Quantum Insdel Errors: Sync & Correction

• Quantum insdel errors are synchronization errors that alter qubit count and order by modifying Hilbert space dimensions, complicating syndrome extraction and recovery.
• Synchronization-string methods attach classical indices to quantum data, effectively converting insdel noise into standard corruption and erasure errors for correction.
• Permutation-invariant codes and angular-momentum recovery protocols demonstrate practical strategies for directly correcting insertion and deletion errors in quantum registers.

Searching arXiv for recent and foundational work on quantum insertion–deletion errors and related insdel coding theory. arXiv search query: "quantum insertion deletion channels permutation invariant insertion errors" Quantum insertion–deletion (“insdel”) errors are synchronization errors that change the number and positions of qubits or qudits in a quantum register. Unlike Pauli, depolarizing, or erasure models on a fixed tensor-product space, insdel noise changes Hilbert-space dimension, destroys positional alignment, and therefore obstructs syndrome extraction unless synchronization is restored. The current theory is organized around three complementary viewpoints: a channel model that reduces quantum insdel noise to standard corruption/erasure noise by attaching classical synchronization information (Leahy et al., 2019); direct quantum error-correction conditions for insertion, deletion, and mixed insdel errors on permutation-invariant (PI) codes (Bulled et al., 9 Feb 2026); and explicit recovery protocols for single insertion errors on gapped PI codes using angular-momentum measurements (Bulled et al., 3 Sep 2025). Classical insdel coding theory remains an essential benchmark because many quantum schemes derive synchronization resilience from classical side information, and those classical ingredients obey sharp insdel bounds (Chen et al., 2021, Con et al., 2021, Kong et al., 2024).

1. Formal channel model and error semantics

A quantum insdel channel over a finite-dimensional Hilbert space $H$ is defined on $n$ input registers $A_1,\dots,A_n$ with an error budget $p+q\le n\delta$, where $p$ is the number of deletions and $q$ is the number of insertions. The channel is modeled as a CPTP map

$$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$

where $H_{\top}$ is a $(d+1)$-dimensional extension of $H$ carrying a special end-of-transmission symbol $n$0. The map factors as

$n$1

with $n$2, a subset $n$3 of correctly transmitted indices of size $n$4, and an injective, strictly monotonic function

$n$5

that preserves the relative order of the surviving qudits. The remaining output systems are arbitrary, produced by a CPTP map $n$6. Padding by $n$7 yields a fixed-length output in $n$8 (Leahy et al., 2019).

This model separates synchronization failure from ordinary corruption. Some input subsystems survive exactly, but only as an unknown subsequence embedded in a longer or shorter stream. Standard quantum error models preserve both the number of channel uses and their positions; quantum insdel channels preserve only an ordered subsequence of the original systems. A common misconception is that deletion noise is simply erasure noise without a flag. In the insdel model, the receiver does not know which positions survived, so the difficulty is not only loss of content but also loss of alignment.

The same literature introduces $n$9, another $A_1,\dots,A_n$0-dimensional extension of $A_1,\dots,A_n$1, with a distinguished erasure symbol $A_1,\dots,A_n$2. This becomes important after resynchronization, because the resulting fixed-position channel is naturally described in terms of corruptions and erasures rather than insertions and deletions (Leahy et al., 2019).

2. Reduction to corruption and erasure by synchronization strings

The first general framework for quantum insdel communication extends the classical synchronization-string method of Haeupler and Shahrasbi to the quantum setting. The idea is to append a classical synchronization label to each transmitted quantum system, measure only that classical label at the receiver, and then reorder the quantum payload according to the decoded indices. The quantum content is not measured during synchronization recovery; only the attached classical index or synchronization symbol is (Leahy et al., 2019).

In the simplest indexing scheme, the sender appends the exact position $A_1,\dots,A_n$3 to the $A_1,\dots,A_n$4-th quantum register through an isometry

$A_1,\dots,A_n$5

so that $A_1,\dots,A_n$6. If $A_1,\dots,A_n$7 deletions and $A_1,\dots,A_n$8 insertions occur, the reordered output on $A_1,\dots,A_n$9 fixed positions satisfies

$p+q\le n\delta$0

where $p+q\le n\delta$1 is the number of corrupted positions and $p+q\le n\delta$2 the number of erasures. With the convention “erasure $p+q\le n\delta$3 error” and “corruption $p+q\le n\delta$4 error,” the simulated channel has at most $p+q\le n\delta$5 half-errors (Leahy et al., 2019).

