Pith. sign in

REVIEW 4 major objections 4 minor 75 references

Scalable Dynamic Optimal Transport via Distributed Linearized ADMM

T0 review · 4 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read This paper claims that dynamic optimal transport with near-zero densities becomes tractable by replacing gradient steps with an exact proximal mapping inside a distributed linearized ADMM.

desk verdict The algorithm is genuinely new and the numerical results are strong, but the unstated omission of M0≥0 in the discretized problem is a real gap that should be addressed before the claim of solving dynamic OT is accepted. read the letter →

arxiv 2607.26407 v1 pith:2LZTXMK6 submitted 2026-07-29 math.OC

Hari Dahal , Rongjie Lai , Yangyang Xu This is my paper · ORCID
classification math.OC MSC 90C2549M2765Y0590C06
keywords dynamicoptimaltransportBenamou–BrenierlinearizedADMMexactproximalmappingnear-zerodensitydistributedoptimizationdomaindecompositionstaggeredgrid
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 tackles two bottlenecks in computing dynamic optimal transport — the Benamou–Brenier problem of moving one density to another at least action: instability when source or target densities approach zero, and the heavy memory bill of storing the whole space–time solution. It proposes to reformulate the discretized problem with auxiliary averaged variables, so the transport energy's proximal mapping is available exactly, as the root of a cubic. Plugging that proximal map into a linearized ADMM removes the need for gradient steps whose step size would otherwise collapse as density vanishes. A second, distributed formulation splits the space–time grid among agents, cutting per-agent memory and running in parallel, with reported speedups above 6× on 12 CPU cores. A sympathetic reader would care because these are exactly the regimes — sparse or nearly empty source or target regions, fine grids, higher dimensions — where standard solvers stall or diverge.

What carries the argument

The engine is the exact proximal mapping of the dynamic transport energy F(ρ,m)=||m||²/(2ρ): given an input pair (ρ̂,m̂), it returns (0,0) when ρ̂≤0 and ||m̂||²≤−2γρ̂, and otherwise a unique positive root ρ of the cubic ρ³+(2γ−ρ̂)ρ²+(γ²−2γρ̂)ρ−γ||m̂||²/2−γ²ρ̂=0, with m=m̂ρ/(ρ+γ). This exact map replaces gradient-type updates, so no step size must absorb the local Lipschitz constant 1/ρ³. Around it, LADMM alternates that proximal step with an exact projection onto the linear continuity-equation constraint, which amounts to solving a Poisson (or Sylvester, in 1D) equation via the fast cosine transform; the projection keeps every iterate feasible and makes the primal residual near machine preci

What would settle it

Solve a small 1D example with a source density that is exactly zero on some cells, run LADMM to convergence, and inspect the raw density values at half-integer grid points: if any are negative and forcing them nonnegative changes the objective beyond the stopping tolerance, the method is solving a relaxed problem, not the original transport problem.

Watch

Extended reading notes

Core claim

The central claim is that the discretized dynamic OT problem can be reorganized into a form whose subproblem over the transport energy F(ρ,m)=||m||²/(2ρ) is exactly solvable: the proximal mapping of F at a given pair either returns (0,0) or the unique positive root of a cubic, with momentum rescaled accordingly. Wrapped in a linearized ADMM, this exact-prox step keeps iterates away from the blow-up caused by the energy's Hessian scaling like 1/ρ³, while an exact projection onto the continuity equation — a Poisson or Sylvester step computed with a fast cosine transform — keeps the mass-conservation constraint satisfied at every iteration. The paper reports that in 1D and 2D experiments where

Load-bearing premise

The load-bearing premise is that letting the discretized density values take negative values during the computation does not change the final answer; the paper assumes this relaxation is harmless without stating it.

Editorial extensions

If this is right

  • Problems with source or terminal densities that are exactly zero in places no longer force tiny step sizes; the paper's experiments show convergence with smallest densities around 10⁻¹⁰.
  • Per-agent memory drops because each agent stores only its slice of the space–time variables, making finer grids and higher dimensions more practical.
  • The distributed version gets a reported 6–7× speedup with 12 CPU cores, with communication limited to shared spatial boundaries.
  • Because a projection step is performed each iteration, mass conservation is never violated, and reported primal residuals are 10⁻¹¹ to 10⁻¹³.
  • The ergodic convergence guarantee O(1/K) of LADMM applies, so averaged iterates give a feasible point with a controlled objective gap.

