Pith. sign in

REVIEW 3 major objections 5 minor 34 references

A Cycle Walk for Sampling Measures on Spanning Forests for Redistricting

T0 review · 3 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The Cycle Walk, a Metropolis-corrected Markov chain on spanning forests, can sample balanced graph partitions for target measures that penalize compactness directly—no longer relying on spanning-tree counts as a proxy—at parameter values wh

desk verdict The Cycle Walk is a genuinely new sampler with real promise, but the paper as written mis-defines the target measure and the reported 2-Tree acceptance ratio does not match it; the efficiency claims cannot be trusted until that is fixed. read the letter →

arxiv 2509.08629 v2 pith:VELC3NZO submitted 2025-09-10 cs.SI math.PR

Daryl R. DeFord , Gregory Herschlag , Jonathan C. Mattingly This is my paper · ORCID
classification cs.SImath.PR MSC 60J2205C0568R10
keywords MarkovchainMonteCarlographpartitionsspanningforestsredistrictingcompactnessMetropolis–Hastingspopulationbalancecyclewalk
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proposes a new Markov chain, the Cycle Walk, for sampling probability measures on spanning forests that encode balanced graph partitions, with political redistricting as the motivating application. The chain alternates moves that re-randomize the spanning tree inside a fixed district and moves that add two boundary edges between adjacent districts to form a cycle, then remove two edges to produce a new balanced pair; both move types have computable reverse probabilities, so the walk can be Metropolized to any desired target measure. Numerical experiments on North Carolina, Connecticut, and synthetic lattices indicate the walk converges for target measures with much stronger penalties on the spanning-tree count term than existing forest-recombination samplers, including when compactness is penalized directly through the isoperimetric ratio. If these results hold, redistricting ensembles can be generated for distributions that state policy concerns directly, rather than through tree-count proxies.

What carries the argument

The central mechanism is the added-edge cycle move. In a spanning forest, a 1-Tree step picks a district, adds one non-tree edge inside it, and removes one edge of the resulting unique cycle, changing the tree but not the partition; it is reversible with respect to the tree-count-weighted measure and acts as internal randomisation. A 2-Tree step picks two adjacent districts, adds two boundary edges to create a unique cycle spanning both, and removes two edges subject to population tolerance, yielding a new balanced pair. The reverse probability is computable because forward and backward proposals differ only by a small set of edge choices, which makes Metropolis–Hastings acceptance exact. Th

What would settle it

The decisive check is to compute the maximum total-variation distance, over a wider set of observables such as county splits and district boundary graphs, between independent runs of the Cycle Walk at γ=0.7 with the isoperimetric term; if that distance does not decay to a small value after 10^8 proposals, the empirical convergence claim fails. A secondary check is wall-clock cost per effective sample versus Forest RECOM.

Watch

Extended reading notes

Core claim

The discovery the paper tries to establish is that a Metropolised walk on spanning forests—proposals formed by adding edges to make a cycle and removing edges—samples measures νγ(τ) ∝ exp(−γJ_tree(ξ_τ)−J(ξ_τ)) for γ much closer to 1 than existing forest-recombination chains. A 1-Tree step adds a missing edge inside one district and removes an edge from the unique cycle, changing the forest but not the partition, reversibly. A 2-Tree step joins two adjacent districts with two boundary edges, forms a unique cycle, and removes two edges under population tolerance to give a new balanced pair. Since proposal probabilities factor into edge weights and a few choices, reverse probabilities are compu

Load-bearing premise

The load-bearing premise is that convergence can be judged from the maximum total-variation distance between the ranked-order partisan marginals of chains started from different initial plans; if the chain is slow in some feature that those marginals do not see, or if the assumed per-step cost advantage over forest recombination does not hold, the paper's claim of efficient sampling is not established.

