Quantum Error Correction's Qubit Tax
The mainstream path to useful quantum computers is the surface code: bundling many noisy physical qubits into a few reliable logical ones. The commonly cited cost is on the order of a thousand physical qubits per high-quality logical qubit — redundancy as a tax.
Quantum LDPC codes (QLDPC) promise to cut that tax by using sparse check matrices for far better encoding rates. But they have long been stuck at one hurdle: they proved they could *store* information efficiently, not *compute* on it. Whenever logical gates (via lattice surgery) were attempted, the qubit savings evaporated and gate depth grew with code distance d.
On August 25, Quantum Zeitgeist reported on work published May 20 in Nature Communications: Malcolm et al. at Photonic propose the SHYPS code family, claimed to be the first QLDPC codes with efficient logical gates. Chief Quantum Officer Stephanie Simmons stated that "efficient QLDPC logic is no longer a theoretical promise — it has been demonstrated."
The caveat up front: all results are circuit-level numerical simulations, not hardware demonstrations. But the numbers are worth examining.
What Are SHYPS Codes?
The name: Subsystem Hypergraph Product Simplex — a subsystem hypergraph product construction using classical simplex codes. For r≥3, the SHYPS(r) family gives codes with parameters [(2^r−1)², r², 2^(r−1)]. Two concrete members used in the paper: [49, 9, 4] and [225, 16, 8] — 49 physical qubits encoding 9 logical qubits, an encoding rate near 1/5, lavish compared to surface codes (which start at k=1).
The key weapon is symmetry: the code family's automorphism group is on the order of |GL_r(2)|², yielding 2^O(k) fault-tolerant logical generators, each with physical circuit depth of 1. The code's inherent permutation symmetries can be used directly as logic gates, acting in a single layer of circuit depth.
Why the Gates Are Cheap
The standard approach to logical gates is lattice surgery: merging code patches with repeated syndrome measurements, at depth O(md) — linear in code distance. SHYPS implements logical gates via code automorphisms and ZX duality, with a five-stage decomposition (D_Z–D_X–D_Z–D_X–D_Z) covering the full Clifford group at depth independent of code distance.
Concrete numbers from the paper: at d=20, a CNOT takes ~40 logical cycles with lattice surgery but ~4 with SHYPS. For an arbitrary worst-case Clifford on b·k logical qubits, SHYPS requires 4bk(1+o(1)) logical cycles versus 2bk + O(log²(bk)) for bare, unencoded qubits — the gap shrinks to a constant factor, nearly unheard of in coding theory.
Simulation Results
Two head-to-head comparisons under the same error model:
| SHYPS | Surface code baseline | Result | Qubit savings | |---|---|---|---| | [49, 9, 4] | [[81, 9, 3]] | Outperforms | 2x | | [225, 16, 8] | [[784, 16, 7]] | Matches | 3.5x |
A harder test: a random 18-logical-qubit Clifford across two [49,9,4] blocks, compiled into 63 depth-1 generators, run as a depth-126 full circuit (operator plus inverse). Logical error suppression was close to pure-storage conditions — computing is as stable as storing.
The decoder was no afterthought: small-window sliding BP+LSD in (2,1) and (3,1) configurations, approaching single-shot behavior, while the surface-code baseline used optimal MWPM decoding.
The Cold Water
The other side of the ledger:
- Lower pseudo-threshold: ~0.35% for [225,16,8] vs ~0.8% for the d=7 surface code baseline. SHYPS' error-rate curve only outperforms below ~0.05% physical error rates — it needs cleaner physical qubits to cash in its efficiency advantage.
- Connectivity demands: the efficiency arguments assume high-connectivity architectures. Photonic's own Entanglement First approach (silicon spin qubits with telecom-band photonic interconnects) targets this — but that hardware must deliver first.
- T gates excluded: the full Clifford group is cheap, but non-Clifford operations still use standard state injection.
- Simulation only: wiring, crosstalk, and decoder latency for distance-8 blocks with hundreds of physical qubits remain untested on real hardware.
- Paper: Computing Efficiently in QLDPC Codes, Malcolm et al., Nature Communications 17, 7286. 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 It Matters
The paper appeared in May; buzz picked up in late August with renewed coverage and Simmons arguing that "architectures and timelines will be affected." Earlier extrapolations of 20x qubit savings have been cited in coverage, with the company framing this as accelerating commercialization timelines.
The most valuable contribution may not be the qubit savings itself, but shifting code selection from storage-driven metrics to computation-driven metrics: logical clock speed now enters the first tier of evaluation criteria. Competing code efforts (IBM's bivariate bicycle codes, Google's surface code improvements, Amazon's LDPC variants) will likely have to compete on gate depth as well.
Code is open-sourced at github.com/PhotonicInc/ComputingEfficientlyInQLDPCCodes with simulation data and reproduction scripts.
References: