Execution with intent.
How lazy evaluation, caching, and parallel computation fit together in LazyAnalysis.
01 /
Compute what you need
Lazy evaluation keeps the computation tied to the outputs being requested. Execution behavior depends on the nodes and expressions involved.
02 /
Reuse useful work
Cached matrix results and generation tracking help avoid repeating work when the relevant data has not changed.
03 /
Let computation branch
Parallel execution and asynchronous work support eligible operations. Actual scheduling depends on the workflow and node implementation.