Getting LArSoft in shape for DUNE

A simulated charged-current interaction for a 3-GeV electron neutrino. The final state has a 1-GeV electron (the down-going electromagnetic shower) and a 1.6-GeV positively charged pion (the up-going hadronic shower going).
 A simulated charged-current interaction for a 3-GeV electron neutrino. The final state has a 1-GeV electron (the down-going electromagnetic shower) and a 1.6-GeV positively charged pion (the up-going hadronic shower going).
A 3-GeV electron neutrino charged-current interaction simulated with LArSoft and reconstructed with Wire Cell 3D imaging. The final state has a 1-GeV electron (the down-going electromagnetic shower) and a 1.6-GeV positively charged pion (the up-going hadronic shower). Courtesy Xin Qian

“This preliminary report represents a tremendous amount of work by the working groups to document their progress,” said Lisa Whitehead, the Far Detector Optimization Task Force convener, at the October 21st collaboration phone meeting. “At about 80 pages, it’s not a quick read, but it’s a great place to learn more about reconstruction in LArTPCs and DUNE physics!”

This month nus2surf reports on two of the efforts covered in this report, single-phase (SP) far detector reconstruction, and detector optimization studies, led by Tingjun Yang and Xin Qian, and dual-phase (DP) simulation and reconstruction, led by Vyacheslav Galymov.  LArSoft, an event simulation and reconstruction software framework, provides the basis for this work.

SP reconstruction in LArSoft

Courtesy Erica Snider
Courtesy Erica Snider

A full-chain automatic reconstruction for the SP LArTPC prototypes and detector modules is now available to evaluate the physics sensitivity for CP violation using the planned LBNF neutrino beam, report Yang and Qian. As with any chain, it comprises several links. Signal processing enables recovery of the ionization electron information from the raw TPC signal, then a hit-finder identifies pulses. A disambiguation step reduces the impact of the wrapped wire design (see related article  Unwinding the APA windings), in which three wire segments comprise one induction wire, and thus are read out by a single electronics channel. Line and blurred clustering groups the hits into 2D line-like or shower-like clusters, respectively.  The Pandora software package addresses the challenges of finding the neutrino interaction vertex and assigning particle identities to groups of clusters.

“Pandora supports a multi-algorithm approach to solving pattern-recognition problems,” said Yang, “so we use it with the Projection Matching Algorithm (PMA) — good for building and optimizing objects in 3D — and EM Shower, that of course finds 3D showers and their properties.”

The reconstruction chain has been incorporated into the standard LArSoft distribution, but the task force continues to improve the performance.

DP sim/reco integrated into LArSoft

dp_det2-open
Schematic of DP detector module, with cathode plane at the bottom, and readout anode plane at the top, above the gas volume.

LArSoft was not originally developed with a DP LArTPC design in mind, and Galymov and his team have worked hard to integrate this design into the software.

“We have successfully completed the DP simulation with charge collection and readout in a short time span,” said Galymov. “Our initial attempts at event reconstruction have not turned up any major software glitches that would hinder further progress.”

“And for the first time,” he added, “we have simulated samples of the neutrino interactions in the DUNE DP detector as part of an official Monte-Carlo production release. The collaboration can now start looking at this detector option while continuing to develop reconstruction tools and analyses.”

More work is needed to bring the software for the DP detector up to the level where the many advantages of this technology can be fully exploited. From the simulation side, the team needs to address a technical issue with setting the electric field direction inside the TPC volume and implement the scintillation light propagation for the DP geometry. On the reconstruction side, they still need to optimize the tools for this detector technology.

“For example, the hit finding — the first step and one of the principal reconstruction tasks — needs work,” said Galymov. “We need to fine-tune it to take advantage of two important features of this design: the high signal-to-noise ratio afforded by the charge amplification, and the fact that the charge readout relies only on collection views.”

The team plans to implement the light simulation and provide better-optimized reconstruction tools for the Task Force final report in March 2017.

Resources

This software developed by the task force is part of the LArSoft distribution and is available to all, with instructions provided.  Young DUNE hosted a one-day software tutorial after the September collaboration meeting at Fermilab, and another was held at CERN at the end of October. Links and resources are available at the meeting web pages.