search

visit-dav / visit

VisIt - Visualization and Data Analysis for Mesh-based Scientific Data
https://visit.llnl.gov
BSD 3-Clause "New" or "Revised" License
412 stars 110 forks source link

recipe for visualizing averge + variance across an ensemble of different sim runs #19638

Open cyrush opened 2 weeks ago

cyrush commented 2 weeks ago

Is your feature request related to a problem?

LLNL user request.

meshes are different -- pos cmfe won't really work b/c cmfes are mesh to mesh, and there are an ensemble of meshes.

Are data binning or resampling good enough?

markcmiller86 commented 2 weeks ago

Can't you CMFE each run onto a common "base", giving each a different variable name (which are perhaps also members of an aggregate array) and the do an expression sum across all those new variables you have CMFE'd to the base?

cyrush commented 2 weeks ago

that's a great idea!

markcmiller86 commented 2 weeks ago

OTOH, if the mesh evolves so much differently that there does not exist a common base mesh that sufficiently overlaps the meshes at all the relevant timestates, then, yes, I think a databinning (obviously of appropriate bin count/resolution for relevant dimensions) which produces a result back on to the "original mesh" (you'd have to pick which timestate you'd want that to be), might be best alternative.