Reading between the lines

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

  • The same exact-proximal trick should extend to other homogeneous transport costs and to mean-field planning problems where the same 1/ρ³ gradient blow-up appears; any cost of the form ||a||²/ρ yields a polynomial root-finding proximal step.
  • The projection step is the main communication bottleneck in the distributed scheme; replacing the cosine-transform solver with a multigrid or basis-free Poisson solver could extend the speedup curve to many more agents.
  • Stability near zero density is demonstrated empirically rather than by a rate; a quantitative bound relating iteration count to the minimum density would turn the stability claim into a theorem.
  • The discretized problem is solved without an explicit nonnegativity constraint on the staggered density, so a direct comparison against a nonnegative-constrained solve would show whether the reported transport paths are realizable by true densities.
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

4 major / 4 minor

Summary. The paper proposes a linearized ADMM for the discretized Benamou–Brenier dynamic optimal transport problem. The key idea is to introduce auxiliary averaged variables so that the subproblem involving the dynamic OT energy can be solved exactly by a proximal mapping, thereby avoiding gradient-based steps whose Lipschitz constants blow up when the density approaches zero. A distributed variant partitions the space–time variables along the first spatial dimension to reduce per-agent memory and enable parallelism. Numerical experiments in 1D and 2D compare the method with FISTA, ALM, and G-prox, including challenging near-zero density settings, and report speedups of roughly 6–7× when scaling from 1 to 12 agents. A convergence statement is imported from Gao–Zhang and Xu.

Significance. If the technical gaps are repaired, the exact-proximal reformulation is a genuinely useful idea for the vanishing-density regime, and the distributed implementation addresses a real memory bottleneck in dynamic OT. The numerical study is extensive and includes systematically chosen comparisons in one and two spatial dimensions. However, as written, the paper does not yet establish that the algorithm solves the stated dynamic OT problem: the nonnegativity constraint on the discrete density is missing, the proximal lemma contains a false uniqueness claim, the pseudocode has a sign inconsistency, and the convergence theorem is asserted without verifying its hypotheses.

major comments (4)
  1. [§2.1, Eqs. (2.11)–(2.13); §2.3, Eq. (2.16)] The discretized problem never imposes M0 ≥ 0. In the continuous problem (1.1), F enforces ρ ∈ R_+; the staggered array M0 is the discrete counterpart of ρ, so feasible M0 should be nonnegative. The set C in (2.10) is only the linear continuity equation, and the projection (2.16) is an unconstrained projection onto C. Lemma 2.1 guarantees nonnegativity only of the auxiliary variable Mbar0, not of M0. Thus the algorithm may converge to points with negative entries in M0 that are not densities for (1.1). The paper neither states this relaxation nor proves that every relaxed feasible point can be replaced by a nonnegative staggered density with the same objective. Because the lower objective values reported in Tables 4–6 may arise from this relaxed, nonphysical problem, the central claim that LADMM solves the dynamic OT problem is not established. Relatedly, f in (2.6) is undefined when Avg0
  2. [§2.3, Algorithm 2.1 line 3 and Eqs. (2.14),(2.17),(2.18)] There is a sign inconsistency in the auxiliary-variable update. With the augmented Lagrangian defined in (2.14) and the dual updates in (2.17), completing the square over Mbar0 with M0 fixed gives an input c + Λ^k/β_k to the proximal map, not c − Λ^k/β_k (and similarly for Mbar). Algorithm 2.1 line 3 writes minus signs. These minus signs are inconsistent with the gradient step (2.18), whose gradient is Avg0^*(Λ^k + β_k(c^k − Mbar0^{k+1})); with the correct plus sign that gradient vanishes at the proximal point, as expected in ADMM. As written, the algorithm is not the LADMM scheme analyzed in Theorem 2.4, and the reported numerical convergence cannot be attributed to the stated updates unless the code uses a different sign convention. The authors should correct the sign or clarify the convention.
  3. [Lemma 2.1 and Appendix A] The claim that ρ in (2.12) is “the unique real-number solution” of the cubic (A.7) is false. For example, with γ=1, ρ̂=10, m̂=0, the cubic is (ρ+1)^2(ρ−10)=0, which has real roots −1 and 10, while the correct proximal value is ρ=10. Non-uniqueness of real roots can also occur in the “otherwise” branch. Since Algorithm 2.1 requires solving this cubic, the paper must specify a deterministic selection rule (e.g., the unique nonnegative real root, or the largest real root) and prove that this root gives the proximal minimizer. As stated, the lemma is incorrect.
  4. [§2.3, Theorem 2.4] The convergence theorem is asserted by direct citation of [18, Thm 4.3] and [47, Thm 2.13] without verifying their hypotheses. In particular, the objective f in (2.6) is not a closed proper convex function unless its domain and extended-value convention are specified; Algorithm 2.1 as written does not match the scheme covered by those references because of the sign issue above; and the numerical schedule (4.1) uses β_k nonconstant, whereas Theorem 2.4 assumes constant β. The O(1/K) guarantee is therefore not established for the implemented method. The authors should either verify the hypotheses of the cited results, or prove convergence directly for the actual algorithm and parameter schedule.
minor comments (4)
  1. [Tables 1–7] The table formatting is garbled in the manuscript text: column separators are missing and several rows appear to have entries merged or omitted (e.g., Table 1, n1=4096). This makes the numerical claims difficult to verify and should be corrected.
  2. [§2.2, near Eq. (2.28)] Typo: “egivenvectors” should be “eigenvectors”.
  3. [Figure 9 caption] Typo: “dentity setting” should be “density setting”.
  4. [Eqs. (2.2)–(2.3)] The domain/codomain conventions for V_d and V in the definitions of Avg_d and Avg_d^* are terse. A brief statement that bJ_d indexes the staggered grid and how boundary entries are handled would improve readability.

