Skip to content

Architecture

The architecture section is for readers who will open the compiler’s source or extend a pass. It documents what runs where: each stage of the pipeline, the invariant that stage owns, and the neutral-atom hardware model the backend lowering targets. The goal is to make the compiler legible — to turn “Quon lowers typed source to OpenQASM 3 and neutral-atom schedules” into a map you can follow in the codebase.

Contributors and advanced users who want to read or modify the compiler — not just invoke it. The pages assume you are comfortable with the language (see the language guide) and want to understand what happens after the typechecker accepts a program.

  • Compiler internals — the pipeline from source text to backend artifact: parsing, typechecking, MLIR lowering, optimization passes, and target-specific emission, stage by stage with the file each lives in.
  • Neutral-atom model — the hardware model, the target schema, and the scheduling approach (interaction graphs, Misra–Gries edge-coloring, AOD movement planning, compaction) behind the neutral-atom backend.

The high-level pipeline is summarized in the compiler pipeline reference; this section goes inside it. For runnable end-to-end examples of the pipeline, see the cookbook.

→ Start with Compiler internals.