To avoid a channel alphabet that grows with $p+q\le n\delta$6, the exact index $p+q\le n\delta$7 is replaced by a symbol $p+q\le n\delta$8 from an $p+q\le n\delta$9-synchronization string over an alphabet of size $p$0. If the corresponding $p$1-indexing algorithm has

$p$2

misdecodings, then the final fixed-position quantum word differs from the sent codeword by at most

$p$3

half-errors. Thus an insdel channel of rate $p$4 is reduced to a standard quantum corruption/erasure channel with effective half-error rate

$p$5

at only constant-factor overhead in local dimension (Leahy et al., 2019).

For binary quantum insertion-deletion channels, the same paper gives a chunked construction with barriers and synchronization tags. It converts a qubit insdel channel of error rate $p$6 into

$p$7

corruption and erasure errors, covered by

$p$8

blocks of length

$p$9

in a computationally efficient way (Leahy et al., 2019). This establishes a one-way simulation theorem rather than a direct quantum insdel metric. It also clarifies a second common misconception: synchronization strings do not eliminate the need for quantum error correction; they convert synchronization failure into a noise model that standard QECCs can already handle.

3. Classical insdel distance as a benchmark for quantum synchronization layers

Classical insdel theory supplies the sharpest known outer benchmarks for any quantum architecture that derives synchronization resilience from classical markers, index strings, or side information. For vectors $q$0, the insertion–deletion distance is characterized by the longest common subsequence (LCS) length $q$1: $q$2 For a linear code $q$3, the minimum insdel distance is

$q$4

and a code with distance $q$5 uniquely corrects any combination of insertions and deletions of total weight at most $q$6 and detects up to $q$7 insdel errors (Chen et al., 2021).

A fundamental upper bound is the improved Singleton-type bound

$q$8

for every $q$9 linear code over $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$0. This strengthens the earlier Haeupler–Shahrasbi bound $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$1, extends the sharper $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$2 bound from Reed–Solomon codes to all linear codes, and removes the field-size restriction $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$3 that had appeared in earlier Reed–Solomon results (Chen et al., 2021).

Reed–Solomon codes can meet this classical benchmark. Over fields of size $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$4, there exist $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$5 Reed–Solomon codes that can decode from

$$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$6

adversarial insdel errors, attaining the half-Singleton bound. There is also a deterministic construction over fields of size

$$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$7

and for $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$8 an explicit $$N \in L\big(H^{\otimes n},(H_{\top})^{\otimes n+n\delta}\big),$$9 Reed–Solomon code over a field of size

$H_{\top}$0

correcting

$H_{\top}$1

insdel errors; any such $H_{\top}$2 construction must satisfy $H_{\top}$3 (Con et al., 2021).

At the non-linear level, the maximum size $H_{\top}$4 of a $H_{\top}$5-ary insdel code of block length $H_{\top}$6 and edit distance $H_{\top}$7 satisfies, for fixed $H_{\top}$8 and $H_{\top}$9,

$(d+1)$0

This combines a new LP-based sphere-packing upper bound with a hypergraph-matching lower bound (Kong et al., 2024). A plausible implication is that classical synchronization side channels embedded inside quantum protocols inherit a quantitatively rigid alphabet-dependent tradeoff, even when the quantum payload is protected by a separate QECC.

4. Direct quantum correction on permutation-invariant codes

A distinct research line treats insertion, deletion, and mixed insdel errors directly at the quantum-code level, without first reducing them to classical indexing. The setting is the $(d+1)$1-qubit symmetric subspace spanned by Dicke states

$(d+1)$2

and a single-qubit PI code with logical basis

$(d+1)$3

where the coefficient vectors satisfy the orthonormality conditions

$(d+1)$4

Permutation invariance makes all qubit locations equivalent and converts positional uncertainty into algebraic constraints on Dicke weights (Bulled et al., 9 Feb 2026).

Deletion and insertion are modeled as channels between Hilbert spaces of different sizes. The $(d+1)$5-deletion channel is

$(d+1)$6

while the $(d+1)$7-insertion channel is

$(d+1)$8

with arbitrary inserted $(d+1)$9-qubit pure state $H$0, insertion structure $H$1, and permutation $H$2 placing the inserted qubits into the prescribed slots (Bulled et al., 9 Feb 2026).

