improve memory performance
Opened this issue · 2 comments
bwheelz36 commented
https://pandas.pydata.org/docs/user_guide/scale.html
- we could convert any single valued quantities to categorical variables
- same for particle type
- we could force the datatypes for different quantities, e.g. int16. the problem is pandas seems to pretty cheerfully case things back into floats anyway, so would regular checks to enforece this.
bwheelz36 commented
bwheelz36 commented
also, deleting ps_data after it is passed to PhaseSpace saves a substantial amount of memory