Summary of results for generating sub banks
Below is the summary of results for generating sub-banks that resemble the 5 sub-banks in table 2 of GWTC2 (https://arxiv.org/pdf/2010.14527.pdf):
PSD file used: We want to use the same PSD file as the one used to generate the sub-banks for GWTC2. The PSD file used for this test was copied from ICDS: /ligo/home/ligo.org/gstlalcbc/observing/3/bank-gen/rainbow-bank/H1L1V1psd.xml.gz
Version of the code git hash: commit 512b1d3f
Result summary: number of templates The BNS (row 1) and NSBH (row 2) sub-banks are well reproduced. The BBH (row 3) and low-q BBH (row 4) sub-banks are underestimated. The IMBH (row 5) sub-bank is over-estimated.
row # | min match | # of templates | # of templates in GWTC2 | difference between GWTC2
row 1 | 0.99 | 263294 | 346500 | 76% of that in GWTC2
row 2 | 0.97 | 860743 | 780780 | 110% of that in GWTC2
row 3 | 0.99 | 242545 | 534914 | 45% of that in GWTC2
row 4 | 0.97 | 20697 | 85220 | 24% of that in GWTC2
(in manifold/metric.py, use DEFAULT_DURATION = 4
instead of DEFAULT_DURATION = 2
. DEFAULT_DURATION = 2
results in errors saying "manifold.metric.EigenvalueError: Unable to resolve negative eigenvalues."
row 5 | 0.99 | 19819 | 11349 | 175% of that in GWTC2
Result summary: bank sim Row 1 (BNS sub-bank): The mismatch is larger along the equal mass line and for m1 and m2 larger than 2 solar masses. The manifold bank sim plot suggests that the mismatch is larger for negative chi for m1 around 2 solar masses, but this is not apparant in the bank sim with sbank.
Row 2 (NSBH sub-bank): The mismatch is larger when the total mass larger than 90 solar masses while the chi is negative. From the sbank bank sim, this corresponds to m1 larger than 85 solar masses, m2 larger than 1.75 solar masses, and s1z less than -0.8. In the sbank bank sim, the mismatch is 0.35 for injection eta below 0.025.
Row 3 (BBH sub-bank): The mismatch looks very good, slightly larger for lower masses.
Row 4 (low-q BBH sub-bank): The manifold bank sim suggests that the mismatch is larger when m1 is larger than 100 solar masses, m2 is smaller than 10 solar masses, and chi is smaller than -0.75. The sbank bank sim doesn't show this tendency, but instead has a point (total mass 170 solar masses, chi -0.9, mchirp 17 solar masses, injection eta ~ 0, m1 160 solar masses, m2 3 solar masses, s1z ~ 1.0, s2z 0.7) has mismatch of 0.25.
Row 5 (IMBH ssub-bank): The mismatch looks very good, slightly larger for lower masses.
Work package on Open Project is here:**[https://projects.gwave.psu.edu/projects/shio/work_packages/1174/activity#activity-16?#activity-16 ]