For PI codes, the paper proves a quantum insertion–deletion equivalence in this restricted setting. Theorem 1 gives necessary and sufficient Knill–Laflamme-type conditions for $H$3-insertion correction, expressed as the combinatorial constraints (C1) and (C2) on the coefficient vectors $H$4 and $H$5. These conditions are equivalent to the $H$6-deletion conditions of Aydin et al. for all $H$7, and the exposition explicitly states that, on PI codes, “if a PI code is $H$8-deletion error-correctable, then it is also $H$9-insertion error-correctable, and vice versa” (Bulled et al., 9 Feb 2026).

The mixed insdel problem is more restrictive. The key commutation structure is

$n$00

which decomposes a full insertion-then-deletion channel into a convex combination of semi-insdel channels indexed by the number $n$01 of inserted qubits that are later deleted (Bulled et al., 9 Feb 2026). From this, Theorem 2 derives necessary and sufficient full-insdel conditions (C5) and (C6) for $n$02-insdel error correction on PI codes. These conditions are strictly more restrictive than pure insertion or pure deletion conditions because they must hold simultaneously for all effective pairs $n$03.

This result resolves a longstanding question only in the PI setting. It does not establish a universal quantum insertion–deletion equivalence for arbitrary code families, and the paper explicitly identifies non-PI generalization as open.

5. Angular-momentum recovery for single quantum insertion errors

A more operational development gives an explicit correction protocol for single insertion errors on gapped PI “gnu” codes. These are PI codes on

$n$04

qubits with logical states

$n$05

so the Dicke weights are supported only on multiples of the gap parameter $n$06 (Bulled et al., 3 Sep 2025).

A single insertion error appends an arbitrary qubit

$n$07

at an unknown position $n$08, giving

$n$09

The protocol uses total angular momentum. After insertion, coupling a spin-$n$10 symmetric state with a spin-$n$11 qubit yields total spin

$n$12

The first syndrome measurement is $n$13, projecting onto one of these two irreducible sectors. The second measures $n$14 modulo the code gap $n$15, through projectors onto subspaces where $n$16 (Bulled et al., 3 Sep 2025).

Because of the gapped support of gnu codes, only four syndrome pairs occur: $n$17 The measurements therefore yield a two-bit syndrome. A key norm-preserving lemma states that for fixed $n$18, the squared norms of the projected logical basis states are equal, so after normalization the post-measurement state has the form

$n$19

Thus the logical qubit survives inside a new two-dimensional codespace determined by the syndrome and the insertion position (Bulled et al., 3 Sep 2025).

Recovery is completed by a teleportation protocol: an ancilla PI code block is prepared in logical $n$20, a logical CNOT is performed between the ancilla and the post-syndrome state, the latter is measured in its logical $n$21-basis, and a conditional logical $n$22 returns the encoded state to a PI code on the desired number of qubits (Bulled et al., 3 Sep 2025). This is not a full insdel code. It corrects arbitrary single insertion errors on a specific PI family and therefore complements, rather than subsumes, the channel-reduction framework and the abstract PI Knill–Laflamme conditions.

6. Scope, limitations, and open problems

Several boundaries of the present theory are already explicit in the literature. The synchronization-string construction addresses one-way quantum communication; interactive quantum communication under insdel noise is not treated (Leahy et al., 2019). The PI equivalence results do not imply a general insertion–deletion equivalence beyond permutation-invariant codes, and the existence of non-PI deletion codes leaves open whether analogous insertion and full-insdel criteria exist in broader families (Bulled et al., 9 Feb 2026).

Parameter optimization also remains incomplete. The synchronization-string reduction gives an effective half-error rate

$n$23

and the qubit construction produces

$n$24

corruption/erasure errors, but these bounds are explicitly described as not optimized (Leahy et al., 2019). The angular-momentum protocol is exact at the level of ideal measurements, but fault-tolerant implementations and extensions to multiple insertions are still open (Bulled et al., 3 Sep 2025).

A further unresolved issue is the relation between classical and genuinely quantum insdel metrics. Classical linear codes satisfy $n$25 (Chen et al., 2021), and classical Reed–Solomon constructions attain the corresponding half-Singleton radius $n$26 (Con et al., 2021). This suggests that any quantum construction whose synchronization layer is classical cannot surpass those classical limits on the marker subsystem. A plausible implication is that future quantum insdel Singleton-type bounds, if formulated intrinsically rather than via reduction, will reflect the same redundancy scaling, modified by genuinely quantum constraints.

The field therefore separates into three partially connected problems: defining robust intrinsic quantum insdel error criteria; reducing insdel noise to standard quantum noise with minimal overhead; and constructing explicit codes with efficient decoders outside the PI setting. The existing results show that all three are tractable in special regimes, but they have not yet converged to a single universal framework for quantum synchronization error correction.