Circularity Check

1 steps flagged · score 4.0 of 10

Central LADMM derivation is self-contained, but one experimental benchmark defines the reference optimal value as LADMM's own output, making that accuracy comparison circular by construction.

  1. fitted input called prediction [Section 4.2.3, paragraph following Table 4 (definition of obj*_num used in Table 5)]
    "We use the best objective values obtained by LADMM in the 7-th column of Table 4 as obj*_num to calculate |obj−obj*_num| and report it in Table 5"

    The reference 'optimal' objective obj*_num is not an independently computed optimum; it is taken from LADMM's own lowest objective in Table 4. Consequently, LADMM's reported |obj−obj*_num| in Table 5 is just the difference between two LADMM runs (one at tolerance 10^-8, one at 10^-4), not a distance to an external ground truth. The favorable accuracy comparison for LADMM in that table is forced by construction: the method is being measured against a yardstick defined by its own output. This does not invalidate the algorithmic derivation, but it makes the 'near-optimal' validation in Table 5 self-referential.

full rationale

The algorithmic derivation itself is largely self-contained. Lemma 2.1 is proved in the appendix; the projection subproblem is reduced to a Poisson/Sylvester equation solved by eigendecomposition; and the LADMM updates follow from standard variable splitting. The convergence guarantee in Theorem 2.4 is imported from [18] and [47]; [47] is co-authored by Xu, but it is a published, parameter-free convergence result and is independently available from [18], so it is real evidence rather than a circular self-citation. The penalty parameter beta0 is tuned by grid search, which is ordinary parameter fitting and not a prediction. The reported speedups are direct timing comparisons across agent counts. The one genuinely circular element is the experimental reference obj*_num: the paper defines the 'optimal objective value' as LADMM's own best objective and then measures LADMM's error against it, so that specific accuracy comparison reduces to self-consistency by construction. This is confined to Section 4.2.3 / Table 5 and does not compromise the core proximal/projection derivation. Separately, and as a correctness concern rather than circularity, the discretized problem (2.13) and projection (2.16) never enforce M0 >= 0, so the method may solve a relaxed problem with negative staggered densities that do not correspond to a valid density for the original Benamou-Brenier problem; this omission should be weighed in assessing the claim that the method solves dynamic OT as originally posed.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on two unstated assumptions: the positivity relaxation in the discretization and the applicability of an imported convergence theorem. The only fitted parameter identified is β0, tuned per experiment.

