Pith. sign in

REVIEW 2 major objections 4 minor 61 references

Reduced Order Modeling of One-Dimensional Conservative PDEs via the Cumulative Distribution Transform

T0 review · 2 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read This paper claims that representing 1D conservative-PDE solutions by mass coordinates rather than spatial coordinates collapses transport-dominated solution manifolds into small linear subspaces, with rigorous Kolmogorov-width bounds and a

desk verdict Solid transport-coordinate ROM paper; the advertised O(n^-2) shrink applies only before shock formation, and the paper says so, but that should not stop a serious referee. read the letter →

arxiv 2607.17066 v1 pith:LH3YN7FI submitted 2026-07-19 math.AP cs.NAmath.NA

Harbir Antil , Rocío Díaz Martín , Ivan V. Medri , Kristofor E. Pas , Gustavo K. Rohde , Aryan Saxena , Sarswati Shah This is my paper · ORCID
classification math.APcs.NAmath.NA MSC 41A4649Q2265F5535L6535Q49
keywords cumulativedistributiontransformoptimaltransportWassersteindistanceKolmogorovwidthreducedordermodelingproperorthogonaldecompositionscalarconservationlawsadvection-diffusion
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

This paper aims to show that one-dimensional conservative PDEs, whose solutions are moving, nonnegative, mass-preserving densities, become far more compressible if you change coordinates before applying linear model reduction. The change of coordinates is the cumulative distribution transform (CDT), which labels each solution by the position of accumulated mass rather than by the spatial coordinate; in these coordinates a translation becomes an affine shift, and the Wasserstein distance becomes an ordinary weighted L2 distance. The authors prove that for linear transport the transformed solution manifold lies in a two-dimensional plane, that for general hyperbolic conservation laws its Kolmogorov n-width decays at worst like O(1/n) even after shocks and like O(1/n^2) in smooth pre-shock regimes, and that for advection-diffusion the transformed trajectory stays within O(sqrt(DT)) of that plane. On top of this, they build a CDT-POD scheme that maps snapshots to transform space, performs POD there, and maps back, and they show numerically that it needs substantially fewer modes than Eulerian POD for transport-dominated examples. A sympathetic reader would care because the failure of linear reduced models on transported fronts is a known bottleneck, and this paper gives both a mechanism for that failure and a concrete fix with quantitative guarantees.

What carries the argument

The central object is the cumulative distribution transform (CDT): the monotone optimal transport map hat-u from a reference density r to a target density u, defined implicitly by the mass-balance identity integral_{-inf}^{hat-u(xi)} u(x) dx = integral_{-inf}^{xi} r(s) ds, i.e., hat-u = F_u^{-1} o F_r. Three properties carry the argument: the CDT isometrically embeds the 1D Wasserstein space into the weighted Hilbert space L^2(dr); it converts translations into additive shifts (hat-{T_tau u} = hat-u + tau), so transported families become affine lines; and its mass labels are globally defined even after shocks, unlike classical characteristics. The width estimates then follow from a Hilbert-v

What would settle it

Take inviscid Burgers with a smooth initial profile that has negative slope somewhere, choose T before the shock time, compute the CDT snapshots of the exact solution, and measure the singular values of the transformed snapshot matrix; if they do not decay at least like (n-1)^{-2}, Theorem 4.5 is violated, and if after shock formation they decay exactly like n^{-1}, that confirms the general entropy-solution bound is the true post-shock rate.

Watch

Extended reading notes

Core claim

The central claim is that the CDT is the right coordinate system for linear compression of conservative-transport data. In the CDT, a density u is represented by the monotone map hat-u that pushes a fixed reference density onto u; the transform is an isometry from the 2-Wasserstein space into L^2(dr). For linear advection u_t + A u_x = 0, hat-u(t) = hat-u_0 + A t, so the solution manifold lies in span{hat-u_0, 1} and d_n = 0 for all n >= 2. For a scalar conservation law u_t + f(u)_x = 0, entropy solutions satisfy W2(u(t),u(s)) <= B_M |t-s|, giving d_n(hat-M; L^2(dr)) <= B_M T/(2n); under the pre-shock condition 1 + t f''(u0) u0' >= alpha > 0, the sharper bound d_n <= C T^2 / (8 sqrt(alpha) (

Load-bearing premise

The sharp O(n^-2) hyperbolic bound assumes no shock forms by time T (1 + t f''(u0) u0' >= alpha > 0); the paper itself flags that a higher-regularity post-shock analogue remains open, and the numerical pipeline further assumes the truncated CDT maps stay monotone so the inverse transform is valid.

