DATA ANALYSIS

Overview

For your experiment, you will want to think about how to judge your data while it is coming in and how to get your result publication worthy.

At XPP, we recommend to use ami for live monitoring of the data and (unless you are a returning user who has already his own setup) "Small Data" for near-real time production of data that can use used both for more in depth look at the data (adding up statistics, vary the event selection,...) as well as for getting the publication quality results. Except in rare cases, there is no need to run user code on "shared memory" servers. The beam line staff has much experience in leveraging ami for looking at live data and we build up statistics by using short runs and running scrips on hdf5 files produced at near real-time speed (the speed depends on the amount of data processing and the amount of data that is written out). The latter step is technically "offline" as the data is read from disk, rather than from shared memory.