free parameters (1)
  • β0 (LADMM penalty parameter) = grid-selected per experiment; e.g., 1e-4 in distributed runs
    The algorithm's convergence speed and reported timings depend on β0; chosen by grid search over {1e-5,...,10} in Eq. (4.1).
assumptions (4)
  • domain assumption M0 is allowed to take negative values in (2.13) and (2.16); only the linear continuity constraint C is enforced.
    The original problem requires ρ ≥ 0; the paper never imposes M0 ≥ 0 or justifies the relaxation.
  • domain assumption The convergence theorem from [18,47] applies to Algorithm 2.1 with the non-globally-Lipschitz objective f.
    Theorem 2.4 is stated as a direct corollary; no verification of the hypotheses is given.
  • domain assumption The staggered-grid discretization (2.6)–(2.10) faithfully approximates the continuum problem (1.1).
    The discretization follows [50]; the paper uses it without a discretization-error analysis.
  • standard math The eigen-decomposition in (2.28) of the tridiagonal matrices A_d is orthonormal and can be applied via DCT.
    Used in (2.27)–(2.29) to solve the projection step.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scalable Dynamic Optimal Transport via Distributed Linearized ADMM." pith.science (2026). https://pith.science/paper/2LZTXMK6

@misc{pith2026260726407,
  author       = {Pith},
  title        = {Pith review of: Scalable Dynamic Optimal Transport via Distributed Linearized ADMM},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2LZTXMK6}},
  note         = {Machine review of arXiv:2607.26407}
}
read the original abstract

In this paper, we address two fundamental challenges in the numerical solution of dynamic opti- mal transport (OT) problems. The first challenge arises when the initial and/or terminal densities approach zero and no positive lower bound is available. In this regime, conventional methods may become unstable or computationally inefficient, since the Lipschitz constant of the objective can scale like the reciprocal of the cube of the density. As a result, near-zero regions may lead to slow convergence or even divergence. The second challenge concerns the substantial memory cost of the dynamic formulation, whose discretization over fine spatial and temporal grids requires storing vari- ables across the entire space-time domain. This storage burden quickly becomes prohibitive as the grid is refined or the spatial dimension increases. To overcome the first difficulty, we reformulate the classical discretized dynamic OT problem so that the resulting objective admits an exact proximal mapping. When combined with a linearized alternating direction method of multipliers (LADMM), this reformulation yields an efficient and robust algorithm that remains stable even in challenging settings where the density may vanish. To reduce the memory burden, we further introduce a dis- tributed formulation in which the optimization variables are partitioned across multiple agents. This design substantially lowers the storage requirement for each agent and can also lead to computational acceleration. We validate the proposed framework through numerical experiments in one- and two- dimensional spatial settings under varying levels of difficulty. The results demonstrate the stability, robustness, and scalability of the proposed method in comparison with conventional approaches.

Figures

Figures reproduced from arXiv: 2607.26407 by the authors.

