Wednesday, January 06, 2010

Project ced

One of my projects is to provide visualization for the upgraded 12 Billion Electron Volt (12 GeV) CEBAF Large Acceptance Spectrometer (CLAS) detector. Shown below is an image from an early version of our application ced (cLAS eVENT dISPLAY--pronounced "See-E-Dee"). The image shows simulated events (in this case from a photon striking a nucleus) created by a GEANT4 Monte Carlo of the upgraded (and as yet un-built) detector. Here we see a tiny portion of the detector showing resulting electrons (red) and pions (orange) traveling through drift chambers, and leaving "hits" in hexagonal cells centered about high voltage sense wires (tiny red dots). If you look closely you see circles in the cells. These represent "distances of closest approach." In a nutshell, the actual trajectories are tangent to these circles. The black cells represent hits that a built-in noise/segment finding algorithm has marked as irrelevant, uncorrelated noise. The shaded areas around the hits cells also represent information from the noise/segment finding algorithm. The colored background represents the magnetic field produced by a huge six-coiled superconducting magnet.

What I see is a relatively high energy pion (orange), which gets bent less by the magnetic field, traveling through from the lower left (the direction of the target, not shown) and bending gracefully to the right. And probably a single low-energy electron (red) which will get bent a lot by the field, starting at the lower right of the image loop-de-looping to the left, exiting and re-entering the chamber leaving a big mess in its wake. The fact that the pion and the electron are bent in different directions tells you that this is a positively charged pion.

The purpose of visualization software is not primarily to see what the detector looks like. The purpose is a) diagnostic and b) supportive. The visualization helps to uncover problems and bugs in the hardware, and supports the development of analysis code. For example with the previous version of ced written for the current 6 GeV CLAS detector, someone--by looking at hundreds of real events, correctly concluded that two adjacent sense wires (out of 30,000!) were crossed-wired, such that when one cell was hit it was mistakenly attributed to its neighbor.

Since this is a simulation we are displaying "truth." That is, we know in advance exactly what we are looking at. With real data the job of the analysis software will be disentangle the truth from raw data. Imagine all the hits are not colored, and all you really have is the location of the hits and the some rather simple timing data and analog signals from your electronics--from this you have to extract what types of particles went where, and what was their momentum and energy.

The visualization is written entirely in JAVA. I have a student working to create a FLEX front end that would run in your browser's FLASH player. If that succeeds then anyone will be able to hit a web page and look at live events. (A few years from now, when the detector is built and we go live.)

From time to time I will post more ced images as the software matures.

No comments:

Post a Comment