Editorial extensions

If this is right

  • For linear advection, a rank-2 POD in CDT space is exact, so purely transported data can be represented by exactly two modes.
  • For entropy solutions of scalar conservation laws, n modes in CDT space guarantee a worst-case Wasserstein/weighted-L2 error of at most B_M T/(2n), even after shock formation.
  • In smooth pre-shock regimes, the guarantee improves to O(n^{-2}) with a constant that degrades only like alpha^{-1/2} as the minimum characteristic Jacobian tends to zero.
  • For conservative advection-diffusion, n=2 modes in CDT space stay within sqrt(2DT) of the exact trajectory, recovering the pure-transport d_2 = 0 as D approaches 0.
  • Numerically, CDT-POD gives faster singular-value decay and lower reconstruction error than Eulerian POD for transport-dominated dynamics, provided the reduced transport maps remain monotone so the inverse CDT is valid.

Reading between the lines

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

  • If these bounds hold for time evolution, the same transform-reduce-invert recipe should extend to parametric families of conservative densities, where one varies the initial condition or coefficients and expects similar compression for any snapshot set dominated by mass displacement.
  • Because the linear truncation can leave the monotone image of the CDT, an editor-level extension is that projecting the reduced transport maps onto the cone of nondecreasing functions should convert the proven transform-space n-width bounds into native-space reconstruction guarantees; this is a testable algorithmic modification the paper leaves open.
  • The robust O(1/n) bound after shocks suggests that post-shock entropy-solution manifolds are still much more compressible in Wasserstein coordinates than in native L2, where translated jumps have n-width decaying only like n^{-1/2}; a direct numerical comparison of native versus CDT widths at fixed n after shock would quantify the practical gain.
  • The construction is essentially one-dimensional because it relies on monotone transport maps; an inference is that higher-dimensional analogues would need slicing (for example, a Radon-CDT) or regularized transport maps, and the isometric width bounds should not be expected to transfer directly.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The paper introduces a Cumulative Distribution Transform (CDT) based reduced-order modeling pipeline for one-dimensional scalar conservative PDEs with nonnegative, equal-mass states. After recalling the CDT and its isometry between the 1D Wasserstein metric and weighted L2(dr), the authors show analytically that linear transport becomes an affine segment in CDT space with zero 2-width, and they prove three families of estimates: (i) for entropy solutions of scalar conservation laws, a robust O(1/n) Kolmogorov-width bound in L2(dr) via the Wasserstein dynamic formulation (Theorem 4.2); (ii) a sharper O(1/n^2) pre-shock bound under the characteristic non-crossing condition 1+t f''(u0)u0' >= alpha > 0 (Theorem 4.5); and (iii) for advection-diffusion, d2 <= sqrt(2DT) (Theorem 5.1) plus Taylor-remainder based O(D^2T^2) bounds (Theorem 5.6). The proposed CDT-POD method is tested on Burgers, traffic flow, Buckley-Leverett, and advection-diffusion, with code provided. The paper is explicit that the sharper hyperbolic estimate is pre-shock and that a post-shock counterpart remains open.

Significance. If correct, the results provide a rigorous explanation for the empirical success of transport-based coordinates in ROM: the exact low-dimensional structure for linear transport, the robust O(1/n) bound for entropy solutions, and the quantification of diffusion's departure from the rank-two transport plane are clean and original contributions. The proofs are largely self-contained and use standard tools correctly; Theorem 4.2 is a particularly nice application of the Benamou-Brenier dynamic formulation with v=f(u)/u, and Theorem 5.1 is a simple Brownian-coupling argument. The paper is honest about its limitations: the O(n^{-2}) hyperbolic bound is conditional on a pre-shock hypothesis, the post-shock sharp estimate is explicitly left open (Section 6), and the inverse CDT requires monotonicity, which linear truncation may destroy. Reproducible code and explicit transform/inverse-transform error floors strengthen the numerical claims. The main caveat is that the theoretical width bounds are in the Wasserstein/L2(dr) metric rather than native L2, so the theoretical comparison with native POD is not a direct L2 statement; the authors are careful about this in the main text.