Figure 1
Figure 1. Illustration of the decomposition of the time and spatial domain for D = 1. For better understanding of our notations, we use the case of D = 1 to explain them. The index sets J0 and J1 correspond to the filled red and blue triangles in [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the domain decomposition for the case D = 1 , N = 4, n0 = 5, n1 = 10, where the domain is vertically decomposed along the spatial dimension. According to our domain decomposition, each agent i holds the following variables: (M0)i ,Mi ,(M0)i ,Mi , Πi ,Λi (3.1) . 12 [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Illustration of partition of C, V , and ϕ for N = 4, and communication using blocks of C and V at agent 4. Partition of C and V is depicted in the first row, partition of ϕ is depicted in the second column of the second row, and communication using the last column block of C and last row block of V is depicted in the first column of the second row. the shared boundary. 3.2.2. Generalization to D > 1. When D > 1, bot… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Illustration of computation and communication using blocks of R and V ⊤ at agent 4 for N = 4. Partition of R and V ⊤ is depicted in the first row and communication using the last column block of R and last row block of V ⊤ is depicted in the second row. methods by “F”,…
Figure 5
Figure 5. Figure 5: Left: illustration of partition of C along the first spatial dimension for N = 4 and D = 2. Right: illustration of how the eige-matrices U, V , and W act on a block of C. that the dual residual decreases more rapidly than the primal residual in the beginning. The linea…
Figure 6
Figure 6. Figure 6: Objective difference vs. iteration number for ALM [5], G-prox [23], and LADMM with ρ0, ρ1 setting in (4.8), n1 ∈ {512, 1024, 2048, 4096}, and the number of time steps set to 64. 4.2.4. Two dimensional spatial domain. In this set of numerical experiments, we consider a …
Figure 7
Figure 7. Figure 7: Constraint violation vs. iteration number for ALM [5] and G-prox [23] with ρ0, ρ1 setting in (4.8), n1 ∈ {512, 1024, 2048, 4096}, and the number of time steps set to 64 [PITH_FULL_IMAGE:figures/full_fig_p021_7.png]
Figure 8
Figure 8. Figure 8: Evolution of images by LADMM with n0 = 20, n1 = n2 = 256, β0 = 10−4 at times t = 0, 0.2, 0.6, 0.8, 1 from left to right [PITH_FULL_IMAGE:figures/full_fig_p022_8.png]
Figure 9
Figure 9. Figure 9: Time taken vs. the number of agents by LADMM with dentity setting (4.8), N ∈ {1, 2, 4, 8, 10, 12}, β0 = 10−4 , and (n0, n1) = (1024, 32768) and (n0, n1) = (512, 65536) represented by blue and red respectively. t0 = 0 t1 = 0.2 t2 = 0.4 t3 = 0.6 t4 = 0.8 t5 = 1 5 10 15 2…
Figure 10
Figure 10. Figure 10: Evolution of images by LADMM with n0 = 20, n1 = n2 = 784, β0 = 10−4 at times t = 0, 0.2, 0.4, 0.6, 0.8, 1 from left to right. 15800 8518 4418 3116 3025 2921 Number of agents 1 2 4 8 10 12 Tim e (in s e c o n d s) 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 [P…
Figure 11
Figure 11. Figure 11: Time taken vs. the number of agents for LADMM during image mapping in [PITH_FULL_IMAGE:figures/full_fig_p023_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

75 extracted references · 3 linked inside Pith

  1. [47]

    Advances in mathematics , volume=

    A convexity principle for interacting gases , author=. Advances in mathematics , volume=. 1997 , publisher=

    1997

  2. [1]

    Contemporary mathematics , volume=

    A numerical method for the optimal time-continuous mass transport problem and related problems , author=. Contemporary mathematics , volume=. 1999 , publisher=

    1999

  3. [2]

    Mathematics of Computation , volume=

    A fast proximal gradient method and convergence analysis for dynamic mean field planning , author=. Mathematics of Computation , volume=

  4. [3]

    Journal of mathematical imaging and vision , volume=

    A first-order primal-dual algorithm for convex problems with applications to imaging , author=. Journal of mathematical imaging and vision , volume=. 2011 , publisher=

    2011

  5. [4]

    1998 , school=

    The optimal time-continuous mass transport problem and its augmented Lagrangian numerical resolution , author=. 1998 , school=

    1998

  6. [5]

    Journal of the Operations Research Society of China , volume=

    First-order algorithms for convex optimization with nonseparable objective and coupled constraints , author=. Journal of the Operations Research Society of China , volume=. 2017 , publisher=

    2017

  7. [6]

    Numerische Mathematik , volume=

    A computational fluid mechanics solution to the Monge-Kantorovich mass transfer problem , author=. Numerische Mathematik , volume=. 2000 , publisher=

    2000

  8. [7]

    Variational Methods for Discontinuous Structures: International Workshop at Villa Erba (Cernobbio), Italy, July 2001 , pages=

    Optimal transportation problems with free Dirichlet regions , author=. Variational Methods for Discontinuous Structures: International Workshop at Villa Erba (Cernobbio), Italy, July 2001 , pages=. 2002 , organization=

    2001

Show all 75 references
  1. [8]

    ESAIM: Control, Optimisation and Calculus of Variations , volume=

    A variational model for urban planning with traffic congestion , author=. ESAIM: Control, Optimisation and Calculus of Variations , volume=. 2005 , publisher=

    2005

  2. [9]

    2006 , publisher=

    Mass Transportation Problems: Volume 1: Theory , author=. 2006 , publisher=

    2006

  3. [10]

    2005 , school=

    Optimisation de problemes de transport , author=. 2005 , school=

    2005

  4. [11]

    Advances in neural information processing systems , volume=

    Sinkhorn distances: Lightspeed computation of optimal transport , author=. Advances in neural information processing systems , volume=

  5. [12]

