Data-flow parallelism for high-energy and nuclear physics computing frameworks

To infer what happened, we must account for: Detector signals from real physics Details of the neutrino "beam" (timing, intensity, ) Description of the detector geometry (not just from the drawings) Detector calibration information (accounting for many imperfections) Environmental conditions (temperature, air pressure, magnetic fields, ...).Marshal all the above data at the right points and use it to identify physics interactions.Do that for all your data, filter out the junk, analyze what's left and publish results. Physics creates Data implies PhysicsReconstructing what really happened How is this expressed as a scientific workflow?