Copied from original trac ticket:
https://laura.hepforge.org/trac/ticket/48

As mentioned in ticket #11, comment 27 there are still some large savings that could be made in the recalculation of the integrals. These should be investigated and implemented if feasible.

The comment in ticket 11 was:
One possibility to save a lot of time is to cache all the kinematics information for each grid point and for each event since approx 40-50% of the CPU time is in recalculating all the helicity angles, daughter momenta etc. from the two DP co-ordinates. We'd need to check the memory consumption very carefully but it could be a big saving.

A further comment added to this ticket was:
Caching the full kinematics will require moving the per-event information data members currently in LauKinematics into a POD object in order to minimise memory usage. LauKinematics can then own one of those objects (or even a pointer to one in order to avoid copying?). Should also try and reduce the memory consumed by other parts of the code - ticket #14 (now T39) should help with this. So these tickets should probably be addressed together.