    International conference on machine learning , pages=

    Fast computation of Wasserstein barycenters , author=. International conference on machine learning , pages=. 2014 , organization=

    2014

  6. [13]

    ACM Transactions on Graphics (ToG) , volume=

    Convolutional wasserstein distances: Efficient optimal transportation on geometric domains , author=. ACM Transactions on Graphics (ToG) , volume=. 2015 , publisher=

    2015

  7. [14]

    International conference on machine learning , pages=

    Wasserstein generative adversarial networks , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  8. [15]

    arXiv preprint arXiv:1803.05573 , year=

    Improving GANs using optimal transport , author=. arXiv preprint arXiv:1803.05573 , year=

    arXiv

  9. [16]

    The review of economic studies , volume=

    Income and wealth distribution in macroeconomics: A continuous-time approach , author=. The review of economic studies , volume=. 2022 , publisher=

    2022

  10. [17]

    2018 , publisher=

    Optimal transport methods in economics , author=. 2018 , publisher=

    2018

  11. [18]

    Proceedings of the National Academy of Sciences , volume=

    A machine learning framework for solving high-dimensional mean field game and mean field control problems , author=. Proceedings of the National Academy of Sciences , volume=. 2020 , publisher=

    2020

  12. [19]

    arXiv preprint arXiv:1807.01083 , year=

    A mean-field optimal control formulation of deep learning , author=. arXiv preprint arXiv:1807.01083 , year=

    arXiv

  13. [20]

    2021 , publisher=

    Topics in optimal transportation , author=. 2021 , publisher=

    2021

  14. [21]

    2009 , publisher=

    Optimal transport: old and new , author=. 2009 , publisher=

    2009

  15. [22]

    PhD Thesis (Univ

    Mean field games and applications to economics , author=. PhD Thesis (Univ. Paris-Dauphine , year=

  16. [23]

    Paris-Princeton lectures on mathematical finance 2010 , pages=

    Mean field games and applications , author=. Paris-Princeton lectures on mathematical finance 2010 , pages=. 2011 , publisher=

    2010

  17. [24]

    Doklady Akademii Nauk , volume=

    On the transfer of masses (in Russian) , author=. Doklady Akademii Nauk , volume=

  18. [25]

    Journal of Convex Analysis , volume=

    Dynamic formulation of optimal transport problems , author=. Journal of Convex Analysis , volume=. 2008 , publisher=

    2008

  19. [26]

    ESAIM: Mathematical Modelling and Numerical Analysis , volume=

    A generalized model for optimal transport of images including dissipation and density modulation , author=. ESAIM: Mathematical Modelling and Numerical Analysis , volume=. 2015 , publisher=

    2015

  20. [27]

    ESAIM: Mathematical Modelling and Numerical Analysis , volume=

    Eulerian models and algorithms for unbalanced optimal transport , author=. ESAIM: Mathematical Modelling and Numerical Analysis , volume=

  21. [28]

    Archive for Rational Mechanics and Analysis , volume=

    Generalized Wasserstein distance and its application to transport equations with source , author=. Archive for Rational Mechanics and Analysis , volume=. 2014 , publisher=

    2014

  22. [29]

    IEEE transactions on automatic control , volume=

    Large-population cost-coupled LQG problems with nonuniform agents: Individual-mass behavior and decentralized varepsilon -Nash equilibria , author=. IEEE transactions on automatic control , volume=. 2007 , publisher=

    2007

  23. [30]

    COMMUNICATIONS IN INFORMATION AND SYSTEMS , volume =

    Large population stochastic dynamic games: closed-loop McKean-Vlasov systems and the Nash certainty equivalence principle , author=. COMMUNICATIONS IN INFORMATION AND SYSTEMS , volume =

  24. [31]

    SIAM Journal on Control and Optimization , volume=

    Mean field games: numerical methods for the planning problem , author=. SIAM Journal on Control and Optimization , volume=. 2012 , publisher=

    2012

  25. [32]

    Journal of Optimization Theory and Applications , volume=

    Augmented Lagrangian methods for transport optimization, mean field games and degenerate elliptic equations , author=. Journal of Optimization Theory and Applications , volume=. 2015 , publisher=

    2015

  26. [33]

