Ultra-High-Rate Quantum Codes

• Ultra-high-rate quantum codes are quantum error-correcting schemes designed to maximize the encoding rate (often ≥1/2) while balancing qubit overhead with practical error suppression.
• They employ diverse construction paradigms—algebraic, geometric, and hardware co-design on neutral-atom arrays—to achieve rates up to 0.86 in assisted models and constant asymptotic rates.
• Key challenges include trade-offs between finite-size code distance and rate, the need for advanced decoding strategies, and open problems in realistic noise modeling and hardware integration.

Searching arXiv for recent and foundational papers on ultra-high-rate quantum codes and related high-rate qLDPC constructions. Ultra-high-rate quantum codes are quantum error-correcting codes designed to maximize the encoding rate $R=k/n$ while retaining useful error suppression, often with $R$ comparable to or exceeding $1/2$, so that roughly two physical qubits encode one logical qubit (Zhao et al., 17 Apr 2026). In adjacent parts of the literature, closely related goals appear under the labels high-rate or constant-rate, including families with explicit asymptotic bounds such as $R \ge 13/72$ or $R=0.2$ (Breuckmann et al., 2020, Yi et al., 2024). Across these variants, the central problem is the same: reducing qubit overhead without sacrificing decodability, circuit-level performance, or hardware realizability.

1. Rate, distance, and the meaning of “ultra-high-rate”

The basic metric is the encoding rate $R=k/n$, with $[[n,k,d]]$ denoting a stabilizer code of block length $n$, dimension $k$, and distance $d$. For finite-size qLDPC codes, the pressure toward high rate is immediately constrained by the quantum Singleton bound,

$R$0

so pushing $R$1 above $R$2 forces modest distances at practical block lengths (Zhao et al., 17 Apr 2026). This is why most practical finite-size qLDPC codes studied before recent ultra-high-rate constructions operated around or below $R$3 when tuned for useful logical performance at $R$4 a few thousand (Zhao et al., 17 Apr 2026).

The literature does not use the term uniformly. In the reconfigurable-neutral-atom setting, ultra-high-rate denotes $R$5 and specifically targets finite-size codes such as $R$6 and $R$7 with overheads $R$8 (Zhao et al., 17 Apr 2026). In geometric qLDPC work, ultra-high-rate may instead mean non-vanishing constant rate, as in hyperbolic-manifold codes with $R$9 (Breuckmann et al., 2020). Other constructions push nominal rates much higher by altering the resource model: less-noisy or entanglement-assisted LDPC schemes reach rates around $1/2$0 at moderate length, while QSBC–QURC turbo constructions realize $1/2$1 and hence approach unity as the outer block length grows (Fujiwara et al., 2013, Chandra et al., 2019).

This suggests that the topic is best understood as a design regime rather than a single asymptotic class. Some families prioritize $1/2$2 at finite $1/2$3, some prioritize constant asymptotic rate with polynomial distance, and others trade standard stabilizer constraints for auxiliary resources.

2. Algebraic and geometric construction paradigms

A major branch of ultra-high-rate research remains within CSS qLDPC design. Hyperbicycle codes interpolate between hypergraph-product and generalized bicycle codes through a quasi-cyclic ansatz, preserve LDPC sparsity, and admit lower and upper bounds with $1/2$4, but the concrete examples in the original study remain in the moderate-rate regime, roughly $1/2$5–$1/2$6, rather than the $1/2$7 regime (Kovalev et al., 2012). They are historically important because they clarify how finite rate, bounded stabilizer weight, and cyclic structure can coexist, even when the resulting rates are not yet ultra-high in the narrowest modern sense.

More recent algebraic qLDPC work reaches substantially higher rates. A two-branch finite-field base over $1/2$8 combined with a circulant-permutation-matrix lift of degree $1/2$9 yields a $R \ge 13/72$0-regular CSS code with parameters $R \ge 13/72$1 and rate $R \ge 13/72$2; the Tanner graphs of $R \ge 13/72$3 and $R \ge 13/72$4 separately have girth $R \ge 13/72$5, and decoder experiments reported no failures in $R \ge 13/72$6 trials at $R \ge 13/72$7, with a finite-length transition near $R \ge 13/72$8 (Okada et al., 25 Jun 2026). In a different HGP direction, TGRE-hypergraph-product codes were constructed with exact asymptotic rate $R \ge 13/72$9, blocklength $R=0.2$0, logical count $R=0.2$1, distance scaling $R=0.2$2, and an observed code-capacity threshold around $R=0.2$3 under depolarizing noise with FDBP decoding (Yi et al., 2024).

