REVIEW
R\'enyi multi-entropy of graph states and the $n -> 1$ limit
Quantum Physics · Submitted 19 Oct 2026 · 35 references
Pith's verdict: Novel results on Rényi multi-entropy for graph states, but the proof of Gauss sum positivity has a real gap in characteristic 2.
T0 · AI review
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
For qubit graph states in the collision-controlled class, the Rényi multi-entropy at odd replica numbers admits a unique polynomial continuation to n=1, but this continuation is not an entanglement monotone—it can increase under local…
The reading
What carries the argument
The key machinery is the reduction of the replica partition function to Gauss sums over F₂, which for odd n become $2^{{-r_G(n)/2}}$ where r_G(n) is the rank of a large adjacency matrix Γ. By Fourier transforming the replica lattice, Γ block-diagonalizes into N×N matrices Υ(t) weighted by differences of roots of unity. The rank sum r_G(n) = ∑_t rank Υ(t) depends only on the collision partition of the momenta t for collision-controlled graphs. The collision-controlled class is identified through a matching/Pfaffian criterion: if the maximum matching of the subgraph G_π (edges between parties with distinct momenta) is the unique perfect matching on its vertices (matching-controlled), or is obtained from such a graph by local complementation, then the rank depends only on the partition. The polynomial P_G(n) is then built from counting the number of momentum assignments with each collision partition.
Extended reading notes
Core claim
The central discovery is that for qubit graph states in the collision-controlled class, the Rényi multi-entropy at odd replica numbers n is determined by a polynomial P_G(x) in x, and the limit as n→1 of (log 2)/2 times P_G'(1) gives a closed-form expression for the multi-entropy. However, this limit can increase under local Pauli measurements, violating a basic requirement for an entanglement measure. Therefore, the unique polynomial continuation from odd integers to n=1 does not yield an entanglement monotone.
Load-bearing premise
The proof that the Gauss sum never vanishes and is positive for all odd n relies on the unsubstantiated claim that a nonzero Hermitian form over a finite field always has a vector with a^T M \bar{a} ≠ 0 (Lemma 1 in Appendix A.3); although the authors state the claim is true, they provide no reference or detailed proof, and the entire positivity of the rank formula depends on it.
Editorial extensions
If this is right
- The multi-entropy of any graph state without even cycles (e.g., any forest or odd cycle) with one qubit per party is given by a simple combinatorial formula: for a forest it is m log 2 for m edges; for an odd cycle C_{2r+1} it is (r+1) log 2.
- For complete bipartite graphs K_{m1,m2}, the multi-entropy is (m1+m2-1) log 2.
- The failure of monotonicity for n→1 means that the natural polynomial continuation cannot be used as a measure of multipartite entanglement in general quantum information tasks.
- The uniqueness of the continuation (by Carlson’s theorem) within a growth class implies that any alternative continuation that does give monotonicity must lie outside that growth class, which may be unphysical.
- The results provide testable predictions for holographic duals: for instance, the soap-film area for a boundary with the topology of a tree or odd cycle should match these combinatorial numbers.
Reading between the lines
- The non-monotonicity of the polynomial continuation suggests that the holographic soap-film prescription, if valid, must define the multi-entropy through a different analytic structure, possibly related to a replica symmetry breaking or a different averaging over replica indices.
- The definition of collision control via matching properties may generalize to qudit graph states with prime local dimension, where the field is F_p and the condition for polynomial continuation would involve p-adic properties of the replica number.
- More broadly, the result indicates that any attempt to define a multipartite entanglement measure from replica tricks must confront the fact that integer-replica data alone cannot uniquely determine a monotonic n→1 limit; some physical principle (e.g., strong subadditivity or holographic consistency) is needed to select the correct continuation.
