Reserve sufficiently large vector<double> for MLP observables (!459) · Merge requests · LHCb / Rec · GitLab

Gerhard Raven requested to merge ghost-id-reserve-vector into master Jan 19, 2017

Based on https://indico.cern.ch/event/600120/contributions/2444101/attachments/1398576/2133160/Ghost_Probability_for_upgrade.pdf it seems there is a lot of time to be gained by pre-allocating sufficient memory for the vector of observables. It would be even better if a stack-based container could be used so that any heap allocation would be avoided -- but that requires changes in the MVA interface and implementation.