You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I saw that in FAX/HAX there is also an option the set the drift_field and s2_mean_area_fraction_top. I don't know if we need them but we could add them to the sciencerun flag and use the actual values. At least the drift_field could be adjusted by the field macros.
The text was updated successfully, but these errors were encountered:
I didn't add drift_field yet since fax is currently using a simple model here, which bypasses the s1_ER_recombination_time model that depends on drift_field. You can add it, but just include a note saying it's not currently used. Also, I just hardcoded the EFIELD parameter in run_sim.sh, but can be improved as you described (pulling from the G4 macro file).
Indeed we should tune s2_mean_area_fraction_top too, as well as all the correction maps as defined in hax, feeding into pax (S1: s1_light_yield_map, S2: s2_light_yield_map, FDC: rz_position_distortion_map). Would you be able to add these?
I saw that in FAX/HAX there is also an option the set the
drift_field
ands2_mean_area_fraction_top
. I don't know if we need them but we could add them to thesciencerun
flag and use the actual values. At least thedrift_field
could be adjusted by the field macros.The text was updated successfully, but these errors were encountered: