Closed jtec closed 1 month ago
By adding @line_profiler.profile, and running
@line_profiler.profile
https://github.com/jtec/prx/blob/85c006479a402380ecac369a7feacaf6bdf8a6c0/src/prx/test/test_main.py#L139
with LINE_PROFILE=1
LINE_PROFILE=1
we can see that assigning carrier frequencies to observations takes a whooping 13 % of the time spent in https://github.com/jtec/prx/blob/9a90fc804f81e29af47a296a41dfe951502b022c/src/prx/main.py#L211
426 2 230039.0 115019.5 13.1 flat_obs.loc[:, "carrier_frequency_hz"] = flat_obs.apply( 427 1 3.0 3.0 0.0 signal_2_carrier_frequency, axis=1 428 )
This PR reduces this to 1.8%:
413 2 234.0 117.0 0.0 glo_cdma = flat_obs[ 414 2 4526.0 2263.0 0.3 (flat_obs.satellite.str[0] == "R") 415 1 4537.0 4537.0 0.3 & (flat_obs["observation_type"].str[1].astype(int) > 2) 416 ] 417 1 369.0 369.0 0.0 flat_obs.loc[glo_cdma.index, "frequency_slot"] = int(1) 418 419 1 4.0 4.0 0.0 def assign_carrier_frequencies(flat_obs): 420 freq_dict = pd.json_normalize(carrier_frequencies_hz(), sep="_").to_dict( 421 orient="records" 422 )[0] 423 assignable = flat_obs.frequency_slot.notna() 424 keys = ( 425 flat_obs.satellite[assignable].str[0] 426 + "_L" 427 + flat_obs["observation_type"][assignable].str[1] 428 + "_" 429 + flat_obs.frequency_slot[assignable].astype(int).astype(str) 430 ) 431 flat_obs.loc[:, "carrier_frequency_hz"] = keys.map(freq_dict) 432 return flat_obs 433 434 1 18716.0 18716.0 1.2 flat_obs = assign_carrier_frequencies(flat_obs)
By adding
@line_profiler.profile, and runninghttps://github.com/jtec/prx/blob/85c006479a402380ecac369a7feacaf6bdf8a6c0/src/prx/test/test_main.py#L139
with
LINE_PROFILE=1we can see that assigning carrier frequencies to observations takes a whooping 13 % of the time spent in https://github.com/jtec/prx/blob/9a90fc804f81e29af47a296a41dfe951502b022c/src/prx/main.py#L211
This PR reduces this to 1.8%: