Closed Suke0811 closed 1 year ago
This actually seems to be almost the same or slower based on some prototyping using pd multi Index. Although this can avoid many for-looping, the overhead coming from pandas is large enough, there were no to little benefits in shifting DefDict to use dataframe. It is best to do C++ and pybind to improve the performance.
Dataframe should be faster and more appropriate for larger data set, but we need to prototype it to see if it is actually better.