Editorial extensions

If this is right

  • Redistricting practitioners can target measures with an explicit isoperimetric compactness term, not only the spanning-tree-count proxy, at γ values where earlier tree-based samplers stall.
  • The mixture of internal and inter-district moves gives a tunable interpolation between local flip dynamics and full recombination, so the sampler can be adapted to the energy landscape of a specific measure.
  • Weighted edges can encode county-preservation or other structural preferences while keeping a known target measure, as demonstrated on South Carolina's 50-district senate maps.
  • Because the 1-Tree step preserves the partition, the walk can refresh internal tree structure at low cost, which numerically accelerates convergence of the 2-Tree step at a roughly 9:1 ratio.
  • The method's efficiency at γ near 1 suggests that exact-MCMC ensembles for uniform-on-partitions-plus-compactness may be reachable with modest compute, an object that has been effectively inaccessible.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the paper's convergence picture holds for the reported marginals, the same design should be tested on other observables such as county splits, racial or ethnic composition, and full boundary graphs before relying on it in legal settings; the paper's diagnostic is deliberately marginal-based.
  • The 1-Tree step's value suggests a general recipe: add cheap partial resampling of internal tree structure to any global balanced-proposal chain to improve Metropolis acceptance; this could be tested by ablating the internal-to-paired ratio.
  • The paper's Balanced Tree Up-Down Walk variant has a uniform stationary measure and may admit rigorous mixing bounds; a proof of rapid mixing there would transfer theoretical support to the Cycle Walk family.
  • The paper explicitly forgoes a formal per-step cost analysis; a direct wall-clock comparison of effective sample rates would settle whether the convergence advantage translates into computational savings over forest recombination.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes the Cycle Walk, a Metropolis--Hastings Markov chain on spanning forests of a graph, designed to sample balanced graph partitions for redistricting. The chain mixes 1-tree cycle moves (internal to a district) and 2-tree cycle moves (exchanging population between adjacent districts), with population balance enforced in the proposal mechanism. The stated target family νγ interpolates between the uniform measure on spanning forests and the uniform measure on partitions lifted to forests, and the paper claims that the Cycle Walk mixes efficiently for γ values and compactness score functions that are difficult for existing forest-RECOM methods. Numerical experiments on 4×4 grids, North Carolina, Connecticut, regular lattices, and South Carolina county-weighted graphs are used to support the efficiency claims; the authors also release a Julia implementation.

Significance. If the central claim holds, the Cycle Walk would be a useful contribution: it would allow MCMC sampling of redistricting measures that penalize policy-relevant compactness directly, rather than implicitly through spanning-tree counts, while retaining exact Metropolis--Hastings correction. The paper has several concrete strengths: validation against exact enumeration on a 4×4 grid, comparison with Metropolized Forest RECOM for γ=0 and γ=0.2, release of code and tests, and explicit discussion of the limitations of tree-count-based compactness. However, the target measure and the reported acceptance probability are mutually inconsistent as written, so the numerical results do not currently establish the claimed capability. The issues are specific and correctable, but they affect the central claim.