Geometric constant-rate qLDPC families provide another axis of comparison. Hyperbolic $R=0.2$4-manifold codes based on the regular $R=0.2$5 tessellation achieve the explicit lower bound

$R=0.2$6

with polynomially growing distance and single-shot decoding behavior consistent with a phenomenological threshold around $R=0.2$7 under syndrome noise (Breuckmann et al., 2020). The earlier “Golden codes,” built from the regular $R=0.2$8 tessellation, give an explicit lower bound

$R=0.2$9

together with $R=k/n$0 and a local decoder that exploits the regular hypercube structure of the tessellation (Londe et al., 2017).

Family	Representative rate or parameters	Distinguishing feature
APM CSS on atom arrays	$R=k/n$1, $R=k/n$2	finite-size $R=k/n$3 with hardware co-design
Two-branch finite-field + CPM lift	$R=k/n$4, $R=k/n$5	same-type girth $R=k/n$6
TGRE-HGP	$R=k/n$7	constant-rate HGP variant with threshold around $R=k/n$8
Hyperbolic $R=k/n$9 family	$[[n,k,d]]$0	single-shot decodable constant-rate qLDPC
Golden codes	$[[n,k,d]]$1	regular $[[n,k,d]]$2 tessellation and local decoder

A persistent misconception is that high rate and asymptotic goodness are interchangeable. The constructions above show otherwise: some achieve $[[n,k,d]]$3 at finite size with modest distance, while others keep constant rate asymptotically but at smaller constants and different geometric or decoder trade-offs.

3. Hardware co-design on neutral-atom platforms

The most explicit recent finite-size ultra-high-rate program is the co-design of CSS qLDPC codes for reconfigurable neutral-atom arrays (Zhao et al., 17 Apr 2026). In that family, both $[[n,k,d]]$4 and $[[n,k,d]]$5 are assembled from $[[n,k,d]]$6 affine permutation matrix blocks arranged in a $[[n,k,d]]$7 block-circulant pattern, with three active block rows and twelve block columns. Each affine permutation matrix implements $[[n,k,d]]$8 with $[[n,k,d]]$9, data-qubit degree is approximately $n$0, ancillas participate in $n$1 two-qubit gates per syndrome round, and the search enforces girth $n$2 rather than girth $n$3 in order to preserve viable finite-size instances (Zhao et al., 17 Apr 2026).

The distinctive contribution is not only algebraic but architectural. The construction imposes orbit-commutation relative to a reference APM with large orbits, Chinese-Remainder-Theorem separability for $n$4 with $n$5, and a maximal abelian column subgroup that linearizes the column action. For $n$6, the relevant subgroup is cyclic $n$7; for $n$8, it is $n$9 (Zhao et al., 17 Apr 2026). Under these constraints, each inter-block transition compiles into only $k$0–$k$1 product moves rather than $k$2 generic moves, and each step requires at most one horizontal cyclic column shift and one vertical row swap or shift (Zhao et al., 17 Apr 2026).

On the hardware side, atoms are placed on a $k$3D grid with data atoms spaced $k$4 horizontally and vertically, ancillas $k$5 from associated data atoms, and Rydberg CZ gates assumed at $k$6 per parallel layer (Zhao et al., 17 Apr 2026). The syndrome-extraction schedule has CNOT depth $k$7 per basis under baseline coloration, total depth $k$8 for sequential $k$9, while Bell-pair-assisted extraction reduces this to about $d$0 (Zhao et al., 17 Apr 2026). The measured movement times per round are $d$1 ms and $d$2 ms with four AOD pairs for the $d$3 and $d$4 codes respectively, and $d$5 ms and $d$6 ms with two AOD pairs (Zhao et al., 17 Apr 2026).

These implementation constraints do not prevent strong logical performance. Under a circuit-level model with $d$7, the $d$8 code achieves a per-logical-per-round error rate of $d$9 and a block error per round of $R$00, corresponding to $R$01 and entry into the teraquop regime. The smaller $R$02 code reaches $R$03 per logical per round and $R$04 (Zhao et al., 17 Apr 2026).

