Quantum error correction's bill: the most expensive component of a quantum computer
The mainstream route to a useful quantum computer is the surface code: bundling many noisy physical qubits into a few quiet logical ones. The cost? Industry figures commonly cited put a single high-quality logical qubit on top of a thousand or more physical qubits. Redundancy is a tax.
Quantum LDPC codes (QLDPC) are seen as a tax cut: clever code constructions keep check matrices sparse, multiplying encoding efficiency. But the roadblock that has persisted for years is that these codes were only proven to "store," not to "compute". They save qubits for memory, but for logical gates (the lattice-surgery workflow), the savings evaporate and gate depth grows with code distance d.
A paper by Malcolm et al. at Canada's Photonic (published in Nature Communications; covered by Quantum Zeitgeist in late August) presents SHYPS codes, claimed as the first QLDPC code family with efficient logical gates as well. Chief Quantum Officer Stephanie Simmons put it bluntly: "Efficient QLDPC logic is no longer a theoretical promise — it is a demonstrated result."
Caveat up front: all results are circuit-level numerical simulations, not hardware. But the simulation ledger is worth reading line by line.
What SHYPS codes are
The name decomposes as Subsystem HYpergraph Product Simplex — a subsystem hypergraph product construction using classical simplex codes as ingredients. The family is parameterized as SHYPS(r); for r ≥ 3 it gives codes with parameters [(2^r−1)², r², 2^(r−1)]: n physical qubits encoding k logical qubits at distance d. The paper's canonical examples use two members: [49, 9, 4] and [225, 16, 8] — 49 physical qubits carrying 9 logical ones, an encoding rate near 1/5, which is lavish next to the surface code (where k starts at 1 and grows only by assembling modules).
The real weapon is symmetry. The family's automorphism group is on the order of |GL_r(2)|², translating into 2^O(k) fault-tolerant logical generators per block, each with physical circuit depth of only 1. In plain terms: the code's massive permutation symmetry can be used directly as logic gates, acting in a single circuit layer.
Why "computing" becomes cheap
The standard approach to logical gates is lattice surgery: patching code blocks together and repeatedly measuring stabilizers, with depth growing linearly in distance, O(md). SHYPS gates instead use code automorphisms and ZX duality, with a five-stage decomposition (D_Z–D_X–D_Z–D_X–D_Z) building the full Clifford group at depth independent of code distance.
How big is the difference? The paper's concrete numbers: at d = 20, a CNOT takes about 40 logical cycles via lattice surgery versus about 4 with SHYPS. For an arbitrary worst-case Clifford acting on b·k logical qubits, SHYPS needs 4bk(1+o(1)) logical cycles — compared with 2bk + O(log²(bk)) for unencoded bare qubits. The gap shrinks to a constant factor, which is nearly unheard of in the world of error-correcting codes.
Simulation results
Two head-to-head comparisons under the same noise model:
| SHYPS | Comparison surface code | Result | Qubit savings | |---|---|---|---| | [49, 9, 4] | [[81, 9, 3]] | Outperforms | 2x | | [225, 16, 8] | [[784, 16, 7]] | Matches performance | 3.5x |
A tougher test: a random 18-logical-qubit Clifford on two [49,9,4] blocks compiled into 63 depth-1 generators, run as a full depth-126 circuit (operator plus inverse). Logical error suppression was close to pure-storage levels. Translation: computing is as stable as storing.
The decoder effort was real too: a small-window sliding BP+LSD decoder in two configurations, (2,1) and (3,1), performing close to single-shot behavior — against a fully optimized MWPM decoder on the surface-code side.
The cold water is colder than expected
The other column of the ledger:
- Lower pseudo-threshold. [225,16,8] has a pseudo-threshold around 0.35%, versus roughly 0.8% for the d=7 surface-code comparison. SHYPS's error-rate curve only beats the surface code's slope once physical error rates fall below ~0.05%. Translation: it needs cleaner physical qubits to cash in its efficiency advantage.
- Connectivity hungry. The efficiency arguments assume high-connectivity architectures. Photonic's own Entanglement First approach (silicon spin qubits with telecom-band photonic interconnects) is aimed at this — but that hardware line must first deliver.
- T gates billed separately. The Clifford group is cheap; non-Clifford operations still use the old magic-state-injection bill.
- Simulation is simulation. Routing, crosstalk, and decoding latency on real hardware for distance-8 blocks of a few hundred physical qubits remain entirely untested.
- Paper: Computing Efficiently in QLDPC Codes, Malcolm et al., Nature Communications https://www.nature.com/articles/s41467-026-73061-9
- Quantum Zeitgeist coverage: https://quantumzeitgeist.com/photonic-code-cuts-qubit-needs-quantum/
- Open-source code: https://github.com/PhotonicInc/ComputingEfficientlyInQLDPCCodes
Why this suddenly got hot
The paper appeared in May; the buzz rose in late August when Quantum Zeitgeist re-covered it and Simmons began arguing that "architectures and timelines will be affected." Earlier coverage from the outlet cited an extrapolated 20x qubit-savings figure, and the company keeps steering toward an "accelerated commercialization timeline."
My take: the most valuable part of this work is not the qubit savings but that it shifts code selection from storage-driven to compute-driven metrics. Previous code benchmarks raced on threshold and encoding rate; SHYPS pulls "logical clock speed" into the top tier of metrics. Other code teams (IBM's bivariate bicycle codes, Google's surface-code improvements, Amazon's LDPC variants) will likely have to add gate depth to their scorecards — without it, no one can credibly claim practicality.
The code is open-sourced at github.com/PhotonicInc/ComputingEfficientlyInQLDPCCodes with simulation data and reproduction scripts included. Publishing the ammunition like this is a practice worth copying.
References: