Wire Cell Toolkit Configuration Data

Overview

This area holds “configuration data” for WCT. Confusingly, this is distinct from but similar to WCT software configuration (see wire-cell-cfg package) as much of if is in the same JSON format as configuration. This is also not “data” in the sense of what WCT simulation, signal processing or reconstruction produces.

Rather, “configuration data” is large, “bulk” input information which WCT needs to do its job but which is too voluminous for humans to type out. Rather, these configuration data files are generated with some external mechanism and then converted to ready-to-consume JSON by code in wire-cell-python).

Using this data

To use this data simply clone this repository into a directory and place that directory in your WIRECELL_PATH. The history of this repository is not particularly useful so to save time and disk space a shallow clone is enough:

$ git clone --depth 1 https://github.com/WireCell/wire-cell-data.git
$ export WIRECELL_PATH=$(pwd)/wire-cell-data

Tour of data files

All WCT configuration data files are in JSON format and optionally compressed with BZip2. The files are described here broken into various categories of purpose. Be wary that the descriptions that follow may not be comprehensive nor up to date. Also, only WCT configuration data files are committed to the repository. The “upstream” files are named below and may be available here. There are also some diagnostic plots related to these files here.

Wires

Wires (wire segments) in WCT are described by their physical endpoints and their logical locations in terms of various numbers and indices. A “WCT wire file” provides an exhaustive list of this info for each wire. Note, WCT can generate wire files in a parameterized way given basic information about an anode plane. See

$ wirecell-util make-wires --help

However, some wires are taken from other programs and then converted to WCT wire files. See

$ wirecell-util convert-oneside-wires --help

• ChannelWireGeometry_v2.txt a dump of MB wires from larsoft in so called “celltree wire format”
• ./microboone-celltree-wires-v2.json.bz2 the conversion of the above
• ./microboone-celltree-wires-v2.1.json.bz2 the conversion of the above, fixing wire index in plane

See this plot for graphical representation of these wires. (Note, a transform may be applied to their locations in WCT, so don’t take the positions shown here as gospel.)

• WireGeometry_dune35t_v5_tpc1.txt “celltree wire format” for DUNE 35t prototype
• ./dune35t-tpc1-celltree-wires-v5.json.bz2 WCT wire file from above
• ./pdsp-wires.json.bz2 protoDUNE single-phase wires for one APA, generated by make-wires
• LArSoft wire definition dump for protoDUNE-SP. ./protodune-wires-larsoft-v4.json.bz2. v4 fixed an ordering problem, (see this issue). See this PDF with a graphical dump of this file. Reproduce with

$ wirecell-util plot-wires protodune-wires-larsoft-v4.json.bz2 protodune-wires-larsoft-v4.pdf

Field Response

The field responses are generally calculated by Garfield (so called “truth responses” may be from some other source). The data produced by Garfield are in the form of a family of files named by their impact position and plane letter. Each file spans that impact position relative to many wires. To process this data into “WCT field response files” see:

$ wirecell-sigproc convert-garfield --help

The required “origin” and “speed” are not recorded in the files directly and must be provided on the command line of this converter. Note the converter takes as input a .tar.gz of the Garfield directory of <impact>_<letter>.dat files (and no other files).

• dune_4.71.tar.gz DUNE APA Garfield data
• ./garfield-1d-3planes-21wires-6impacts-dune-v1.json.bz2 WCT field response file from above.
• ./garfield-1d-boundary-path-rev-dune.json.bz2 slight tweak in Garfield running to avoid overpopulating saddle points.
• ./garfield-11impacts-dune.json.bz2 fine-grained field response with 11 impact positions instead of 6.
• ./garfield-1d-boundary-path-rev-dune-no-grid.json.bz2 field response without grid plane.
• ./dune_nogrid_0d8x.json.bz2 field response without grid plane and the bias voltage is reduced by 20 precent.
• ./garfield-1d-boundary-path-rev-dune-ALL-COLLECTION.json.bz2 replace field response of two induction plane with the collection one. It is used for simulation so that one can easily figure out the signal region in the induction plane.

MB has many sets of field response files. The nominal set is called ub_10 (uboone wire model with fields starting at 10cm from the wires).