    Proceedings of the ARPA image understanding workshop , volume=

    The earth mover’s distance, multi-dimensional scaling, and color-based image retrieval , author=. Proceedings of the ARPA image understanding workshop , volume=

  27. [34]

    Computer Vision--ECCV 2008: 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part III 10 , pages=

    A linear time histogram metric for improved sift matching , author=. Computer Vision--ECCV 2008: 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part III 10 , pages=. 2008 , organization=

    2008

  28. [35]

    Economic Theory , volume=

    Hedonic price equilibria, stable matching, and optimal transport: equivalence, topology, and uniqueness , author=. Economic Theory , volume=. 2010 , publisher=

    2010

  29. [36]

    Advances in neural information processing systems , volume=

    Near-linear time approximation algorithms for optimal transport via Sinkhorn iteration , author=. Advances in neural information processing systems , volume=

  30. [37]

    IEEE signal processing magazine , volume=

    Optimal mass transport: Signal processing and machine-learning applications , author=. IEEE signal processing magazine , volume=. 2017 , publisher=

    2017

  31. [38]

    Sixth international conference on computer vision (IEEE Cat

    A metric for distributions with applications to image databases , author=. Sixth international conference on computer vision (IEEE Cat. No. 98CH36271) , pages=. 1998 , organization=

    1998

  32. [39]

    SIAM Journal on Scientific Computing , volume=

    Numerical simulation of diffusive and aggregation phenomena in nonlinear continuity equations by evolving diffeomorphisms , author=. SIAM Journal on Scientific Computing , volume=. 2010 , publisher=

    2010

  33. [40]

    SIAM journal on mathematical analysis , volume=

    The variational formulation of the Fokker--Planck equation , author=. SIAM journal on mathematical analysis , volume=. 1998 , publisher=

    1998

  34. [41]

    Foundations of Computational Mathematics , volume=

    Numerical analysis of a robust free energy diminishing finite volume scheme for parabolic equations with gradient structure , author=. Foundations of Computational Mathematics , volume=. 2017 , publisher=

    2017

  35. [42]

    ESAIM: Control, Optimisation and Calculus of Variations , volume=

    On the total variation Wasserstein gradient flow and the TV-JKO scheme , author=. ESAIM: Control, Optimisation and Calculus of Variations , volume=. 2019 , publisher=

    2019

  36. [43]

    Nature , volume=

    A reconstruction of the initial conditions of the universe by optimal mass transportation , author=. Nature , volume=. 2002 , publisher=

    2002

  37. [44]

    2017 , school=

    Unbalanced optimal transport: Models, numerical methods, applications , author=. 2017 , school=

    2017

  38. [45]

    Foundations of Computational Mathematics , volume=

    An interpolating distance between optimal transport and Fisher--Rao metrics , author=. Foundations of Computational Mathematics , volume=. 2018 , publisher=

    2018

  39. [46]

    SIAM Journal on Imaging Sciences , volume=

    Optimal transport with proximal splitting , author=. SIAM Journal on Imaging Sciences , volume=. 2014 , publisher=

    2014

  40. [48]

    HAL , year=

    Primal-dual formulation of the Dynamic Optimal Transport using Helmholtz-Hodge decomposition , author=. HAL , year=

  41. [49]

    2012 , publisher=

    Finite element methods for Navier-Stokes equations: theory and algorithms , author=. 2012 , publisher=

    2012

  42. [50]

    2015 , publisher=

    Optimal Transport for applied mathematicians, progress in nonlinear differential equations and their applications , author=. 2015 , publisher=

    2015

  43. [51]

    Numerische Mathematik , volume=

    Domain decomposition for entropy regularized optimal transport , author=. Numerische Mathematik , volume=. 2021 , publisher=

    2021

  44. [52]

    SIAM journal on numerical analysis , volume=

    A domain decomposition method for the polar factorization of vector-valued mappings , author=. SIAM journal on numerical analysis , volume=. 1995 , publisher=

    1995

  45. [53]

    Pacific Journal of Mathematics , volume=

    Concerning nonnegative matrices and doubly stochastic matrices , author=. Pacific Journal of Mathematics , volume=. 1967 , publisher=

    1967

  46. [54]

    SIAM Journal on Optimization , volume=

    Accelerated first-order primal-dual proximal methods for linearly constrained composite convex programming , author=. SIAM Journal on Optimization , volume=. 2017 , publisher=