A related neutral-atom strategy uses static long-range connectivity rather than dynamic rearrangement. La-cross hypergraph-product codes on static $R$05D arrays implement weight-$R$06 checks with depth-$R$07 syndrome rounds, and circuit-level simulations show that these codes outperform surface codes when the nearest-neighbor two-qubit gate error probability is below $R$08 under a range-dependent neutral-atom model; under range-independent errors, the most non-local instance improves below $R$09 (Pecorari et al., 2024). The contrast between the two neutral-atom programs is informative: one turns non-locality into shallow motion, the other into bounded long-range blockade gates.

4. Concatenated, subsystem, assisted, and specialized high-rate families

Ultra-high-rate design is not confined to LDPC constructions. Many-hypercube codes concatenate the $R$10 error-detecting code to produce

$R$11

with explicit instances $R$12 at $R$13 and $R$14 at $R$15 (Goto, 2024). They are not LDPC asymptotically, because check weights grow with level, but at practical sizes they deliver rates around $R$16 and $R$17 together with a level-by-level minimum-distance decoder whose bit-flip threshold is about $R$18, and a logical CNOT threshold around $R$19 in the stated circuit-level model (Goto, 2024). A subsystem variant based on concatenated $R$20 blocks replaces exponentially growing syndrome weight by constant measurement weight $R$21, with subsystem parameters

$R$22

rate $R$23, and an estimated bit-flip threshold around $R$24 using neural-network decoding (Nakai et al., 6 Oct 2025). A later optimization showed that mixed-level families such as $R$25 and $R$26 can outperform the lower-rate all-$R$27 versions at the same level, while level-$R$28 encoder overhead can be reduced by about $R$29 relative to the original design (Goto, 29 Nov 2025).

Concatenated quantum Hamming architectures give another high-rate route. For $R$30 levels of the $R$31 code,

$R$32

and a bidirectional hard-decision decoder raises the code-capacity threshold under independent bit-flip noise from approximately $R$33 to $R$34 while empirically preserving the full $R$35 distance scaling for at least $R$36 (Zhang et al., 14 Jan 2026).

Assisted formalisms push the nominal rate far higher by relaxing standard stabilizer constraints. In the less-noisy model, any binary $R$37 code can be converted into a quantum code with parameters $R$38, and an affine-geometry example gives $R$39 with rate $R$40 (Fujiwara et al., 2013). In the entanglement-assisted model, using one ebit yields the example $R$41 with rate $R$42 (Fujiwara et al., 2013). These constructions are ultra-high-rate in the most literal sense, but the auxiliary-resource assumptions place them in a distinct operational category.

Other specialized families target particular noise or coding goals. QSBC–QURC turbo codes realize

$R$43

for outer block lengths $R$44, giving supported rates $R$45, $R$46, and $R$47, and at $R$48 operate at distance $R$49 from the quantum hashing bound at QBER $R$50 (Chandra et al., 2019). Constant-excitation stabilizer codes immune to collective coherent $R$51 rotations were constructed with parameters

$R$52

so that the rate tends to $R$53 while retaining exact immunity to collective $R$54 rotations on the code space (Chang, 7 Mar 2025).

5. Decoding, single-shot operation, and fault-tolerant computation

In high-rate quantum coding, decoder design is often as decisive as the code family itself. The neutral-atom APM codes use a hierarchical three-tier decoder: T1 belief propagation on a sliding-window Tanner graph, T2 relay BP, and T3 exact integer-programming maximum-likelihood decoding on the rare tail (Zhao et al., 17 Apr 2026). For the $R$55 and $R$56 codes, the escalation rates are $R$57 and $R$58 respectively, and FPGA/GPU acceleration reduces the average per-round decoding work to about $R$59 ns and $R$60, well below the $R$61–$R$62 ms syndrome-extraction cycle (Zhao et al., 17 Apr 2026).

Single-shot decoding provides a complementary paradigm for constant-rate qLDPC families. In the hyperbolic-manifold codes derived from the $R$63 tessellation, parity-check redundancy supports both a cellular-automaton decoder and BP under noisy measurements, and the data are consistent with a threshold around $R$64 in the phenomenological model with syndrome noise, or above $R$65 for BP with perfect measurements (Breuckmann et al., 2020). The existence of such decoders is significant because constant rate alone is not enough; operationally useful high-rate codes must also avoid repeated measurement overhead and decoder bottlenecks.

