TO:  All GCN Sites
RE:  Correction to the LVC Probabilities in the Notices
DT:  29 Aug 2019


INTRODUCTION:

This applies to ONLY those recipients receiving LVC Notice types,
and only those sites that receive them via full-format Email (& LMAIL),
the CBEEP short-form emails, and the 160byte socket format.
(If you are receiving VOEvents or the really short-form email formats,
you are not affected by this mistake.)
My apologies to those of you that my mistake has caused problems at your end.

An error in the processing of LVC notices was discovered.
There are 7 probabilities parameters that LIGO-Virgo provides in the notices
they send to GCN.  Not all 7 are provided in every notice;
about 90% have have all 7 values, but about 10% are missing 1 or more.
(The number missing  depends on which pipeline produced the notice.)

DETAILS EMAIL:

For the PROB_BNS, PROB_NSBH, PROB_BBH, PROB_MassGap, and PROB_TERES
probabilities, if the parameter value was not explicitly given
in the notice, then the outgoing GCN Notice would have a value 1.57
(which is impossible -- all probabilities nust be between 0.00 and 1.00);
example:
    PROB_BNS:    1.57 [0.0-1.0]
For all param values explicitly listed in the notice sent to GCN,
the outgoing value will be displayed properly;
example:
    PROB_BNS:    0.93 [0.0-1.0]

A similar problem applied to the PROB_NS and PROB_REMNANT fields.
For the PROB_NS and PROB_REMNANT fields (in slot 20), the filling of the 2 values
now includes a 0xFFFF masking each of the two values when put into the slot 20.

The above two mistakes in the GCN programs produced emails (full-format & CBEEP sizes only)
with impossible or incorrect values (~10% of the time).
The line in the email indicating an invalid value has now changed from
    PROB_NS:     -1.00 [0.0-1.0]
to
    PROB_NS:     -1 [0.0-1.0] VALUE NOT ASSIGNED!


DETAILS SOCKET PACKET:

For the 160byte-binary socket packet customers:
Knowledge of the missing probability parameters is solved by taking
the information of probability from the original LVC message
and capturing it in a slot in the socket packet.
There is a newly added field (slot 23 in the socket packet)
which contains 7 flag bits -- one for each of the 7 probability values.
Each bit (0-6) is set to 1 if that probability value was NOT available
for inclusion in the notice.  And so your socket_receiver program
should NOT extract that location(s) in the packet (slots 20-22)
for the invalid probabilities.
See  https://gcn.gsfc.nasa.gov/gcn/sock_pkt_def_doc.html for details.



MISC:
If are you not receiving LVC notice types, and would like to,
please go to  https://gcn.gsfc.nasa.gov/gcn/config_builder.html
and select "2. Modify an existing Site Configuration".

REMINDER OF RECENT ADDITIONS TO GCN:
In case you have missed them, here is a brief list of the most recent additions
to the GCN system.  You might consider signing up for these Notice types as well.
The full announcements/descriptions for each of these can be found at:
http://gcn.gsfc.nasa.gov/gcn/whats_new.html
ICECUBE_GOLD/_BRONZE & HAWC (Jun 2019) // Improved Notice types from IceCube and the new HAWC Notice type
COINCIDENCE  (Jan 2018)                // Real-time Temporal & Spatial coincidences between any Missions/Instruments
FERMI_GBM_SUBTHRESHOLD  (July 2017)    // Fermi-GBM subthreshold events
CALET_LC  (Mar 2016)                   // The Flight- & Ground-produced CALET-GBM lightcurves


Sincerely,

Scott Barthelmy                      NASA-GSFC, Code 661, Greenbelt, MD   20771
PHONE:  301-286-3106                 (office)
PHONE:  301-346-3733                 (cell)
EMAIL:  scott@milkyway.gsfc.nasa.gov
TEXT:   3013463733@cingularme.com
WEB:    http://gcn.gsfc.nasa.gov/gcn