search

mina19 / approval_processorMP

MIT License
1 stars 4 forks source link

Only send alerts for single IFO triggers that both pass alert threshold and are found in coincidence with grb #2

Open cmessick opened 7 years ago

cmessick commented 7 years ago

I was asked to open a PR about this so there's an official record somewhere. This is just a request to modify AP to only label single IFO gracedb uploads with ADVREQ if they

  1. Pass the alert FAR threshold and
  2. are found in coincidence with a GRB

It looks like Raven already updates event logs with information about possible coincidences (see e.g. https://gracedb.ligo.org/events/view/T276080, under "External Coincidence"), so I believe the only change needed on the approval processor side will be differentiating between single and double IFO candidates, looking for this information from raven if a single ifo candidate, and then a conditional that describes when to apply the advreq label.

pshawhan commented 7 years ago

There is not a trivial solution since the external trigger may become known to RAVEN after the GW trigger has been passed to Approval Processor, and Approval Processor is designed to decide whether or not to process a GW trigger when it first learns about it.

chadhanna commented 7 years ago

Hi,

Could we consider simply having a patch to approval processor ignore single ifo results for now and then if someone needs to intervene manually they can do so? At least then we can upload to gracedb.

thanks,

Chad

pshawhan commented 7 years ago

Yes, that is one approach, and in fact Deep has written code to do that. I assume that needs a quick review and then can be deployed.

Peter

On Fri, May 19, 2017 at 11:36 AM, chadhanna notifications@github.com wrote:

Hi,

Could we consider simply having a patch to approval processor ignore single ifo results for now and then if someone needs to intervene manually they can do so? At least then we can upload to gracedb.

thanks,

Chad

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2#issuecomment-302736544, or mute the thread https://github.com/notifications/unsubscribe-auth/AKdlWcEd5ymTdGLRUb1Jt-Tc9N-qmUk3ks5r7bcSgaJpZM4NDVX1 .

chadhanna commented 7 years ago

Ok. Great. Can we move forward on this? Looks like we are about to have a stretch of llo only time.

On May 19, 2017 12:49 PM, "Peter Shawhan" notifications@github.com wrote:

Yes, that is one approach, and in fact Deep has written code to do that. I assume that needs a quick review and then can be deployed.

Peter

On Fri, May 19, 2017 at 11:36 AM, chadhanna notifications@github.com wrote:

Hi,

Could we consider simply having a patch to approval processor ignore single ifo results for now and then if someone needs to intervene manually they can do so? At least then we can upload to gracedb.

thanks,

Chad

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2# issuecomment-302736544, or mute the thread https://github.com/notifications/unsubscribe- auth/AKdlWcEd5ymTdGLRUb1Jt-Tc9N-qmUk3ks5r7bcSgaJpZM4NDVX1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2#issuecomment-302754587, or mute the thread https://github.com/notifications/unsubscribe-auth/AF78Xjl-mMGppZYwiLtbr0Obo87n3gqHks5r7cgEgaJpZM4NDVX1 .

pshawhan commented 7 years ago

We will have to make a case to Patrick -- will one of you talk to him? The policy up to now has been that even minor changes to the Approval Processor code have been in the queue behind the major changes that Mina has been working on. But I would think he would be agreeable to deploying this simple change this week.

Peter

On Fri, May 19, 2017 at 1:14 PM, chadhanna notifications@github.com wrote:

Ok. Great. Can we move forward on this? Looks like we are about to have a stretch of llo only time.

On May 19, 2017 12:49 PM, "Peter Shawhan" notifications@github.com wrote:

Yes, that is one approach, and in fact Deep has written code to do that. I assume that needs a quick review and then can be deployed.

Peter

On Fri, May 19, 2017 at 11:36 AM, chadhanna notifications@github.com wrote:

Hi,

Could we consider simply having a patch to approval processor ignore single ifo results for now and then if someone needs to intervene manually they can do so? At least then we can upload to gracedb.

thanks,

Chad

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2# issuecomment-302736544, or mute the thread https://github.com/notifications/unsubscribe- auth/AKdlWcEd5ymTdGLRUb1Jt-Tc9N-qmUk3ks5r7bcSgaJpZM4NDVX1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2# issuecomment-302754587, or mute the thread https://github.com/notifications/unsubscribe-auth/AF78Xjl- mMGppZYwiLtbr0Obo87n3gqHks5r7cgEgaJpZM4NDVX1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2#issuecomment-302760329, or mute the thread https://github.com/notifications/unsubscribe-auth/AKdlWajXl_PdowypOPEP19wptzfcLrgNks5r7c3_gaJpZM4NDVX1 .

chadhanna commented 7 years ago

I will ping him

On May 19, 2017 1:31 PM, "Peter Shawhan" notifications@github.com wrote:

We will have to make a case to Patrick -- will one of you talk to him? The policy up to now has been that even minor changes to the Approval Processor code have been in the queue behind the major changes that Mina has been working on. But I would think he would be agreeable to deploying this simple change this week.

Peter

On Fri, May 19, 2017 at 1:14 PM, chadhanna notifications@github.com wrote:

Ok. Great. Can we move forward on this? Looks like we are about to have a stretch of llo only time.

On May 19, 2017 12:49 PM, "Peter Shawhan" notifications@github.com wrote:

Yes, that is one approach, and in fact Deep has written code to do that. I assume that needs a quick review and then can be deployed.

Peter

On Fri, May 19, 2017 at 11:36 AM, chadhanna notifications@github.com wrote:

Hi,

Could we consider simply having a patch to approval processor ignore single ifo results for now and then if someone needs to intervene manually they can do so? At least then we can upload to gracedb.

thanks,

Chad

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2# issuecomment-302736544, or mute the thread https://github.com/notifications/unsubscribe- auth/AKdlWcEd5ymTdGLRUb1Jt-Tc9N-qmUk3ks5r7bcSgaJpZM4NDVX1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2# issuecomment-302754587, or mute the thread https://github.com/notifications/unsubscribe-auth/AF78Xjl- mMGppZYwiLtbr0Obo87n3gqHks5r7cgEgaJpZM4NDVX1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2# issuecomment-302760329, or mute the thread https://github.com/notifications/unsubscribe-auth/AKdlWajXl_ PdowypOPEP19wptzfcLrgNks5r7c3_gaJpZM4NDVX1 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mina19/approval_processorMP/issues/2#issuecomment-302764058, or mute the thread https://github.com/notifications/unsubscribe-auth/AF78XsIK5lhKQBgAoK-qQZWQS-ii5_BJks5r7dH7gaJpZM4NDVX1 .

prbrady commented 7 years ago

Deep and I talked yesterday about the status of this. The initial approach will not work. He is looking at a different, and probably more sustainable approach, which we will go over this afternoon and hopefully discuss a patch tomorrow morning.