The fixed b-tagging weight producer is very slow compared to other producers because of the following line that gets called multiple times:
ak.min([flat_input.pt, 999 * ak.ones_like(flat_input.pt)], axis=0). Apparently getting the minimum of two arrays is very inefficient and should be avoided at all cost or at least reduced to be ran only once. Since the efficiency maps still require jet pT to be lower than 1000 GeV it is still required to call this line. The line can however be moved up in the function and be derived only once for all jets.
The fixed b-tagging weight producer is very slow compared to other producers because of the following line that gets called multiple times:
ak.min([flat_input.pt, 999 * ak.ones_like(flat_input.pt)], axis=0). Apparently getting the minimum of two arrays is very inefficient and should be avoided at all cost or at least reduced to be ran only once. Since the efficiency maps still require jet pT to be lower than 1000 GeV it is still required to call this line. The line can however be moved up in the function and be derived only once for all jets.