External runners

The pinefarm run is mainly a uniform interface to some Monte Carlo (and non) generator that are able to produce PineAPPL grids.

Internally the runners are managed through a class system, with a base class External, that defines the basic steps and attributes, while implementing the common actions.

Attributes:

• name: name of the dataset

• theory: identifier of the theory

• pdf: PDF used for the comparison

• timestamp, optional: the timestamp of the previous run, if rerunning an already present grid

Computed attributes:

• dest: folder used for all the output

• source: folder containing pinecard

• grid: path of the computed grid

• gridtmp: path used for auxiliary grid (removed at the end of the run)

Steps:

• install(): further install steps, needed for the runner (not needed if the runner available as a python package on PyPI)

• run(): compute the actual predictions

• generate_pineappl(): collect predictions into a PineAPPL grid

• results(): provide runner results on chosen PDF, for comparison with PineAPPL convolute ones

• annotate_versions(): collect versions of all the program used to compute the results (for reproducibility)

• postprocess(): apply any further step specified in postprocess file, save metadata, and compress the final grid