major comments (3)
  1. [§6, Eq. (11)] The family νγ is stated inconsistently. With J_tree = -log tree(ξ), Eq. (11) gives νγ(τ) ∝ tree(ξτ)^γ e^{-J(ξτ)}. At γ=1 the partition marginal is proportional to tree(ξ)π(ξ), not π(ξ) as required by (6) and as stated in §10.2.1 ('γ=1 ... uniform on partitions'). To interpolate between ν0 and ν1 as defined in (6), the exponent needs the opposite sign, e^{+γ J_tree - J}, or J_tree must be defined as +log tree. As written, the target measure described in the text is not the measure defined by Eq. (11).
  2. [§8, Eq. (10) vs Eq. (13)] Equation (10) contains no factor involving treeα(ξτ)/treeα(ξτ'). Substituting the stated target (13) into the Metropolis--Hastings ratio yields an additional factor of the form (treeα(ξτ)/treeα(ξτ'))^{γ} or its inverse, depending on the sign convention used to correct Eq. (11). Since the paper states that 'our rejection probabilities take on the equation reported in (10),' the implemented 2-tree acceptances are inconsistent with the stated νγ for γ≠0. Consequently, the γ-dependence displayed in Figures 5–8 cannot be attributed to sampling the measure νγ as defined, and a reader cannot reproduce the target distribution from the manuscript.
  3. [§10.2.1, Figs. 5–8] The central efficiency claim rests on convergence diagnostics that are proxies rather than full-distribution checks: maximum total-variation distance between 14 ranked-order marginals from four chains, and single-chain effective sample sizes. The paper also explicitly forgoes a formal cost analysis in §10.2.1. Even after the target-measure and acceptance-ratio issues are corrected, the experiments need to be re-run with the corrected chain before the headline conclusion—efficient sampling for γ near 1 and for non-tree compactness scores—can be considered established. The current numerical evidence does not isolate the stated νγ.
minor comments (5)
  1. [§4.1, §6, §10.1.2] Typos: 'alos' should be 'also' (§4.1), 'mesure' should be 'measure' (§6), 'intrest' should be 'interest' (§10.1.2).
  2. [§10.2.4 and Fig. 10] The text says 'We begin by sampling ν0 over a variety of grid sizes,' but the surrounding discussion and Figure 10 caption refer to γ=1 with isoperimetric weights c=0.3,0.5,0.7,0.9. Please reconcile this discrepancy.
  3. [References] Reference [2] is corrupted ('Micha/suppress l Derezi nski'); please fix the author name and title.
  4. [§11] The walk is called 'Up-Down Walk' in the first paragraph and 'Down-Up Walk' in the next sentence; please use consistent terminology.
  5. [Figures 5–6] The labels 'Linked Forest Recom' are used in multiple panels without explicitly distinguishing top/bottom rows; adding '(no Polsby-Popper)' and '(w/ Polsby-Popper)' directly in each panel would improve readability.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the Cycle Walk's target measures and MH acceptance ratios are independently specified; self-citations are baselines, not load-bearing. (An internal sign inconsistency in the definition of ν_γ is a correctness issue, not circularity.)

full rationale

The central derivation is not circular. The target measures ν_γ are defined explicitly in Eqs. (6), (11), and (13) in terms of tree counts, tree-weight products, and an independently chosen score J(ξ); they are not fitted to the sampler's output. The Metropolis–Hastings acceptance probabilities in Eqs. (9)–(10) follow from the standard detailed-balance ratio ν(τ')Q(τ',τ)/(ν(τ)Q(τ,τ')), with the proposal kernels defined geometrically by adding edges to form cycles and removing edges to restore a forest. No parameter is calibrated to make a 'prediction' match held-out data. The compactness weight c_γ is tuned in Appendix A to keep mean isoperimetric score roughly constant across γ; this affects only the choice of test measures, not the sampler's correctness or the convergence diagnostic. Self-citations to Forest RECOM [4,6] and multiscale single-node flip [12] are used as experimental baselines, and the sampler is additionally validated against exact enumeration on a 4×4 grid (Sec. 10.1.1), which is independent supporting evidence. No load-bearing argument reduces to a self-citation chain. The one serious concern is an internal algebraic inconsistency: Section 6 defines J_tree = −log tree and then writes ν_γ ∝ e^{−γJ_tree−J}, which equals tree^γ e^{−J}, whereas (6) and the stated γ=1 uniform-on-partitions behavior require ν_γ ∝ tree^{−γ} e^{−J}; Eq. (10) also omits the corresponding tree-count ratio for γ≠1. This would change the measure actually sampled for γ≠0, but it is not circular: the target measure is still independently specified, and the bug does not make the output equivalent to the input by construction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The ledger is small. The sampler's correctness depends on standard Markov chain theory and on the known invariance of the 1-tree cycle walk from the cited literature. The efficiency claims depend on hand-selected experimental parameters (mixture ratio, compactness weight calibration, county edge weights). No invented entities are introduced.

