PC 2015-01-29 Computing Questions

Email from Durga:

Hi Chris,

Some questions about MC production came up at the GRID meeting earlier today. Specifically about the resource needs from the GridPP point of view.

I think it ties with our analysis and physics needs and plans.

I'm afraid it's more like a set of questions now -- something you want to think about or bring up at the analysis meeting?

I'm listing the issues below for completeness:

--> Do we want to do a batch MC for every DAQ run setting that we take?
----- I imagine we do MC production for the 'matrix', but when upstream magnet currents change during data-taking, do we need to regenerate G4BL and run MC+reco against each change?
----- Does this have to go with the real-data reconstruction on the 24 hour timescale? Means we'll have to know/flag which runs to simulate?

--> What is the level of access expected for the MC batch output?
--- all MC output available via http? [ Imperial/Brunel] Or leave the bulk on it tape and have only limited [ which? ] data on disk for open access?

--> In addition to the matrix, what are the MC configurations we want to simulate? 
---- How many configurations? How much for each? Presumably some can take the same beam input, while others require specific beam generation?