major comments (2)
  1. [§4.1, Theorem 4.5] The advertised O(n^{-2}) hyperbolic width is conditional on 1+t f''(u0(ξ))u0'(ξ) >= alpha > 0 on [0,T]. For generic C^1 data with min u0' < 0 and f'' not identically zero on the range of u0, this quantity reaches zero at the first shock time, so the estimate covers only pre-shock times; after shocks, the only proven bound is the O(1/n) one in Theorem 4.2. Section 6 acknowledges this gap. Because the numerical examples in §4.2 (e.g., Burgers in Fig. 7) explicitly include post-shock times, the statement that the experiments 'support the theoretical picture' should be qualified: the observed faster decay after shocks is not covered by Theorem 4.5. I recommend making the pre-shock condition and the open post-shock question prominent in the abstract and contributions.
  2. [§5.1, Theorem 5.6 and Remark 5.5] The phrase 'sharper O(D^2T^2) estimates under additional regularity or away from initial layers' in the abstract can be misread as an asymptotic statement in D. For t0>0, the constant C(u(·,t0)) in (60) is bounded in Remark 5.5 by a quantity proportional to (D t0)^{-3}, so for fixed t0 the right-hand side of (60) grows as D→0; only the t0=0 smooth-data case (Remark 5.7) yields D→0 recovery. Please state clearly in Theorem 5.6 that the D^2 factor comes with a D-dependent constant for t0>0, and that the zero-width recovery as D→0 is restricted to the t0=0 case.
minor comments (4)
  1. [§2.5, Algorithm 1] The numerical experiments do not specify which monotone projection or rearrangement is applied when the truncated CDT map is not monotone. Please document this, and state the reference density r used in each experiment, for reproducibility of the reported reconstruction errors.
  2. [§5.1, proof of Theorem 5.6] The display bounding d2 by ||pu(t)-pu(t0)-...|| should include a supremum over t in [t0,T]; as written it appears to be missing the sup.
  3. [§4.2, Eq. (50)] For the Burgers initial condition u0 = A g(x)+0.1, please state explicitly how the total mass is normalized and how the CDT reference mass is matched; this affects the interpretation of the transformed snapshots.
  4. [§4.1, formal native-space comparison] The labels 'formal' and the statement that the native-space estimates are not exact widths are welcome. In Section 6, avoid wording that suggests the native-space comparison is a theorem; the current text is mostly careful but could be tightened.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Kolmogorov-width estimates are derived from the problem data and standard external theorems, with no fitted quantities renamed as predictions.

full rationale

Walking the derivation chain, the paper's main results are self-contained rather than circular. The linear-transport claim (Eq. 18-19) follows from the in-paper proof of the CDT composition/translation property (Lemma 2.5, Corollary 2.6), not from an unverified self-citation. Theorem 4.2 derives the O(1/n) entropy-solution bound using Kružkov's external well-posedness and L1-contraction results, the standard dynamic formulation of W2, and the CDT isometry (14); the constant BM is computed directly from the flux and initial state. Theorem 4.5 obtains the sharper O(n^-2) pre-shock bound from explicit characteristic-formula computations in Proposition 4.3, with C and alpha defined solely in terms of u0, f, and the pre-shock Jacobian; no parameter is fitted to the solution manifold. The advection-diffusion bounds (Theorems 5.1 and 5.6) are derived from a Brownian coupling and heat-kernel estimates, again with constants determined by D, T, and u0. The paper's heavy reliance on the authors' earlier CDT papers is real self-citation, but the specific properties used are either re-proved in the text (translation equivariance, isometry) or are standard optimal-transport facts from external references [51, 52, 53]. The explicit limitations--the pre-shock restriction in Theorem 4.5, the openness of a higher-regularity post-shock analogue (Section 6), and the possible loss of monotonicity under CDT-space truncation (Section 2.5)--are limitation statements, not circular steps. No 'prediction' is a renamed fit, and no load-bearing argument reduces to a self-citation chain. Therefore the correct circularity finding is no significant circularity, score 0.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No fitted free parameters appear: the width bounds depend only on the flux, initial condition, diffusion coefficient, and time horizon. The fixed reference density r is a user-chosen input to the algorithm; it is not reported for the numerical experiments, but the theoretical bounds are independent of it. The paper introduces no new physical or mathematical entity.