free parameters (3)
  • c_gamma (compactness weight) = linear fit through origin, slope about 0.35 (Appendix A)
    Calibrated so that the average isoperimetric ratio stays roughly constant as gamma varies; this is a tuning choice for the experiments, not a parameter of the algorithm.
  • kappa (mixing ratio of 1-tree to 2-tree moves) = 0.99 in the main NC experiments; varied in Section 10.2.2
    Chosen by hand; Section 10.2.2 shows convergence changes with this ratio, so headline numbers are tied to this choice.
  • county edge upweights for South Carolina = 1, 10, 40
    Hand-selected to illustrate weighted-graph behavior; not fit to data.
assumptions (4)
  • standard math Metropolis-Hastings convergence requires the proposal chain to be irreducible and aperiodic on the relevant state space.
    Invoked in Sections 3 and 5; no proof of irreducibility or aperiodicity for the Cycle Walk is given.
  • standard math The 1-Tree Cycle Walk is reversible with respect to the measure m_xi^(alpha) from the cited literature [31, 3].
    Used in Sections 4.1 and 8 to set P_1-tree = Q_1-tree without Metropolization.
  • domain assumption The population-balance tolerance and score functions define the support of the target measure.
    Used throughout Section 2 and in the 2-Tree Cycle Walk's restriction that removed edges must yield balanced trees.
  • domain assumption The graphs studied are finite, connected, undirected planar graphs with positive edge weights.
    Setup in Section 2; all experiments are on precinct graphs or square and triangular lattices.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Cycle Walk for Sampling Measures on Spanning Forests for Redistricting." pith.science (2026). https://pith.science/paper/VELC3NZO

@misc{pith2026250908629,
  author       = {Pith},
  title        = {Pith review of: A Cycle Walk for Sampling Measures on Spanning Forests for Redistricting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VELC3NZO}},
  note         = {Machine review of arXiv:2509.08629}
}
read the original abstract

We introduce the Cycle Walk, a new Markov chain Monte Carlo method for sampling distributions on balanced graph partitions, motivated by applications in political redistricting. The method operates on spanning forests and combines two types of updates: local "cycle" moves within districts and global moves that exchange population between adjacent districts while preserving balance constraints. This construction enables efficient Metropolis--Hastings correction while allowing proposals at multiple spatial scales. We show that the Cycle Walk naturally interpolates between existing approaches based on local updates and a class of global update methods derived from recombination (RECOM). Through a range of numerical experiments on synthetic graphs and real-world precinct data, we demonstrate that the Cycle Walk exhibits improved empirical convergence diagnostics for distributions that place weaker weight on spanning-tree counts, a regime that is challenging for existing methods. In particular, the algorithm remains effective when incorporating alternative compactness measures that more closely reflect policy-relevant criteria. These results suggest that the Cycle Walk provides a flexible and computationally efficient framework for sampling from a broader class of redistricting distributions than previously accessible with MCMC techniques.

Figures

Figures reproduced from arXiv: 2509.08629 by the authors.