A recurrent criticism of high-rate qLDPC is that they are better suited to memory than to computation. Several recent works directly address that objection. Batched protocols for arbitrary CSS qLDPC codes introduce batched single-shot syndrome extraction, batched code switching, and batched addressable Cliffords with constant per-block space-time overhead, assuming fast classical computation (Xu et al., 7 Oct 2025). In a near-term self-dual bivariate-bicycle implementation, the reported space-time cost per logical qubit per layer is about $R$66, compared with about $R$67 for a surface-code protocol and $R$68 for a low-rate BB protocol (Xu et al., 7 Oct 2025).

Addressable logic has also been demonstrated in a small 3D-local high-rate qLDPC family. Lift-connected surface codes scale as $R$69, and the $R$70 instance provides deterministic fault-tolerant circuits for the logical Clifford set together with magic-state preparation (Old et al., 13 Nov 2025). Although the asymptotic rate vanishes, the small-instance ratio $R$71 is high relative to comparable surface-code instances, and the reported circuit-level pseudothresholds for algorithmic FT lie in the range $R$72 to $R$73 (Old et al., 13 Nov 2025).

6. Performance frontiers, trade-offs, and open problems

The strongest present finite-size evidence for ultra-high-rate practicality comes from teraquop-scale memory estimates. The reconfigurable-neutral-atom APM family reaches $R$74 for the $R$75 code at $R$76 (Zhao et al., 17 Apr 2026). A different route concatenates algebraic quantum Reed–Solomon outer codes over the gross bivariate-bicycle inner code by treating each inner block as a single Galois qudit of dimension $R$77; at uniform $R$78 physical noise, the concatenated gross code enters the teraquop regime and crosses below the two-gross code in space overhead at $R$79 per logical qubit-round (Wills et al., 21 May 2026). In that architecture, list decoding, time-like RS protection against measurement errors, and spacetime error mixing compensate for the correlated logical failures inside each high-rate inner block (Wills et al., 21 May 2026).

The trade-offs are correspondingly sharp. Finite-size $R$80 families generally have modest distances, often in the low teens, because of the Singleton constraint (Zhao et al., 17 Apr 2026). Constant-rate qLDPC families with stronger asymptotic guarantees usually operate at lower rates, larger blocklengths, or both (Breuckmann et al., 2020, Londe et al., 2017). Some constructions keep bounded check weight but accept moderate rates, as in hyperbicycle and many bivariate-bicycle implementations; others keep high rate but pay with heavier checks, such as TGRE-HGP, where the maximum stabilizer weight scales as $R$81 (Kovalev et al., 2012, Yi et al., 2024). Assisted codes achieve the most aggressive nominal rates, but only by assuming phase-only ancillas, ebits, or analogous auxiliary structure (Fujiwara et al., 2013).

Several open problems recur across the literature. One is how to raise distance at similar rate and size; this is explicit in the neutral-atom APM work, where $R$82 is already close to the finite-size ceiling imposed by high rate (Zhao et al., 17 Apr 2026). A second is how to model realistic hardware noise beyond idealized depolarizing channels, including idling, atom loss, correlated faults, and movement-induced errors (Zhao et al., 17 Apr 2026). A third is certification: the rate-$R$83 CPM-lifted code reports an explicit upper bound $R$84, but no nontrivial lower bound is claimed, and symmetry-reduced logical-operator searches remain open (Okada et al., 25 Jun 2026). A fourth is computational overhead: many constant-overhead logical-gate results assume classical decoding latency far below the code cycle time (Xu et al., 7 Oct 2025).

Ultra-high-rate quantum codes therefore do not define a single architecture but a spectrum of strategies for compressing the physical-to-logical overhead of fault-tolerant quantum information. The current landscape ranges from finite-size $R$85 qLDPC codes co-designed with reconfigurable atom arrays, through constant-rate hyperbolic and product-code families, to concatenated, assisted, and specialized constructions that exploit nonlocality, auxiliary qubits, or tailored noise structure. What unifies them is the attempt to replace the conventional high-overhead paradigm with codes whose logical density is large enough to matter at experimentally relevant scales, while still admitting concrete syndrome extraction, decoding, and fault-tolerant control.