Meeting reports
4/2/2024
- Start with Kolya's analysis update
- Update from Caspar Gutsche, PSI
- Discuss the new beam line designs and their requirements
- Clarify the arguments for training the model with pions and/or muons. Everybody agreed
that both have to be done in the end, but we did not fully discuss the pros and cons for starting
with a restricted and more efficient model vs a full model.
- Intro to running piE5 G4BL. Urs and Peter
3/19/2024
In the BD dynamic meeting today Andy presented his ideas on how to proceed
building on Paul's work, followed by a discussion at each step. Anna stressed the
importance of a measurement based approach for the final optimization and of the
G4BL based studies to guide the required statistics and sensitivities. She might
have a master student interested in the AI/ML approach. Kolya is taking a class
on this subject and also wants to get involved.
Peter went through the beginning of his
AI/ML draft on the pioneer twiki which
is designed as a working document. The initial suggested steps largely coincided
with Andy's verbal plan.
Doug mentioned that in addition to the ML approach, extensions to the current beamline designs
need to be developed. Likely we will always need those as the initial conditions for the
ML optimization range.
We sketched some topics for the next meeting in two weeks.
- Start with Kolya's analysis update
- Discuss the new beam line designs and their requirements
- Clarify the arguments for training the model with pions and/or muons. Everybody agrees
that both have to be done in the end, but we did not fully discuss the pros and cons for starting
with a restricted and more efficient model vs a full model.
Anna will invite Giovanni for the next meetings.
--
Peter Kammel - 2024-04-02
Comments