Figure 1
Figure 1. Illustration of one step of the 1-Tree Cycle Walk [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the 2-Tree Cycle Walk [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 4
Figure 4. Comparison of the ranked ordered marginals from the most republican to the most democratic district over samples of ν0 and ν0.2 applied to North Carolina using the votes for the 2020 presidential election. The orange gives the marginals using samples generated by the cycle-walk algorithm and the blue using samples generated by the Metropolized Forest RECOM algorithm. course, it is possible that the state space has o… view at source ↗
Figures from the paper (9 more)
Figure 5
Figure 5. Figure 5: We examine the average error between ranked ordered marginals of the least to most Democratic districts under the Pres￾idential 2020 votes on 14 North Carolina congressional districts. The error is averaged across 4 independent runs as a function of the number of propo…
Figure 6
Figure 6. Figure 6: We display the range of effective sample rates across the 14 ranked marginal distributions of the partisan makeup (most Republican to most Democratic) under the 2020 Presidential vot￾ing data. For each value of γ considered, we display a box plot to show the variation …
Figure 7
Figure 7. Figure 7: Quantiles of compactness (left) and forests (right) as we tune the weight of the sum of isoperimetric scores in the mea￾sure. We have tuned the weights so that the average isoperimetric is roughly constant (as shown in the left plot). A further refine￾ment of the weigh…
Figure 8
Figure 8. Figure 8: The average error in the 14 Ranked Ordered Marginals Congressional distracting plans using the vote from the 2020 Pres￾idential race. The error is averaged across 4 independent chains. The plots show a range of ratios of 1- vs 2-Tree Cycle Walks. The plot is with γ = 0…
Figure 9
Figure 9. Figure 9: We examine the convergence properties of the Cycle￾Walk on 5 districts within Connecticut with γ = 1 and the weight on the sum of the isoperimetric ratio (as part of the measure) at 0.3, 0.5, 0.7, 0.9. We compare the results from [12] with an isoperimet￾ric weight of c…
Figure 10
Figure 10. Figure 10: We examine the convergence properties of the Cycle￾Walk on 5 districts with γ = 1 and the weight on the sum of the isoperimetric ratio (as part of the measure) at 0.3, 0.5, 0.7, 0.9. 10.3. The Structure and Effectiveness of Cycle Walk Proposals. We have already seen i…
Figure 11
Figure 11. Figure 11: The first three figures show the acceptance probability (median and quartiles) as a function of the relative population change of the proposed for three values of γ with the isoperimetric score turned on to target the measures presented in [PITH_FULL_IMAGE:figures/fu…
Figure 12
Figure 12. Figure 12: We show the histograms of the number of split coun￾ties in an ensemble of 50-district plans in South Carolina. We up-weight edges that have two nodes in the same county by a fac￾tor of 1, 10 and 40 to demonstrate the effect of county splitting on sampling from the wei…
Figure 13
Figure 13. Figure 13: We show the multiplicative weight on the sum of isoperimetric scores as a function of γ to arrive at the results shown in [PITH_FULL_IMAGE:figures/full_fig_p032_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 1 linked inside Pith

  1. [1]

    Common Cause, No

    Rucho v. Common Cause, No. 18-422, 588 U.S. (2019)

    2019

  2. [2]

    Domain sparsifica- tion of discrete distributions using entropic independence.LIPIcs, Volume 215, ITCS 2022, 215:5:1–5:23, 2022

    Nima Anari, Micha/suppress l Derezi´ nski, Thuy-Duong Vuong, and Elizabeth Yang. Domain sparsifica- tion of discrete distributions using entropic independence.LIPIcs, Volume 215, ITCS 2022, 215:5:1–5:23, 2022

    2022

  3. [3]

    Log-concave polynomials iv: approximate exchange, tight mixing times, and near-optimal sampling of forests

    Nima Anari, Kuikui Liu, Shayan Oveis Gharan, Cynthia Vinzant, and Thuy-Duong Vuong. Log-concave polynomials iv: approximate exchange, tight mixing times, and near-optimal sampling of forests. InProceedings of the 53rd Annual ACM SIGACT Symposium on The- ory of Computing, STOC 2021, page 408–420, New York, NY, USA, 2021. Association for Computing Machinery

    2021

  4. [4]

    Metropolized forest recombination for monte carlo sampling of graph partitions.SIAM Jour- nal on Applied Mathematics, 83(4):1366–1391, 2023

    Eric Autry, Daniel Carter, Gregory J Herschlag, Zach Hunter, and Jonathan C Mattingly. Metropolized forest recombination for monte carlo sampling of graph partitions.SIAM Jour- nal on Applied Mathematics, 83(4):1366–1391, 2023

    2023

  5. [5]

    Autry, Daniel Carter, Gregory Herschlag, Zach Hunter, and Jonathan C

    Eric A. Autry, Daniel Carter, Gregory Herschlag, Zach Hunter, and Jonathan C. Mattingly. Multi-scale merge-split markov chain monte carlo for redistricting. 2020. 4Seehttps://julialang.org/ 5For GerryChain seehttps://mggg.github.io/GerryChain/index.html 6Seehttps://networkx.org/ A CYCLE W ALK FOR SAMPLING SPANNING FOREST MEASURES 33

    2020

  6. [6]

    Metropolized multiscale forest recombination for redistricting.Multiscale Modeling & Simu- lation, 19(4):1885–1914, 2021

    Eric A Autry, Daniel Carter, Gregory J Herschlag, Zach Hunter, and Jonathan C Mattingly. Metropolized multiscale forest recombination for redistricting.Multiscale Modeling & Simu- lation, 19(4):1885–1914, 2021

