GitLab

Original: 0aac7088bc71670a195b7d295c3fe4b7cf6446a3

Original: b48fa9544935b54969661f2e6d6a021cdee5c597

In O2, CBC detection pipelines are using TaylorF2threePointFivePN for
systems with (m1 + m2) < 4 Msun and SEOBNRv4_ROM otherwise.
Original: f7506211db049f7a9463664749fa9d4a27a90366

(cherry picked from commit 78b44c14f1adeab6981445bd90dc91e894590772)
Original: b3b385f2c88f099b3ec9254bd0415061068b10bb

Original: 78b44c14f1adeab6981445bd90dc91e894590772

(cherry picked from commit e003ded0b3c28874a99777aa249840bff990a2f7)
Original: 71933781441c468a9a5854a0f3b9c3a32e3ffae7

Seaborn imports pyplot, which configures the Matplotlib backend. So we must
set a non-X11 (e.g. agg or no-op) before importing Seaborn.
(cherry picked from commit 04a4b4a1c68378fe7169fdd49ecb2e256982341e)
Original: 55f24a7c6042001cac223ca22c9f4e3993ae0f6d

Remove the old lvalert_listen mode. Instead, support a mode where
it reads in GraceDB IDs to analyze from stdin.
Original: 146c64ce16def7fff0de241ec0ae998d0f2cc83c

Original: e003ded0b3c28874a99777aa249840bff990a2f7

Seaborn imports pyplot, which configures the Matplotlib backend. So we must
set a non-X11 (e.g. agg or no-op) before importing Seaborn.
Original: 04a4b4a1c68378fe7169fdd49ecb2e256982341e

Original: 124e0233770cdee16817ae99be0296e50a60c1a6

This feature is unreviewed. Revert this commit once it is reviewed.
Original: 5fc7b8eede53e3aee325dd16f0916e5d77636510

Original: 90d8cbc6cf030edcb87e83d3b112bd6b17d6fc83

Original: 30a7bff4ca9903457606406305a1e7684b791018

Original: d1a8464ea9c2b53ed2d2caa9a6a911df5c0302e8

The operation is not atomic.
Original: f6aa68008cfcf85fc22536a9191986f826db08ff

Original: acae82c36e841a4a9ec73565ada6f2be56b2cc3b

Original: a71a7dbde7b8195b1ed33414ffd2d28b6dfc56ff

Original: 52d9ee638ff91f5103af7ab485172ea9239af15f

Original: 2904d428cde78f5215c95efbedd809395c735d4c

Use plt.switch_backend(...) insteads of matplotlib.use(...) to avoid issues
with matplotlib import order and initialization. This is essential for
plotting with Seaborn.
Original: 1e2a818931b1b53a9195cbc6f1aa282ccf72a59c

Original: 68deef7f60a98abfcfb651af62477b386a3bc7a5

Some folks may have noticed that BAYESTAR sky maps in O1 occasionally
exhibit many narrow, widely, regularly spaced rings. Here are a few
recent examples:
https://gracedb.ligo.org/events/view/G205398
https://gracedb.ligo.org/events/view/G203488
https://gracedb.ligo.org/events/view/G187935
I have heard a few people speculate that this unusual morphology might
be a standard feature of glitches or very faint events, but I was never
convinced. I am now certain that these are artifacts due to the
conditioning of the PSDs from the online gstlal pipeline.
These PSDs artificially roll off to zero starting at about 95% of their
maximum frequency of 1024 Hz. Presumably this reflects gstlal’s
low-pass filter. It is a problem because it implies to BAYESTAR that
the detector is exquisitely sensitive from about 975 to 1024 Hz. The
resultant autocorrelation sequence has rapid sinusoidal modulations
that reflect the implied high power near the Nyquist rate. This causes
the ring structure in the sky maps because the transverse profile of
the localization annuli are determined by the autocorrelation sequence.
From what I have seen, this only affects low-mass (e.g., BNS and NSBH
systems) because high-mass systems terminate at frequencies below 1024
Hz. It does not affect MBTA candidates because MBTA’s PSDs appear to be
sampled up to 2048 Hz (though MBTA’s PSDs also display a hint of a
rolloff artifact). To be clear, all pipelines (gstlal, MBTA, pycbc)
display PSD rolloff artifacts.
This quick fix for BAYESTAR sets the PSD to infinity at frequencies
above 95% of the maximum rate. We will probably roll back this patch
later once the issue is fixed upstream.
Refs #2890.
Original: ba1c2ddf5535b7765e3754cee1a412515cea5587

Original: f47b0c44020e049d95d1b13b09b33ead26c43879

...including ER8/O1 uberbank mass split
Original: 67eb618026d9bcdcb80d1e44317f13eebe9e3cd7

Default: anti-FINDCHIRP; same as gstlal and MBTA but opposite pycbc.
Original: dbe2a7bd6e2e6958bcaa57c8b0143982a9815edb

This makes the main help message more readable.
Original: 04c8900963d094eb2e7c8a586a295372d5a89da2

Original: cc3901bb90b43a2f373aa906c416cd796927140f

Original: 16adabde739da54d1d6f7f6c9ede18743a8a9b58

Original: af415586ca8585e16abd1c9877d1b9cdaa90d651

The `o1-uberbank` settings is not yet available on master.
Original: 9f4728069436b63e4f30acfa5ff25dccb79ef3ac

Original: 76a745e881dbf042c1d07d30e8776aafb49b9e5b

Original: 8eb755266662c7d49adb27768cd13645d8951b0d

Original: 28a2c89c0a05dbee63989daed4841449d85f4f7a

This makes it easier for them to share command line options.
Original: c91290289c3ebf8fa77c607d8f7ef69c866029d5

It is a common enough to have docstrings that look like this:
"""
Documentation for this script
"""
We don't need the extra leading and trailing newlines in the --help
output.
Original: bdf4c99911aff383c47c532916fba6cd28446daf

- And Python scripts to python/ subdirectory
- Use lalsuite_python.am
Original: 40fe5770b1539aec68f7d6abbad2a67cdf6c42cc

Original: 188391d6b2ccebb06a76aaac08f7bc1b43e44aae

Original: d060ba468defafee39deb83f21d39df313fe1987

- Minimal changes to lalapps build
- cHEALPix now an optional dependency of lalinference
- OpenMP acceleration now optional in lalinference
- Building of Python C extension conditional on Python support
Original: b29f20ff11e62095dbd44e850b248ecc58b08a13