search

FDSI / Kickoff_Workshop

FDSI Kickoff Workshop (May 2018)
https://www.colorado.edu/events/cfdsi/
7 stars 3 forks source link

What are the greatest obstacles to extracting insight from experimental and computational data bases? #1

Open KennethEJansen opened 6 years ago

KennethEJansen commented 6 years ago

This workshop is bringing together a wide spectrum of researchers who either produce large databases, analyze them, or develop software to analyze them. This issue is meant to spawn discussions that identifies what is working vs. what needs are currently unmet and requiring further development.

amsteinb commented 6 years ago

I'll give a little push to get things going... I think we first have to distinguish what type of insight we are seeking. In the combustion community, there are still a lot of open questions regarding what are physically reasonable closure models. I believe this also to be true in many non-reacting problems. In such cases, insights might be sought (from physical experiments or DNS) that stress whether underlying model assumptions hold at particular condition. This involves a certain set of challenges. On the other hand, insights may also be sought regarding the capabilities/limitations of a simulation methodology to predict engineering metrics for a class of problem. The challenges to obtain insight here can be quite different. In other words, there are different challenges to obtain insights regarding models versus simulations. Are we interested in both? Predominantly one?

flagor commented 6 years ago

Often we spend a lot of time processing or digesting the data to calculate derived quantities of interest from the data. Certain derived quantities are the same ones that others are looking for. Is there a way to adaptively edit what is stored with the dataset so that users can upload a derived quantity so that others don't need to recreate this work? Can we also maintain a version control of these contributed data so that we can "roll back" versions if errors are found?