remove pulse_cleaning code

snowflake/library.py

@@ -396,71 +396,3 @@ def refine_vertex_time(vertex, time, direction, pulses, omgeo):
     if np.isinf(min_d):
         return time
     return min_t + adj_d/CCC
-
-
-################## pulse cleaning
-def _weighted_quantile_arg(values, weights, q=0.5):
-    indices = np.argsort(values)
-    sorted_indices = np.arange(len(values))[indices]
-    medianidx = (weights[indices].cumsum()/weights[indices].sum()).searchsorted(q)
-    if (0 <= medianidx) and (medianidx < len(values)):
-        return sorted_indices[medianidx]
-    else:
-        return np.nan
-
-
-def weighted_quantile(values, weights, q=0.5):
-    if len(values) != len(weights):
-        raise ValueError("shape of `values` and `weights` don't match!")
-    index = _weighted_quantile_arg(values, weights, q=q)
-    if not np.isnan(index):
-        return values[index]
-    else:
-        return np.nan
-
-
-def weighted_median(values, weights):
-    return weighted_quantile(values, weights, q=0.5)
-
-
-def pulse_cleaning(frame, Pulses, Residual=1.5e3*I3Units.ns):
-    pulses = dataclasses.I3RecoPulseSeriesMap.from_frame(frame, Pulses)
-    mask = dataclasses.I3RecoPulseSeriesMapMask(frame, Pulses)
-    counter, charge = 0, 0
-    qtot = 0
-    times = dataclasses.I3TimeWindowSeriesMap()
-    all_ts = []
-    all_cs = []
-    for omkey, ps in pulses.items():
-        ts = np.asarray([p.time for p in ps])
-        all_ts.extend(ts)
-        cs = np.asarray([p.charge for p in ps])
-        all_cs.extend(cs)
-    tw_start = max(weighted_quantile(np.asarray(all_ts), np.asarray(all_cs), 0.1) - 1000, min(all_ts))
-    tw_stop = min(weighted_quantile(np.asarray(all_ts), np.asarray(all_cs), 0.95) + 1000, max(all_ts))
-    for omkey, ps in pulses.items():
-        ts = np.asarray([p.time for p in ps])
-        cs = np.asarray([p.charge for p in ps])
-        median = weighted_median(ts, cs)
-        dts = np.ediff1d(ts)
-        median_dts = np.median(dts)
-        qtot += cs.sum()
-        for p in ps:            
-            if median_dts > 1200 and len(dts) > 1:
-                # attempt to mask out correlated noise
-                mask.set(omkey, p, False)
-            elif p.time >= min(median+Residual, tw_stop) or p.time < tw_start:
-                latest_time = min(median+Residual, tw_stop)
-                if omkey not in times:
-                    tws = dataclasses.I3TimeWindowSeries()
-                    tws.append(
-                        dataclasses.I3TimeWindow(latest_time, np.inf)
-                        )  # this defines the **excluded** time window
-                    times[omkey] = tws
-                mask.set(omkey, p, False)
-                counter += 1
-                charge += p.charge
-
-    frame[Pulses+"PulseCleaned"] = mask
-    frame[Pulses+"PulseCleanedTimeWindows"] = times
-    frame[Pulses+"PulseCleanedTimeRange"] = dataclasses.I3TimeWindow(tw_start, tw_stop)