Transform data

You can now read data into a FAST-HEP workflow. The next step is to derive new quantities and select the events and objects relevant to your analysis.

Most High Energy Physics datasets are naturally represented as columnar data. Rather than looping over events one-by-one, analyses operate on columns of quantities such as particle momenta, energies, and identification flags.

Many of these quantities are also jagged: one event may contain two muons, another five, and another none at all. FAST-HEP builds on the Awkward Array and NumPy ecosystems, allowing workflows to operate naturally on variable-length collections using familiar array-oriented expressions.

All available quantities can be referenced by name. As workflows grow, it is often useful to map dataset-specific branch names onto stable analysis-facing field names, or derive new variables with more meaningful names.

FAST-HEP also provides a number of convenience functions for working with irregular event data, including operations such as selecting leading objects, handling missing values, and accessing specific entries within variable-length collections.

The following tutorials introduce the most common transformation operations:

• deriving new quantities

• selecting objects and events

• mapping dataset fields