The Signal
Preprint proposes code quilts for logical measurements without extra qubits
A new preprint describes code quilts, a way to measure joint logical operators across quantum-code blocks without adding ancillary qubits. Its decoding tests report logical error probabilities comparable to idle memory in selected code circuits.

Logical measurements without added qubits
A new preprint describes a way to measure joint logical operators across quantum-code blocks without adding ancillary qubits to the hardware. In the paper submitted to arXiv on November 19, Stasiu Wolanski calls the construction a code quilt and presents circuits that use measurement qubits already involved in routine syndrome extraction.
The technique targets logical Pauli-product measurements: operations that measure a joint logical operator spanning two or more code blocks. The construction is based on a subsystem code, whose gauge operators need not commute. Wolanski gives conditions under which the measurement preserves code-capacity distance, a measure of protection against errors in the code-capacity setting.
Two protocol variants
The paper sets out two versions. In the first, the subsystem code’s gauge generators are measured directly. The paper says this variant is not fault tolerant in the large-distance limit, though its circuit implementation increases the number of interactions per physical qubit by at most two.
The second version gauge-fixes the subsystem code into a stabiliser code and measures its generators. Neither protocol requires an ancillary data-qubit system or measurement qubits beyond those used for idle syndrome extraction.
Measurements use d syndrome-extraction rounds, where d is the code distance. For the subsystem protocol, a measurement round is generally one to three timesteps longer than an idle-memory round, depending on the logical operator and the number of blocks involved. The paper notes that the operations are not generally single-shot.
Code tests and decoding
The empirical study applied the construction to bivariate bicycle and surface codes. It found code-capacity-distance-preserving quilts for 25,000 randomly selected logical operators, then optimised detector bases and examined selected circuits for distance preservation. The paper reports tests across bivariate bicycle codes with 72, 144 and 288 physical qubits, and, in some cases, matched-distance rotated surface codes.
Decoding experiments covered a subset of the circuits. For operations involving a 144-qubit bivariate bicycle code and a distance-12 rotated surface code, the paper reports logical error probabilities comparable to idle memory at physical error rates between 1.5 × 10⁻³ and 3 × 10⁻³. Those findings concern the tested circuits and conditions; they do not establish the second protocol’s fault tolerance in general.
Limits and next steps
Its other variant has better asymptotic properties, but its general fault tolerance remains unproven. The author calls for testing at larger distances and says the study’s optimisation methods are proofs of concept.
The paper describes the quilt construction as suitable for measuring a single logical operator. It says further work would be needed to broaden the range of operators or combine the method with a code that has plentiful automorphisms for use in an architecture with fixed connectivity.
Sources
- Original source · Preprint Stitching Quantum Codes into Quilts: Universal Logic with no Ancillary Qubits arXiv:2611.26329
Related stories
- Infleqtion reports 30 entangled logical qubits on Sqale
- ETH Zurich team teleports logical states on a 41-qubit circuit
- Alice & Bob targets 200 logical qubits as soon as 2030
- Preprint says Rapids decodes distance-19 code in under a microsecond per round
- Preprint details FPGA platform for testing QEC decoders