From 808599ff438f31a0ecb695b2ea18962391d25dd7 Mon Sep 17 00:00:00 2001
From: Surabhi Sachdev <surabhi.sachdev@ligo.org>
Date: Wed, 20 Feb 2019 08:11:14 -0800
Subject: [PATCH] gstlal_inspiral_mass_model: - tweak ligo mass model by
making the bns broader, the power law for bbh steeper, and redcucing the
relative BNS rate - also changes the assumption about template ids being
from 0 to len(template bank)-1
---
gstlal-inspiral/bin/gstlal_inspiral_mass_model | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/gstlal-inspiral/bin/gstlal_inspiral_mass_model b/gstlal-inspiral/bin/gstlal_inspiral_mass_model
index 849ed92f6f..76c4990050 100755
--- a/gstlal-inspiral/bin/gstlal_inspiral_mass_model
+++ b/gstlal-inspiral/bin/gstlal_inspiral_mass_model
@@ -74,7 +74,7 @@ mchirps = {}
ligo_min = 3
ligo_max = 40
ligo_peak = 40
-ligo_alpha = -1
+ligo_alpha = -1.5
ligonorm = schechter_norm(ligo_min, ligo_max, ligo_peak, ligo_alpha)
bbh_min = 3
@@ -120,7 +120,7 @@ for row in sngl_inspiral_table:
#
# assume a 0.15 solar mass std deviation, this should capture both population distribution and snr effects
- sigma = 0.15
+ sigma = 0.25
mean = 1.2
bnsprob = 1. / (2 * numpy.pi * sigma**2)**.5 * numpy.exp(-(mchirp - mean)**2 / 2. / sigma**2)
@@ -135,7 +135,7 @@ for row in sngl_inspiral_table:
#
# make intrinsic BNS rate 10 times higher (it is actually probably 20 times higher)
- bns_rate = 10.
+ bns_rate = 7.
bbh_rate = 1.
# FIXME if the noise is ever normalized over mass then we would need the following, but it isn't
#prob[row.template_id] = (bns_rate * bnsprob + bbh_rate * bbhprob) / massBA[(mchirp,)] / (bns_rate + bbh_rate)
@@ -146,9 +146,9 @@ for row in sngl_inspiral_table:
ids = sorted(prob.keys())
norm = numpy.sum(prob.values())
chirpmasses = numpy.array([mchirps[tid] for tid in ids])
-coefficients = numpy.zeros((1, 1, max(ids)+1), dtype=float)
-for tid in ids:
- coefficients[0,0,tid] = numpy.log(prob[tid]) - numpy.log(norm)
+coefficients = numpy.zeros((1, 1, len(ids)+1), dtype=float)
+for idx, tid in enumerate(ids):
+ coefficients[0,0,idx] = numpy.log(prob[tid]) - numpy.log(norm)
#import matplotlib
#matplotlib.use('agg')
--
GitLab