assumptions (7)
  • standard math Kruzhkov entropy-solution theory: existence, uniqueness, L1 contraction, comparison principle for scalar conservation laws with f in C^1_loc and u0 in L1∩L∞.
    Used in Theorem 4.2 to ensure solution existence, nonnegativity, mass conservation, and finite second moment.
  • standard math Dynamic formulation of the Wasserstein distance: a distributional solution of the continuity equation with ∫∫|v|^2 u dx dt < ∞ is an absolutely continuous curve in P2(R) with metric derivative bounded by the L2 velocity norm.
    Used in Theorem 4.2 to convert the entropy solution's flux velocity v=f(u)/u into a W2-Lipschitz bound.
  • standard math CDT properties: isometry W2 ↔ L2(dr), translation equivariance, composition rule; established in [36,42,43].
    Foundation for all width bounds and the CDT-POD pipeline; accepted from prior literature with proof sketches.
  • standard math In 1D, the optimal transport map between densities is the unique monotone map pushing one onto the other.
    Used in Lemma 3.3 to identify the CDT trajectory with the flow of the velocity field.
  • domain assumption Pre-shock characteristic invertibility: 1 + t f''(u0(ξ)) u0'(ξ) ≥ α > 0 on I×[0,T].
    Assumed in Proposition 4.3 and Theorem 4.5 for the sharp O(n^-2) width bound; not valid after shock formation.
  • domain assumption For sharp advection-diffusion estimates: u0 ∈ C^3(R), u0 > 0, derivatives in L∞∩L1, or t0 > 0 so the heat kernel makes u strictly positive and smooth.
    Needed in Lemma C.2 and Theorem 5.6 for the Taylor remainder constant C(u(·,t0)) to be finite.
  • domain assumption Snapshots are nonnegative with common total mass; boundary conditions/decay guarantee mass conservation.
    Needed for the CDT to apply at every time; stated in Section 2.1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reduced Order Modeling of One-Dimensional Conservative PDEs via the Cumulative Distribution Transform." pith.science (2026). https://pith.science/paper/LH3YN7FI

@misc{pith2026260717066,
  author       = {Pith},
  title        = {Pith review of: Reduced Order Modeling of One-Dimensional Conservative PDEs via the Cumulative Distribution Transform},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LH3YN7FI}},
  note         = {Machine review of arXiv:2607.17066}
}
abstract

We propose a reduced order modeling (ROM) framework for 1D conservative PDEs based on the cumulative distribution transform (CDT). The CDT maps nonnegative, equal-mass states into a Hilbert space in which 1D Wasserstein distances become weighted $L^2$ distances and translations become affine shifts. This makes the transform especially suited for transport-dominated dynamics, where Eulerian linear-subspace ROMs often suffer from slow decay of Kolmogorov widths. We study this phenomenon for scalar conservative dynamics by analyzing the solution manifold in CDT coordinates. For linear transport, the transformed solution manifold is contained in the 2-dimensional space spanned by the transformed initial datum and the constant function, and has zero Kolmogorov $2$-width. For nonlinear hyperbolic conservation laws, we prove two complementary types of estimates: robust $O(n^{-1})$ bounds that rely only on the conservative transport structure and remain meaningful after shock formation, and sharper $O(n^{-2})$ bounds in smooth pre-shock regimes. For conservative advection-diffusion, we show that the CDT trajectory remains within distance $O(\sqrt{DT})$ of the pure-transport plane, and we also obtain sharper $O(D^2T^2)$ estimates under additional regularity or away from initial layers. In both cases, the zero 2-width behavior of linear transport is recovered as the diffusion coefficient tends to zero. Motivated by these estimates, we develop a CDT-POD numerical scheme: snapshots are mapped to CDT space, Proper Orthogonal Decomposition (POD) is performed in transformed coordinates, and the inverse CDT is used to reconstruct physical states. Numerical experiments for several transport-dominated dynamics show that CDT-POD can capture solution manifolds with substantially fewer modes than Eulerian POD.

Figures

Figures reproduced from arXiv: 2607.17066 by the authors.

