CLAS12 RICH alignment problem

• CLAS12 RICH alignment parameters highly correlated
• Many possible reflection paths produced from a given aerogel tile (each tile with some unique index of refraction)
• Misalignment of mirrors and aerogel layers
• Goal of this work: simultaneously determine alignment parameters for ALL optical components of the RICH (52 parameters total)
• Utilizing Trust Region Bayesian Optimization (TuRBO, arXiv:1910.01739)
• Constructs local model within adaptive trust region after global initialization

1

Aerogel layer 1

Aerogel layer 3/4

Aerogel layer 2

Spherical mirrors

Planar mirrors

MAPMTs

Alignment objective construction

2

Minimize variation between topologies from the same tile

Alignment results