Integrate 2D Signal Processing into ICARUS Reconstruction #777
Conversation
…s of two-faced ICARUS wires config.
… jsonnets for various WC SP configs. Move parameters out of json config into fcl config.
…e due to the different filter width).
…with new H5 interface). Fixes to which frames get created. Don't save H5 files by default.
|
As far as I can tell this is still a draft PR. |
|
If we want to start moving forward with this we should try to get the documentation part of the PR up to speed... As I interpret the presentation at the workshop we should not wait for the DNN ROI just yet? |
|
As this looks to be essentially ready for integration (sans DNN ROIs), can we please get this properly documented now? This is important before signing off to merge... Likely we have a couple of weeks here so hopefully not an onerous task. Thanks!! |
…ilter to accomodate this.
|
trigger build |
1 similar comment
|
trigger build |
|
✔️ CI build for LArSoft Succeeded on slf7 for c14:prof -- details available through the CI dashboard |
|
✔️ CI build for LArSoft Succeeded on slf7 for e26:prof -- details available through the CI dashboard |
|
❌ CI build for ICARUS Failed at phase build ICARUS on slf7 for c14:prof -- details available through the CI dashboard 🚨 For more details about the failed phase, check the build ICARUS phase logs parent CI build details are available through the CI dashboard |
|
❌ CI build for ICARUS Failed at phase build ICARUS on slf7 for e26:prof -- details available through the CI dashboard 🚨 For more details about the failed phase, check the build ICARUS phase logs parent CI build details are available through the CI dashboard |
Title. Ready for merge.
Depends on external updates:
WireCell/wire-cell-toolkit#341
Also needs a WireCell version with this commit merged in:
WireCell/wire-cell-toolkit@e532442
Also needs the following files added to icarus_data:
icarus_fnal_fit_ks_P0nom_P1bin{0,1,...,14}.json.bz2
icarus_wire_twofaced.json.bz2
Integrates WireCell-based 2D deconvolution in to ICARUS reconstruction, with flavors of both traditional and DNN (Deep-Neural-Net) region of interest (ROI) identification.
See previous presentations: DocDB 37529, 38074, 38939, and 40465.
The 2D signal processing deconvolves across both the time and wire directions, which improves the quality (signal-to-noise and charge resolution) of charge signals, especially on the two induction planes. The filters used in this application are wide compared to other experiments, both in the wire and time dimensions. This broadens the overall charge signal relative to the raw charge.
The implementation builds on the previous half-WireCell implementation which combined elements of the 2D and 1D signal processing change.
In order to integrate the new signal processing with mainline reconstruction, the input labels on the recob::Wires to hit finding are updated.