Remove mass-gap category from p_astro

I think I heard the CBC group wishing to remove such designation?

changed milestone to %O3b

In issue o3break#18 I've suggested to change MG from a p_astro category to a em_bright property. Then within the p_astro category anything above 3 Msun would be considered a black hole.

Please leave a note here and/or send email on list if and when a decision is taken to do this.

MIT f2f decided not to do make this change for O3B

Oh. Any particular reasoning?

Was it for technical reasons? Because I think we'd rather see MassGap as a property rather than a category.

The consensus from the low-latency F2F at MIT was that making this change for the O3b could lead to more problem. Here are some of the arguments:

1. Inconsistencies with previous events from O3a: Changing the definition of a source-category mid-run would make it difficult to compare probabilities of events that happened earlier in the run with those of events that happen after the change in definition of the source category. This can be fixed by recomputing the probabilities of all the past events, but that will also mean we will have to send GCN updates for each events. It is something we would like to avoid since this change is simply because of the change in the definition of what a black hole is, not because of an error in calculation, or any other bug, or improved background analysis.

2. Possibly break infrastructure of the observers: Astronomers around the world have built infrastructure based on what kind of data-products we are sending. If suddenly we have changes in the JSON files with one less key in p-astro and one extra key in em_bright, that might break their codes and will require them to make changes in their infrastructure.

3. Astronomers did not ask for these changes: There has been no requests from the astronomers to make these changes. This is not a reason to not make this change. Rather, if this was the case we would have perhaps given this a higher priority.

Personally, I like this idea of moving the MassGap to source-properties. However, it will be cleanest to do this in O4.

With these points, we are requesting to close this issue.

Thanks Shaon. It is unfortunate that this proposal was floated for so long without being resolved. The reasons for not making any change could have been thought of before the break.

On the last point I want to remind people that we never asked astronomers what they thought of the possible change, so we cannot know. We have not even asked so far whether the existing classification is useful or comprehensible for them. The only feedback I remember is that some people want the chirp mass..

This could be a possible item for the next Open LVEM call where we can ask astronomers what they think of this proposal to change the MassGap from being a quantity in source-classification to something in the source-properties. We can give them the rationale that the probabilities for source-classification are “pure-classes”, while MassGap is a mixed-class with three-classes representing it. Where as in source-properties we do have mixed-classes, like HasNS, so a HasMassGap will fit there naturally. We can propose this as something we are thinking of implementing in O4.

changed milestone to %O4

changed title from Remove mass-gap category from p_astro? to Remove mass-gap category from p_astro

changed milestone to %O4

added Alert Contents label

mentioned in issue lscsoft/p-astro#14

closed