    1914

  7. [7]

    Models of random spanning trees, 2024

    Eric Babson, Moon Duchin, Annina Iseli, Pietro Poggi-Corradini, Dylan Thurston, and Jamie Tucker-Foltz. Models of random spanning trees, 2024

    2024

  8. [8]

    Mattingly, and Robert Ravier

    Sachet Bangia, Christy Vaughn Graves, Gregory Herschlag, Han Sung Kang, Justin Luo, Jonathan C. Mattingly, and Robert Ravier. Redistricting: Drawing the line.arXiv, 1704.03360, 2017

    arXiv 2017

Show all 34 references
  1. [9]

    Spanning tree methods for sampling graph partitions, 2022

    Sarah Cannon, Moon Duchin, Dana Randall, and Parker Rule. Spanning tree methods for sampling graph partitions, 2022

    2022

  2. [10]

    Sampling balanced forests of grids in polynomial time, 2024

    Sarah Cannon, Wesley Pegden, and Jamie Tucker-Foltz. Sampling balanced forests of grids in polynomial time, 2024

    2024

  3. [11]

    On the complexity of sam- pling redistricting plans, 2023

    Moses Charikar, Paul Liu, Tianyu Liu, and Thuy-Duong Vuong. On the complexity of sam- pling redistricting plans, 2023

    2023

  4. [12]

    Mattingly

    Gabriel Chuang, Gregory Herschlag, and Jonathan C. Mattingly. Multiscale parallel temper- ing for fast sampling on redistricting plans, 2024

    2024

  5. [13]

    Colorado in Context: Congressional Redistricting and Competing Fairness Criteria in Col- orado.Journal of Computational Social Science, 5:180–226, May 2021

    Jeanne Clelland, Haley Colgate, Daryl DeFord, Beth Malmskog, and Flavia Sancier-Barbosa. Colorado in Context: Congressional Redistricting and Competing Fairness Criteria in Col- orado.Journal of Computational Social Science, 5:180–226, May 2021

    2021

  6. [14]

    Clelland, Nicholas Bossenbroek, Thomas Heckmaster, Adam Nelson, Peter Rock, and Jade VanAusdall

    Jeanne N. Clelland, Nicholas Bossenbroek, Thomas Heckmaster, Adam Nelson, Peter Rock, and Jade VanAusdall. Compactness statistics for spanning tree recombination, 2021

    2021

  7. [15]

    Cormen, Charles E

    Thomas H. Cormen, Charles E. Leiserson, Ronald L. Rivest, and Clifford Stein.Introduction to algorithms. MIT Press, Cambridge, MA, third edition, 2009

    2009

  8. [16]

    Redistricting reform in virginia: Districting criteria in context.Virginia Policy Review, 12(2):120–146, 2019

    Daryl DeFord and Moon Duchin. Redistricting reform in virginia: Districting criteria in context.Virginia Policy Review, 12(2):120–146, 2019

    2019

  9. [17]

    Recombination: A family of markov chains for redistricting.Harvard Data Science Review, 3(1), 2021

    Daryl DeFord, Moon Duchin, and Justin Solomon. Recombination: A family of markov chains for redistricting.Harvard Data Science Review, 3(1), 2021

    2021

  10. [18]

    Outlier analysis for pennsylvania congressional redistricting

    Moon Duchin. Outlier analysis for pennsylvania congressional redistricting. https://www.governor.pa.gov/wp-content/uploads/2018/02/md-report.pdf

    2018

  11. [19]

    Comparing Algorithms for Generating Ensembles to Detecting Ger- rymandering

    Jonas Eichenlaub. Comparing Algorithms for Generating Ensembles to Detecting Ger- rymandering. https://sites.duke.edu/quantifyinggerrymandering/2023/08/23/comparing- algorithms-for-generating-ensembles-to-detecting-gerrymandering/, 2023

    2023

  12. [20]

    A new automated redis- tricting simulator using Markov chain Monte Carlo.Work

    Benjamin Fifield, Michael Higgins, Kosuke Imai, and Alexander Tarr. A new automated redis- tricting simulator using Markov chain Monte Carlo.Work. Pap., Princeton Univ., Princeton, NJ, 2015

