Interim report on beam optimization released

beam-optim-cory
Visualization of the optimized focusing system. Credit: Cory Crowley

The DUNE beam optimization task force has delivered its Interim Report. In this report, the task force recommends an optimized focusing system that includes three focusing horns and a significantly longer target than that of the current reference design. The report represents work done since last September in collaboration with the LBNF Beamline project team and the Accelerator and Beam working group; a final report is due in March 2017. The team recognizes the importance of addressing the high-level design choices for the facility, including this system, before CD2.

An accelerator engineering team has developed an initial design (see figure) based on results from a genetic algorithm that, relative to the reference design, improves the predicted flux of muon neutrinos between 0.5 and 4 GeV by 44%, and improves sensitivity to CP violation by 21%. This “survival of the fittest” algorithm simulates random beams, “mates” them together to form better beams, and repeats the process.

The task force has a lot on its plate. In addition to guiding the engineering work through continuing studies of the physics impact of engineering changes, it is comparing neutrino flux uncertainties in the optimized and reference designs, and looking at optimizing for metrics other than CP sensitivity. Other metrics may include, for instance, flux at the second oscillation maximum or measurement of tau appearance. In addition, the task force is addressing questions such as what to use as the atmosphere in the target area (air, N2 or He) and which hadron absorber technology to use.

This process of course involves a series of iterations between physics and engineering. To make it more efficient, collaborators at Lawrence Berkeley Lab (LBL) are developing a new “particle swarm optimization” algorithm, a computational method that was first used to simulate groups of organisms, like schools of fish swarming around a food source, and was found to in fact perform optimization. The LBL algorithm uses a more complex procedure for identifying optimal beams that appears to converge much faster than genetic algorithms do, and preliminary studies indicate that this will substantially reduce the time required for refining parameters of the beam design.

“Optimizing the neutrino beam is as important as optimizing the detector, as the physics sensitivity depends on the product of neutrino beam flux and the detector mass and performance,” said Alfons Weber. “We have a dedicated team that wants to get the most out of this facility. We are continually learning more on how we can make it more powerful. The work over the last nine months has improved our sensitivity as much as having added roughly 50% more detector mass.”