TTrees

TTrees are the primary data container in the ROOT data analysis framework. They provide a columnar storage model optimized for large datasets, storing entries that typically correspond to events or observations. TTrees are designed to support efficient I/O, selective reading, and scalable analysis workflows common in high-energy physics and related fields.

A TTree organizes data into branches, each representing a variable or a collection of related variables. Leaves hold the actual data values and the structure can be simple scalars, vectors, or more complex objects. The split level controls how much of a composite object is stored as sub-branches; higher split levels enable finer granularity and potentially better compression and I/O performance.

TTrees are written to and read from a ROOT file (TFile). They use a basket mechanism and optional compression (like zlib or LZMA) to balance disk usage and speed. Only the branches required for a given analysis need to be loaded, enabling fast selective access in large datasets. ROOT also supports streaming and versioning of data when writing TTrees.

Common tools for working with TTrees include TTreeReader for row-wise iteration, TTreeFormula for evaluating expressions, and TChain to assemble a set of TTrees across files into a single dataset. Several code-generation helpers, such as MakeClass and MakeSelector, assist in creating analysis skeletons that interface with TTrees.

While highly suitable for large-scale data, TTrees are not ideal for tiny datasets where the overhead outweighs benefit. Users should plan on data layout and compression settings, understand memory management for complex objects, and consider the implications of object ownership and I/O caching in long analyses.