implement helicity recycling at LO and new phase-space integration strategy
>
> Requested reviews:
> MadTeam (madteam)
>
> For more details, see:
>
> https://code.launchpad.net/~maddevelopers/mg5amcnlo/2.7.3_tflipping/+merge/394923
>
> This branch implements some significant speed-up to the LO code
>
> with two types of speed-up
> 1) faster method to evaluation tree-level (for LO) matrix-element --same
> can be implemented at NLO--
> 2) improved phase-space integration method.
>
> The Paper is close to be finalized.
>
>
> The new phase-space improves (even more since 2.8.0 started to improve it
> already) the speed mainly for VBF processes. This branch has been
> extensively used for muon collider study where the previous version of the
> code was not even converging.
>
> The modification of the matrix-element provides a factor of ~2 in speed
> for complicated process (like g g > t t~ g g) but does not improves the
> situation much for very simple process.
>
>
> This will be released as 2.9.0
>
> --
> Your team MadDevelopers is subscribed to branch
> lp:~maddevelopers/mg5amcnlo/2.7.3_tflipping.
>