Figure 1
Figure 1. Sketch of a solution upx, tq of a conservative evolution equation and its discretized snapshot matrix U [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Sketch of solution snapshots in native space (left) and CDT space (right), and corre [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Examples of hyperbolic conservation laws. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Characteristics of pure advection equation. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: For each t, we define uppξ, tq as the (unique) point such that the identity (6) is satisfied (shared areas). The blue curves in the top panel represent up¨, tq at different times t “ t1, t2, t3 and the gray curve on the bottom represents a fixed reference density r. Th…
Figure 6
Figure 6. Figure 6: Comparison between the standard model reduction (top) and the CDT-based model [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Inviscid Burgers equation: comparison between native Euclidean ROM and CDT [PITH_FULL_IMAGE:figures/full_fig_p034_7.png]
Figure 8
Figure 8. Figure 8: Inviscid Burgers equation for two regimes with different maximum characteristic [PITH_FULL_IMAGE:figures/full_fig_p035_8.png]
Figure 9
Figure 9. Figure 9: Traffic Flow equation: comparison between native Euclidean ROM and CDT-ROM. [PITH_FULL_IMAGE:figures/full_fig_p037_9.png]
Figure 10
Figure 10. Figure 10: Traffic Flow equation for two regimes with different minimum characteristic speeds. [PITH_FULL_IMAGE:figures/full_fig_p038_10.png]
Figure 11
Figure 11. Figure 11: Buckley-Leverett equation: comparison between native Euclidean ROM and CDT [PITH_FULL_IMAGE:figures/full_fig_p040_11.png]
Figure 12
Figure 12. Figure 12: Buckley-Leverett equation for two regimes with different maximum characteristic [PITH_FULL_IMAGE:figures/full_fig_p041_12.png]
Figure 13
Figure 13. Figure 13: Advection-Diffusion equation: comparison between native Euclidean ROM and CDT [PITH_FULL_IMAGE:figures/full_fig_p046_13.png]
Figure 14
Figure 14. Figure 14: Advection-Diffusion equation for three regimes with different advection and diffusion [PITH_FULL_IMAGE:figures/full_fig_p047_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 4 linked inside Pith

  1. [1]

    L. C. Evans,Partial differential equations. American mathematical society, 2010, vol. 19

    2010

  2. [2]

    C. M. Dafermos,Hyperbolic conservation laws in continuum physics. Springer, 2005

    2005

  3. [3]

    The Boltzmann equation,

    C. Cercignani, “The Boltzmann equation,” inThe Boltzmann equation and its applications. Springer, 1988, pp. 40–103

    1988

  4. [4]

    Hamiltonian description of the ideal fluid,

    P. J. Morrison, “Hamiltonian description of the ideal fluid,”Reviews of modern physics, vol. 70, no. 2, p. 467, 1998

    1998

  5. [5]

    Projection-based model reduction with dynamically transformed modes,

    F. Black, P. Schulze, and B. Unger, “Projection-based model reduction with dynamically transformed modes,”ESAIM Math. Model. Numer. Anal., vol. 54, no. 6, pp. 2011–2043, 2020

    2011

  6. [6]

    Model Reduction for Transport-Dominated Problems via Online Adaptive Bases and Adaptive Sampling,

    B. Peherstorfer, “Model Reduction for Transport-Dominated Problems via Online Adaptive Bases and Adaptive Sampling,”SIAM J. Sci. Comput., vol. 42, no. 5, pp. A2803–A2836, 2020

    2020

  7. [7]

    Nonlinear model order reduction based on local reduced-order bases,

    D. Amsallem, M. J. Zahr, and C. Farhat, “Nonlinear model order reduction based on local reduced-order bases,”International Journal for Numerical Methods in Engineering, vol. 92, no. 10, pp. 891–916, 2012. 55

    2012

  8. [8]

    Model reduction for flow analysis and control,

    C. W. Rowley and S. T. Dawson, “Model reduction for flow analysis and control,”Annual Review of Fluid Mechanics, vol. 49, no. 1, pp. 387–417, 2017

    2017

Show all 61 references
  1. [9]

    Discovering governing equations from data by sparse identification of nonlinear dynamical systems,

    S. L. Brunton, J. L. Proctor, and J. N. Kutz, “Discovering governing equations from data by sparse identification of nonlinear dynamical systems,”Proceedings of the national academy of sciences, vol. 113, no. 15, pp. 3932–3937, 2016

    2016

  2. [10]

    Nonlinear reduction strategies for data compres- sion: A comprehensive comparison from diffusion to advection problems,

    I. C. Gonnella, F. Pichi, and G. Rozza, “Nonlinear reduction strategies for data compres- sion: A comprehensive comparison from diffusion to advection problems,” inScientific Machine Learning, ser. SEMA SIMAI Springer Series. Cham: Springer, 2026, vol. 42, pp. 177–198