    2015

  13. [21]

    Automated redistrict- ing simulation using markov chain monte carlo.Journal of Computational and Graphical Statistics, 29(4):715–728, 2020

    Benjamin Fifield, Michael Higgins, Kosuke Imai, and Alexander Tarr. Automated redistrict- ing simulation using markov chain monte carlo.Journal of Computational and Graphical Statistics, 29(4):715–728, 2020

    2020

  14. [22]

    Marzuola, Jonathan C

    Greg Herschlag, Jeremy L. Marzuola, Jonathan C. Mattingly, and Andrew Sun. A spectral analysis of tree based markov chains for redistricting problems.Preprint, 2025

    2025

  15. [23]

    Quantifying gerrymandering in north carolina

    Gregory Herschlag, Han Sung Kang, Justin Luo, Christy Vaughn Graves, Sachet Bangia, Robert Ravier, and Jonathan C Mattingly. Quantifying gerrymandering in north carolina. Statistics and Public Policy, 7(1):30–38, 2020

    2020

  16. [24]

    Mattingly, Matthias Sachs, and Evan Wyse

    Gregory Herschlag, Jonathan C. Mattingly, Matthias Sachs, and Evan Wyse. Non-reversible markov chain monte carlo for sampling of districting maps, 2020

    2020

  17. [25]

    Kruskal, Jr

    Joseph B. Kruskal, Jr. On the shortest spanning subtree of a graph and the traveling salesman problem.Proc. Amer. Math. Soc., 7:48–50, 1956

    1956

  18. [26]

    J. C. Mattingly and C. Vaughn. Redistricting and the Will of the People.ArXiv e-prints, October 2014

    2014

  19. [27]

    Mattingly

    Jonathan C. Mattingly. Expert report for Common Cause v. Lewis.Common Cause v. Lewis, 2019

    2019

  20. [28]

    Mattingly

    Jonathan C. Mattingly. Expert report for Harper v. Hall/Moore.Harper v. Hall/Moore, 2021

    2021

  21. [29]

    Complexity and geometry of sampling con- nected graph partitions, 2019

    Elle Najt, Daryl DeFord, and Justin Solomon. Complexity and geometry of sampling con- nected graph partitions, 2019

    2019

  22. [30]

    Empirical sampling of connected graph parti- tions for redistricting.Physical Review E, 104(6):064130, 2021

    Elle Najt, Daryl DeFord, and Justin Solomon. Empirical sampling of connected graph parti- tions for redistricting.Physical Review E, 104(6):064130, 2021

    2021

  23. [31]

    Linking and cutting spanning trees.Algorithms, 11(4):53, 2018

    Lu´ ıs MS Russo, Andreia Sofia Teixeira, and Alexandre P Francisco. Linking and cutting spanning trees.Algorithms, 11(4):53, 2018. 34 A CYCLE W ALK FOR SAMPLING SPANNING FOREST MEASURES

    2018

  24. [32]

    Swendsen and Jian-Sheng Wang

    Robert H. Swendsen and Jian-Sheng Wang. Nonuniversal critical dynamics in monte carlo simulations.Phys. Rev. Lett., 58:86–88, Jan 1987

    1987

  25. [33]

    Generating random spanning trees more quickly than the cover time

    David Bruce Wilson. Generating random spanning trees more quickly than the cover time. InProceedings of the Twenty-Eighth Annual ACM Symposium on Theory of Computing - STOC ’96, pages 296–303, Philadelphia, Pennsylvania, United States, 1996. ACM Press

    1996

  26. [34]

    Mathematically quantifying non-responsiveness of the 2021 georgia congressional districting plan

    Zhanzhan Zhao, Cyrus Hettle, Swati Gupta, Jonathan Christopher Mattingly, Dana Randall, and Gregory Joseph Herschlag. Mathematically quantifying non-responsiveness of the 2021 georgia congressional districting plan. InEquity and Access in Algorithms, Mechanisms, and Optimizati...

    2021

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.