Minutes of the January 30, 2013 Liquid Argon Simulations meeting
Present: Tom Junk, Brian Rebel, Matt Szydagis, Dan (Matt's student), Herb Greenlee, Kate Scholberg, Zelimir Djurcic, Tyler Alion, Mike Kirby, Zepeng Li, Jim Stewart, Denver Whittington, Milind Diwan, Stan Seibert, Jonathan Insler, Bob Wilson
Apologies to those omitted -- I didn't get all the names.
Our agenda is filling up for each week -- we should meet more frequently and reserve a room that's available for more than 1 hr. Tom will ask Stephanie for a meeting room at the Wednesday 1:00 PM time.
Milind brought up the proposal for LBNE to have a software manager. Such a person can help us to bring more people to bear on the tasks at hand. LBNE works together with LArSoft, but has a larger set of needs -- DAQ and online software, far detector simulation, beam simulation, near detector simulation, reconstruction, analysis, physics sensitivity projections, collaborative tools are areas in which LBNE has effort and needs more. The software manager can attend the LBNE meetings focusing on these areas, as well as the LArSoft stakeholder's group meetings and computing infrastructure.
Milind also proposed that the different simulation groups within LBNE work more closely together. The beam, ND and FD simulation efforts do not communicate much currently. There is a beam simulations meeting on January 31 -- Tom will attend that meeting. The production of physics results will require coherent assembly of data from the different pieces of LBNE -- beam, ND and FD, and it is good if people focusing on getting results out are familiar with all of the pieces. Online and offline for the FD should share the ART framework and software to make sure that data can be passed easily between the detector and production, and simulation and production are naturally closely tied. The beam simulation is a different kind of thing, and we should think about the proper interface (currently, it's flux files using root, but we need systematic uncertainties of course). ND may benefit by collaborating with FD simulations, but if it's not a liquid argon detector, they won't need LArSoft. But should use ART.
Tyler presented an update of the LBNE detector geometry. The GDML is done (modulo the hill and service building ARB8 rotations, to do), and this week has been devoted to volume object sorting. There is now a new geo::ChannelMapAPAAlg() for LBNE that handles the TPC, planes, and wires with LBNE's unique geometry. The drift direction is in the +X direction for the even-numbered TPC's and in the -X direction for the odd-numbered TPC's. The wire planes are given the numbers 0 for U, 1 for V, and 2 for collection ("Z" now -- it had been called "W" in the code but Z is more natural).
To do: need a nearest channel function (for signal simulation), test the view convention, and get the WirePlaneZLength out of the GDML instead of hard-coding it in the program.
We should be able to push an event through very soon (within a week). High-priority item: we need an event display. It is natural to work on the event display after the choices of numbering the volumes have been made, so we are doing everything in the right order. Tom suggests a simple event display which shows only one view of one APA at a time, using existing event display code. It will require a careful eye to interpret the U and V views, but the collection view should be straightforward. Things to check are to make sure the right kinds of signals show up in each plane (we don't put bipolar signals on the collection wires for example), and to check that tracks that exit one TPC volume enter the neighboring one,
to make sure our numbering conventions are consistent and understood.
We also need 35T geometry and event display. We need effort to make these happen.
Jonathan has been testing zero suppression and Huffman coding using the MicroBooNE geometry, and seeing how much space single-muon events take up on disk. Current zero suppression looks for zeros, but this will be a FCL-controllable threshold. Jonathan will communicate with Bo Yu about signals the APA's make, and has already talked with Andrzej Szelc about noise parameterization.
Zepeng has been studying building a photon library for LBNE, and has worked with Tyler on the geometry and provided SiPM modeling in the GDML-making perl scripts. Zepeng chose a voxelization density the same as that of MicroBooNE -- and with 20 paddles per APA, the lookup library is expected to be 107.52 times the size of MicroBooNE's. A cryostat's photon library done in this manner will take 31 GBytes, and 11 million hours of CPU to simulate. Symmetry may buy us a factor of 8 (or perhaps 16) which is still not enough. Tom suggests parameterizing light emitted in one APA's TPC's and detected in another APA with functions instead of library tables. The absorption and scattering lengths mean that much of this library will consist of zeros.
Matt updated the comparison of NEST predictions with Icarus data and will go next meeting. If we get a meeting room, it should be Feb. 6.
There are minutes attached to this event.
Show them.