    2026

  3. [11]

    The proper orthogonal decomposition in the analysis of turbulent flows,

    G. Berkooz, P. Holmes, and J. L. Lumley, “The proper orthogonal decomposition in the analysis of turbulent flows,”Annual review of fluid mechanics, vol. 25, no. 1, pp. 539–575, 1993

    1993

  4. [12]

    Antil, M

    H. Antil, M. Heinkenschloss, and D. C. Sorensen,Application of the Discrete Empirical Interpolation Method to Reduced Order Modeling of Nonlinear and Parametric Systems. Cham: Springer International Publishing, 2014, pp. 101–136

    2014

  5. [13]

    Dynamic mode decomposition and its variants,

    P. J. Schmid, “Dynamic mode decomposition and its variants,”Annual Review of Fluid Mechanics, vol. 54, no. 1, pp. 225–254, 2022

    2022

  6. [14]

    Dynamic mode decomposition: Theory and applications,

    J. H. Tu, “Dynamic mode decomposition: Theory and applications,” Ph.D. dissertation, Princeton University, 2013

    2013

  7. [15]

    Reduced basis methods: Success, limitations and future chal- lenges,

    M. Ohlberger and S. Rave, “Reduced basis methods: Success, limitations and future chal- lenges,”Proceedings of the Conference Algoritmy, pp. 1–12, 2016

    2016

  8. [16]

    Galerkin v. least-squares Petrov-Galerkin projec- tion in nonlinear model reduction,

    K. Carlberg, M. Barone, and H. Antil, “Galerkin v. least-squares Petrov-Galerkin projec- tion in nonlinear model reduction,”J. Comput. Phys., vol. 330, pp. 693–734, 2017

    2017

  9. [17]

    Kolmogorovn-widths for linear dynamical systems,

    B. Unger and S. Gugercin, “Kolmogorovn-widths for linear dynamical systems,”Advances in Computational Mathematics, vol. 45, pp. 2273–2286, 2019

    2019

  10. [18]

    Domain decomposition and model reduction for the numerical solution of PDE constrained optimization problems with localized optimization variables,

    H. Antil, M. Heinkenschloss, R. Hoppe, and D. C. Sorensen, “Domain decomposition and model reduction for the numerical solution of PDE constrained optimization problems with localized optimization variables,”Comput. Vis. Sci., vol. 13, no. 6, pp. 249–264, 2010

    2010

  11. [19]

    Domain decomposition and balanced truncation model reduction for shape optimization of the stokes system,

    H. Antil, M. Heinkenschloss, and R. H. Hoppe, “Domain decomposition and balanced truncation model reduction for shape optimization of the stokes system,”Optimization Methods and Software, vol. 26, no. 4-5, pp. 643–669, 2011

    2011

  12. [20]

    Decay of the Kolmogorovn-width for wave problems,

    C. Greif and K. Urban, “Decay of the Kolmogorovn-width for wave problems,”Applied Mathematics Letters, vol. 96, pp. 216–222, 2019

    2019

  13. [21]

    The Kolmogorovn-width for linear transport: exact representation and the influence of the data,

    F. Arbes, C. Greif, and K. Urban, “The Kolmogorovn-width for linear transport: exact representation and the influence of the data,”Advances in Computational Mathematics, vol. 51, no. 2, p. 13, 2025

    2025

  14. [22]

    Breaking the Kolmogorov barrier with nonlinear model reduction,

    B. Peherstorfer, “Breaking the Kolmogorov barrier with nonlinear model reduction,”No- tices of the American Mathematical Society, vol. 69, no. 5, pp. 725–733, 2022

    2022

  15. [23]

    Nonlinear model reduction for transport- dominated problems,

    J. S. Hesthaven, B. Peherstorfer, and B. Unger, “Nonlinear model reduction for transport- dominated problems,”Acta Numerica, vol. 35, 2026. 56

    2026

  16. [24]

    Model order reduction via Lie groups,

    Y. P. Wotte, P. Buchfink, S. Glas, F. Califano, and S. Stramigioli, “Model order reduction via Lie groups,”Communications in Nonlinear Science and Numerical Simulation, vol. 160, p. 109973, 2026

    2026

  17. [25]

    Model order reduction for problems with large convection effects,

    N. Cagniart, Y. Maday, and B. Stamm, “Model order reduction for problems with large convection effects,” inContributions to partial differential equations and applications. Springer, 2018, pp. 131–150