    2017

  47. [55]

    SIAM Journal on Scientific Computing , volume=

    Iterative Bregman projections for regularized transportation problems , author=. SIAM Journal on Scientific Computing , volume=. 2015 , publisher=

    2015

  48. [56]

    arXiv preprint arXiv:2301.12065 , year=

    Decentralized Entropic Optimal Transport for Privacy-preserving Distributed Distribution Comparison , author=. arXiv preprint arXiv:2301.12065 , year=

    arXiv

  49. [57]

    SIAM Journal on Numerical Analysis , volume=

    Solving large-scale optimization problems with a convergence rate independent of grid size , author=. SIAM Journal on Numerical Analysis , volume=. 2019 , publisher=

    2019

  50. [58]

    Journal of Computational Physics , volume=

    Gradient enhanced ADMM algorithm for dynamic optimal transport on surfaces , author=. Journal of Computational Physics , volume=. 2025 , publisher=

    2025

  51. [59]

    ACM Transactions on Graphics (TOG) , volume=

    Dynamical optimal transport on discrete surfaces , author=. ACM Transactions on Graphics (TOG) , volume=. 2018 , publisher=

    2018

  52. [60]

    Transactions on Machine Learning Research , year=

    Neural monge map estimation and its applications , author=. Transactions on Machine Learning Research , year=

  53. [61]

    SIAM Journal on Scientific Computing , volume=

    A scalable deep learning approach for solving high-dimensional dynamic optimal transport , author=. SIAM Journal on Scientific Computing , volume=. 2023 , publisher=

    2023

  54. [62]

    Journal of Computational Physics , volume=

    Bridging mean-field games and normalizing flows with trajectory regularization , author=. Journal of Computational Physics , volume=. 2023 , publisher=

    2023

  55. [63]

    Journal of Computational Physics , volume=

    Unsupervised solution operator learning for mean-field games , author=. Journal of Computational Physics , volume=. 2025 , publisher=

    2025

  56. [64]

    2019 , publisher=

    Computational optimal transport: With applications to data science , author=. 2019 , publisher=

    2019

  57. [65]

    Journal of Computational Physics , volume=

    Computational mean-field games on manifolds , author=. Journal of Computational Physics , volume=. 2023 , publisher=

    2023

  58. [66]

    Journal of Scientific Computing , volume=

    A parallel method for earth mover’s distance , author=. Journal of Scientific Computing , volume=. 2018 , publisher=

    2018

  59. [67]

    Annals of operations research , volume=

    The auction algorithm: A distributed relaxation method for the assignment problem , author=. Annals of operations research , volume=. 1988 , publisher=

    1988

  60. [68]

    Naval research logistics quarterly , volume=

    The Hungarian method for the assignment problem , author=. Naval research logistics quarterly , volume=. 1955 , publisher=

    1955

  61. [69]

    SIAM review , volume=

    Tensor decompositions and applications , author=. SIAM review , volume=. 2009 , publisher=

    2009

  62. [70]

    Advances in neural information processing systems , volume=

    Linearized alternating direction method with adaptive penalty for low-rank representation , author=. Advances in neural information processing systems , volume=

  63. [71]

    East Asian Journal on Applied Mathematics , volume=

    Linearized alternating direction method of multipliers for constrained linear least-squares problem , author=. East Asian Journal on Applied Mathematics , volume=. 2012 , publisher=

    2012

  64. [72]

    SIAM Journal on Imaging Sciences , volume=

    An accelerated linearized alternating direction method of multipliers , author=. SIAM Journal on Imaging Sciences , volume=. 2015 , publisher=

    2015

  65. [73]

    Foundations of Computational Mathematics , volume=

    An extragradient-based alternating direction method for convex minimization , author=. Foundations of Computational Mathematics , volume=. 2017 , publisher=

    2017

  66. [74]

    Signal, Image and Video Processing , volume=

    Fast linearized alternating direction method of multipliers for the augmented _1 -regularized problem , author=. Signal, Image and Video Processing , volume=. 2015 , publisher=

    2015

  67. [75]

    SIAM Journal on Numerical Analysis , volume=

    On the O(1/n) convergence rate of the Douglas--Rachford alternating direction method , author=. SIAM Journal on Numerical Analysis , volume=. 2012 , publisher=

    2012

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.