Original abstract
R\'enyi multi-entropy extends the R\'enyi entropy to pure states of more than two parties. It is defined by a replica contraction at integer replica number $n$, and the multi-entropy is its $n\to1$ limit. In holography this limit is conjectured to compute the area of a minimal soap-film web in the bulk. Understanding multi-entropy for general states, however, is harder. Unlike the bipartite case, the replica contraction is not a moment of a density matrix, and the continuation to $n=1$ has to be found by other means. We study this problem for qubit graph states. We show that at every odd $n$ the contraction equals $2^{-k/2}$, where $k$ is a sum of ranks of matrices over a finite field, one row and one column per qubit. We define a class of graphs that we call collision-controlled, for which the rank sum is a polynomial in $n$. This class includes graphs without even cycles and all complete bipartite graphs. The natural polynomial $n \to 1$ continuation is a finite sum over partitions of the parties. While the R\'enyi multi-entropy is non-increasing under single-qubit Pauli measurements at every odd integer $n>1$, the continuation to $n=1$ is not. A $Z$ measurement on one qubit of an odd ring with $2r + 1 \geq 7$ qubits raises it from $(r+1)\log 2$ to $(2r-1) \log 2$. Thus the continuation is not an entanglement monotone.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
R\'enyi multi-entropy extends the R\'enyi entropy to pure states of more than two parties. It is defined by a replica contraction at integer replica number $n$, and the multi-entropy is its $n\to1$ limit. In holography this limit is conjectured to compute the area of a minimal soap-film web in the bulk. Understanding multi-entropy for general states, however, is harder. Unlike the bipartite case, the replica contraction is not a moment of a density matrix, and the continuation to $n=1$ has to be found by other means. We study this problem for qubit graph states. We show that at every odd $n$ the contraction equals $2^{-k/2}$, where $k$ is a sum of ranks of matrices over a finite field, one row and one column per qubit. We define a class of graphs that we call collision-controlled, for which the rank sum is a polynomial in $n$. This class includes graphs without even cycles and all complete bipartite graphs. The natural polynomial $n \to 1$ continuation is a finite sum over partitions of the parties. While the R\'enyi multi-entropy is non-increasing under single-qubit Pauli measurements at every odd integer $n>1$, the continuation to $n=1$ is not. A $Z$ measurement on one qubit of an odd ring with $2r + 1 \geq 7$ qubits raises it from $(r+1)\log 2$ to $(2r-1) \log 2$. Thus the continuation is not an entanglement monotone.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | R\'enyi multi-entropy extends the R\'enyi entropy to pure states of more than two parties. | background 1.00 | 0.06 | 0.03 | asserted 1.00 |
| s1 | It is defined by a replica contraction at integer replica number $n$, and the multi-entropy is its $n\to1$ limit. | background 1.00 | 0.09 | 0.04 | asserted 1.00 |
| s2 | In holography this limit is conjectured to compute the area of a minimal soap-film web in the bulk. | background 0.99 | 0.04 | 0.04 | speculative 0.81 |
| s3 | Understanding multi-entropy for general states, however, is harder. | gap 0.83 | 0.04 | 0.02 | asserted 1.00 |
| s4 | Unlike the bipartite case, the replica contraction is not a moment of a density matrix, and the continuation to $n=1$ has to be found by other means. | gap 0.58 | 0.05 | 0.03 | asserted 1.00 |
| s5 | We study this problem for qubit graph states. | contribution 0.99 | 0.83 | 0.03 | asserted 0.89 |
| s6 | We show that at every odd $n$ the contraction equals $2^{-k/2}$, where $k$ is a sum of ranks of matrices over a finite field, one row and one column per qubit. | result 0.99 | 0.94 | 0.23 | asserted 1.00 |
| s7 | We define a class of graphs that we call collision-controlled, for which the rank sum is a polynomial in $n$. | contribution 0.90 | 0.93 | 0.07 | asserted 0.99 |
| s8 | This class includes graphs without even cycles and all complete bipartite graphs. | result 0.65 | 0.76 | 0.03 | asserted 1.00 |
| s9 | The natural polynomial $n \to 1$ continuation is a finite sum over partitions of the parties. | result 0.62 | 0.80 | 0.14 | asserted 1.00 |
| s10 | While the R\'enyi multi-entropy is non-increasing under single-qubit Pauli measurements at every odd integer $n>1$, the continuation to $n=1$ is not. | result 0.99 | 0.84 | 0.39 | asserted 1.00 |
| s11 | A $Z$ measurement on one qubit of an odd ring with $2r + 1 \geq 7$ qubits raises it from $(r+1)\log 2$ to $(2r-1) \log 2$. | result 1.00 | 0.86 | 0.16 | asserted 1.00 |
| s12 | Thus the continuation is not an entanglement monotone. | interpretation 0.96 | 0.85 | 0.29 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
Figures
Showing 2 of the paper's 5 figures. See all figures in the PDF ↗ Figures reproduced from arXiv:2610.27080 by the authors.
References
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.