    2018

  18. [26]

    Model reduction of a parametrized scalar hyperbolic conserva- tion law using displacement interpolation,

    D. Rim and K. T. Mandli, “Model reduction of a parametrized scalar hyperbolic conserva- tion law using displacement interpolation,”arXiv preprint arXiv:1805.05938, 2018

    2018 arXiv

  19. [27]

    Displacement interpolation using monotone rearrangement,

    ——, “Displacement interpolation using monotone rearrangement,”SIAM/ASA Journal on Uncertainty Quantification, vol. 6, no. 4, pp. 1503–1531, 2018

    2018

  20. [28]

    Manifold approximations via transported sub- spaces: Model reduction for transport-dominated problems,

    D. Rim, B. Peherstorfer, and K. T. Mandli, “Manifold approximations via transported sub- spaces: Model reduction for transport-dominated problems,”SIAM Journal on Scientific Computing, vol. 45, no. 1, pp. A170–A199, 2023

    2023

  21. [29]

    A registration method for model order reduction: data compression and geom- etry reduction,

    T. Taddei, “A registration method for model order reduction: data compression and geom- etry reduction,”SIAM Journal on Scientific Computing, vol. 42, no. 2, pp. A997–A1027, 2020

    2020

  22. [30]

    Interpolation of functions with parameter dependent jumps by transformed snapshots,

    G. Welper, “Interpolation of functions with parameter dependent jumps by transformed snapshots,”SIAM Journal on Scientific Computing, vol. 39, no. 4, pp. A1225–A1250, 2017

    2017

  23. [31]

    Overcoming slowly decaying Kolmogorov n-width by transport maps: Application to model order reduction of fluid dynamics and fluid–structure interaction problems,

    M. Nonino, F. Ballarin, G. Rozza, and Y. Maday, “Overcoming slowly decaying Kolmogorov n-width by transport maps: Application to model order reduction of fluid dynamics and fluid–structure interaction problems,”Advances in Computational Science and Engineer- ing, vol. 1, no. 1...

    2023

  24. [32]

    Advection–diffusion in Lagrangian coordinates,

    J.-L. Thiffeault, “Advection–diffusion in Lagrangian coordinates,”Physics Letters A, vol. 309, no. 5-6, pp. 415–422, 2003

    2003

  25. [33]

    Lagrangian dynamic mode decomposition for construction of reduced-order models of advection-dominated phenomena,

    H. Lu and D. M. Tartakovsky, “Lagrangian dynamic mode decomposition for construction of reduced-order models of advection-dominated phenomena,”Journal of Computational Physics, vol. 407, p. 109229, 2020

    2020

  26. [34]

    Lagrangian basis method for dimensionality reduction of convection dominated nonlinear flows,

    R. Mojgani and M. Balajewicz, “Lagrangian basis method for dimensionality reduction of convection dominated nonlinear flows,”arXiv preprint arXiv:1701.04343, 2017

    2017 arXiv

  27. [35]

    Kolmogorovn-width and Lagrangian physics-informed neural networks: A causality-conforming manifold for convection- dominated PDEs,

    R. Mojgani, M. Balajewicz, and P. Hassanzadeh, “Kolmogorovn-width and Lagrangian physics-informed neural networks: A causality-conforming manifold for convection- dominated PDEs,”Computer Methods in Applied Mechanics and Engineering, vol. 404, p. 115810, 2023

    2023

  28. [36]

    The cumulative distribution transform and linear pattern classification,

    S. R. Park, S. Kolouri, S. Kundu, and G. K. Rohde, “The cumulative distribution transform and linear pattern classification,”Applied and computational harmonic analysis, vol. 45, no. 3, pp. 616–641, 2018

    2018

  29. [37]

    Parametric signal estimation using the cumulative distribution transform,

    A. H. M. Rubaiyat, K. M. Hallam, J. M. Nichols, M. N. Hutchinson, S. Li, and G. K. Rohde, “Parametric signal estimation using the cumulative distribution transform,”IEEE Transactions on Signal Processing, vol. 68, pp. 3312–3324, 2020

    2020

  30. [38]

    Nearest subspace search in the signed cumulative distribution transform space for 1d signal classi- fication,

    A. H. M. Rubaiyat, M. Shifat-E-Rabbi, Y. Zhuang, S. Li, and G. K. Rohde, “Nearest subspace search in the signed cumulative distribution transform space for 1d signal classi- fication,” inICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing ...

