-
Notifications
You must be signed in to change notification settings - Fork 356
PyCBC offline multi ifo coincident search
Gareth S Davies edited this page May 22, 2020
·
57 revisions
| Issue | description | Assigned to | Status |
|---|---|---|---|
| Coincidence finding | link to code | Marton | Done (#2449) |
| Foreground censor | Remove coincs from 'closed box' single-ifo results plotted / followed up | Marton | Done #2568 |
| Default multi-ifo statistic | Calculate simple ranking stat and IFAR for each type of coinc, combine IFAR via trials factor | Gareth | Done (part of #2473) |
| Improved multi-ifo statistic:noise & network sensitivity model | Calculate and implement noise element of multi-ifo ranking statistic | Gareth | see Multiifo-Ranking-Statistic-Development for components of this |
| Improved multi-ifo statistic:signal model | Calculate and implement signal element of multi-ifo ranking statistic | Gareth+Alex | Done |
| Statmap | Combine Coincs to map between stat and fap and hierarchical removal | Marton | Done |
| Statmap Inj | equivalent of statmap, but for injinj/fullinj/injfull | Marton | Done (#2474) |
| Combine_statmap | Combine coincidences from different combinations | Gareth | Done #2473 |
| Add combine_statmap to workflow | above code is currently not included in workflow | Marton | Done #2539 |
| hdf_injfind | match injections to times of coincidences | Gareth | Done #2499 |
| Add hdf_injfind to workflow | make it possible to give --ifos as an input to hdf_injfind |
Marton | Done #2539 |
Plotting/result display codes - list incomplete
|
Codes may be currently hard-coded for H1L1 - see 'Failing plotting codes' below | Gareth + Ian | |
| Add plots/tables to workflow | See below table | Some plots present, in progress | |
| Fix storage format of coinc segments in coinc code and jobs downstream | Don't have redundant/misleading information duplicated between different datasets | Marton (Gareth to check implications for downstream code) | Done, PR #2517 |
| Hierarchical Removal | Hierarchical removal has been missed or ignored in some codes. Add this in | Tom/Gareth | Done |
| Optimisation | HLV coinc jobs are using a lot of memory | Ian/Tom/Gareth |
i.e. adding things to results pages
| Issue | description | Assigned to | Status |
|---|---|---|---|
| Loudest coincidence table | Open box results page needs a loudest coincidence table | Koustav is working on this | Done #3279 #3284 |
| SNRIFAR plot | Weird overlapping numbers issue, see e.g. this plot. Also wants ylabel on right hand side to say what these mean | Koustav | Done |
| SNRIFAR plot | Plot should return IFO combination in title for differentiation on summary pages | Connor | Done |
| Loudest events Detector Combination | Explicitly state the detector combination in Loudest Event Followup (both open box and missed injections) | Unassigned | |
| Loudest events IFAR | If IFAR is low it is showing as 0.00, check whether this should be in sig figures (i.e g style string formatting) |
Unassigned |
The following plotting codes are currently known to fail.
| code | description | Status |
|---|---|---|
| pycbc_page_foreground | This is supposed to run once and dumps the list of triggers in a table format. Looks like it's not understanding the new time format. | To Do |
| pycbc_page_ifar | This runs a bunch of times to make various counts vs FAR plots. This appears hardcoded for H1L1, so fails on the individual STATMAP files. It also fails on the combined STATMAP job as a bunch of the background stuff is not stored there. Not sure what plot we might want to make on the combined file. | |
| pycbc_page_snrifar | This also runs a bunch of times and makes various statistic vs FAR plots. Some instances of this are running, but it fails when running on the combined STATMAP file. Might need to think about what we actually want in that case as not all of the background information is now stored. | Global IFAR plot is being made #3078 |
| pycbc_page_snrratehist | Same situation as pycbc_page_snrifar
|
|
| pycbc_injection_minifollowup | This is failing complaining that some trigger times are not within data segments ... Possibly the code is trying to followup H1L1 missing injections also in V1?? | Working |
| pycbc_page_coinc_snrchi | This is supposed to run a bunch of times, I think, to make chisq plots. Seems to be expecting old format input (trigger_id1 doesn't exist) | Working |
| pycbc_page_injtable | This runs once per injection run and makes a table of injection details. I believe this code is expecting old format input (time1 doesn't exist) | Working #3105 |
To see failures, if you have the current version of the code installed in CIT go to this directory:
/home/spxiwh/aLIGO/O3/devel/offline_multiifo/DEBUG_JOBS
and run . ${EXECUTABLE_OF_CHOICE}_failure.sh. The error message can be viewed from within this directory, but you can also copy the whole thing away to aid debugging.
- pycbc_page_foundmissed - This runs a lot of times and makes found-missed injection diagnostic plots. (#2516, #2553, #2581)
- pycbc_foreground_minifollowup - This runs once and sets up the minifollowups. Done by Marton #2579