• ub_10.tar.gz nominal MB Garfield data
• ./ub-10.json.bz2 WCT field response conversion

Alternatives in two dimensions have been created in order to understand the MB shorted wires and a Garfield normalization issue. The shorted wire alternatives have labels

uv-ground

the U and V wires are shorted

vy-ground

the V and Y wires are shorted

The normalization issue is that early Garfield runs seem to produce collection responses which when integrated gives a total charge of about 2.04 electrons instead of the expected 1.0. Initially the responses were normalized such that the average collection wire integral was 1.0. This technique utterly fails for shorted wires where “collection” becomes ill defined. The variants are labeled as:

wnormed

normalize by the average collection (W wire) integral method

absolute

take Garfield normalization as-is

half

simply scale by 0.5

These are the files:

• ./ub-10-absolute.json.bz2
• ./ub-10-half.json.bz2
• ./ub-10-wnormed.json.bz2
• ./ub-10-uv-ground-absolute.json.bz2
• ./ub-10-uv-ground-half.json.bz2
• ./ub-10-uv-ground-wnormed.json.bz2
• ./ub-10-vy-ground-absolute.json.bz2
• ./ub-10-vy-ground-half.json.bz2
• ./ub-10-vy-ground-wnormed.json.bz2

Zeroed

An “almost 1D” FR can be generated from a full 2D FR with:

$ wirecell-sigproc frzero -n 0 \
-o dune-garfield-1d565-wires0.json.bz2 \
dune-garfield-1d565.json.bz2

This example is provided. This retains the fine-binned responses but just for impact positions in the 0-wire region. -n 1 would also include response for charge above the +/- 1 wire region, etc.

PCBro

The PCBro package provides WCT wires and fields files for “strips plus holes” detectors (50-L, eventual DUNE SP VD FD module). Files here are still be a work in progress. Check the above GitHub project for some known issues.

The available PCBro files are named as:

[wires|fields]-<views>-<hole>-<angle>.json.bz2

• first is literal wires or fields giving file type
• <views> is 2 or 3
• <hole> is hole diameter written, eg h2mm5 for 2.5 mm.
• <angle> is characteristic angle of an indication plane

Currently there are 2 2-view corresponding to the initial 50-L R&D detector at CERN and 2 3-views corresponding to the desired hexagonal pattern DUNE SP VD FD. A 3-view “diagonal 48 deg” induction to match the latest 50-L R&D detector is not yet available.

Noise

The WCT simulation can produce proper intrinsic noise waveforms based on a measured noise amplitude distribution expressed in frequency domain. These are provided as simple text files which are then converted to “WCT noise files”. For info on this conversion see

$ wirecell-sigproc convert-noise-spectra --help

As the procedures are improved the vN version label is increased.

• ./microboone-noise-spectra-v1.json.bz2
• ./microboone-noise-spectra-v2.json.bz2

Shorted wires

• MicroBooNE_ShortedWireList.xlsx spreadsheet of MB shorted wires from Brooke.
• ./microboone-shorted-wires.json.bz2 a “WCT shorted wire file” distilled from that.

See this set of plots showing wires listed in the full spreadsheet. The distilled file holds just wire regions that are shorted, listed by their plane number and the wire numbers bounding wires that are shorted.

Deposition

Note, strictly, these types of files should not be included here and may be removed in the future. The current WCT drift and detector response simulation relies on an external interaction/tracking simulation (ie, Geant4 based) to provide initial distribution of energy depositions or of ionization electrons. A couple samples of these are provided.

• ./g4tuple-qsn-v1-fixed.json.bz2
• ./g4tuple-qsn-v2-fixed.json.bz2

DNN-ROI models

TorchScript .ts models loaded by the ~WireCellPytorch~ package nodes (TorchService + ~DNNROIFinding~ or ~DNNROIFindingMultiPlane~). Each subdirectory contains the .ts file plus a small README.md describing provenance (source training repo, run-id, training shape, output activation).

• ./dnnroi/pdhd/CP43.ts — PDHD MobileNetV3-UNet trained on APA0 U+V combined input (1, 3, 1600, 1500), output activation sigmoid. See ./dnnroi/pdhd/README.md for full provenance.