Stim emission

A selected P2 program can leave CUDA-Q Logical as standards-compatible Stim circuit text. This is an interchange path: Stim is the assembly language the realization is projected onto — explicit physical qubits, Clifford operations, and measurements — so you can inspect, diff, and archive the result, or feed it to any Stim-speaking tool. Emission consumes a P2 build; it never changes the program’s semantics, and it is not an execution or sampling service.

Emit from the command line

qlx-translate performs the projection. It ships as a console script with the cudaq-logical wheel, so an installed package is enough. This module holds one round of CSS syndrome extraction on the Steane [[7,1,3]] code, followed by a destructive data readout:

qlx-translate preview/logical/examples/mlir/stim_steane_memory.mlir \
  --fabric-to-stim

R 7 8 9
H 7 8 9
CX 7 0 7 1 7 2 7 3 8 0 8 1 8 4 8 5 9 0 9 2 9 4 9 6
H 7 8 9
R 10 11 12
CX 0 10 1 10 2 10 3 10 0 11 1 11 4 11 5 11 0 12 2 12 4 12 6 12
M 7 8 9
M 10 11 12
M 0 1 2 3 4 5 6

Thirteen carriers, all explicit. Data occupies 0–6; the X-type ancillas are 7–9 and the Z-type ancillas 10–12, matching the code’s partitions. Each CX line is one stabilizer row of hx or hz expanded into carrier pairs — the encoded cudaq.logical.extract_syndrome has become the physical circuit it stands for.

Note what is not there: the data carriers are never reset. The gadget takes its encoded patch as an entry argument, because emission projects a realization rather than preparing an encoded state. Only the ancillas are reset, and only because each is a single carrier.

examples/mlir/stim_memory.mlir is the degenerate companion — a bare one-carrier code, whose whole projection is R 0 then M 0.

Fail-closed boundaries

Emission refuses rather than approximates:

  • the module must pass native MLIR verification;

  • the root must be a legalized P2 entry gadget — protocols go through fabric-lower-protocols first;

  • external resource requests (fabric.resource_request) are unsupported — Stim text cannot name an incoming resource stream;

  • reset-based preparation is emitted only for a verified trivial one-carrier code (n=1, k=1, r=0); encoded preparation is rejected;

  • recursive Fabric call graphs are rejected.

Stim cannot express noise models, detectors, sampling, or decoding. The trimmed product does not include them, so no DEM or shot-level target exists to emit toward.