    2022

  31. [39]

    End-to-end signal classification in signed cumulative distribution transform space,

    A. H. M. Rubaiyat, S. Li, X. Yin, M. Shifat-E-Rabbi, Y. Zhuang, and G. K. Rohde, “End-to-end signal classification in signed cumulative distribution transform space,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 9, pp. 5936–5950, 2024

    2024

  32. [40]

    Data-driven identification of para- metric governing equations of dynamical systems using the signed cumulative distribution transform,

    A. H. M. Rubaiyat, D. H. Thai, J. M. Nichols, M. N. Hutchinson, S. P. Wallen, C. J. Naify, N. Geib, M. R. Haberman, and G. K. Rohde, “Data-driven identification of para- metric governing equations of dynamical systems using the signed cumulative distribution transform,”Compute...

    2024

  33. [41]

    Pytranskit,

    G. Rohde, “Pytranskit,” https://github.com/rohdelab/PyTransKit, 2021

    2021

  34. [42]

    The signed cumula- tive distribution transform for 1-d signal analysis and classification,

    A. Aldroubi, R. D ´ ıaz Mart ´ ın, I. Medri, G. K. Rohde, and S. Thareja, “The signed cumula- tive distribution transform for 1-d signal analysis and classification,”Foundations of Data Science, vol. 4, no. 1, pp. 137–163, 2022

    2022

  35. [43]

    Data representation with optimal trans- port,

    R. D ´ ıaz Mart ´ ın, I. V. Medri, and G. K. Rohde, “Data representation with optimal trans- port,”arXiv preprint arXiv:2406.15503, 2024

    2024 arXiv

  36. [44]

    Model reduction of traveling-wave problems via Radon cumulative distribution transform,

    J. Ren, W. R. Wolf, and X. Mao, “Model reduction of traveling-wave problems via Radon cumulative distribution transform,”Physical Review Fluids, vol. 6, no. 8, p. L082501, 2021

    2021

  37. [45]

    A reduced-order model for advection-dominated problems based on the Radon Cumulative Distribution Transform,

    T. Long, R. Barnett, R. Jefferson-Loveday, G. Stabile, and M. Icardi, “A reduced-order model for advection-dominated problems based on the Radon Cumulative Distribution Transform,”Advances in Computational Mathematics, vol. 51, no. 1, p. 5, 2025

    2025

  38. [46]

    Neural Parametric Fokker–Planck Equation,

    S. Liu, W. Li, H. Zha, and H. Zhou, “Neural Parametric Fokker–Planck Equation,”SIAM Journal on Numerical Analysis, vol. 60, no. 3, pp. 1385–1449, 2022

    2022

  39. [47]

    Parameterized Wasserstein Gradient Flow,

    Y. Jin, S. Liu, H. Wu, X. Ye, and H. Zhou, “Parameterized Wasserstein Gradient Flow,” Journal of Computational Physics, vol. 524, p. 113660, 2025

    2025

  40. [48]

    Numerical analysis on neural network projected schemes for approximating one dimensional Wasserstein gradient flows,

    X. Zuo, J. Zhao, S. Liu, S. Osher, and W. Li, “Numerical analysis on neural network projected schemes for approximating one dimensional Wasserstein gradient flows,”Journal of Computational Physics, vol. 546, p. 114501, 2026

    2026

  41. [49]

    A computational fluid mechanics solution to the Monge- Kantorovich mass transfer problem,

    J.-D. Benamou and Y. Brenier, “A computational fluid mechanics solution to the Monge- Kantorovich mass transfer problem,”Numerische Mathematik, vol. 84, no. 3, pp. 375–393, 2000

    2000

  42. [50]

    The geometry of dissipative evolution equations: The porous medium equation,

    F. Otto, “The geometry of dissipative evolution equations: The porous medium equation,” Communications in Partial Differential Equations, vol. 26, no. 1–2, pp. 101–174, 2001

    2001

  43. [51]

    Ambrosio, N

    L. Ambrosio, N. Gigli, and G. Savar´ e,Gradient flows: in metric spaces and in the space of probability measures. Springer Science & Business Media, 2005

    2005

  44. [52]

    Santambrogio,Optimal Transport for Applied Mathematicians, ser

    F. Santambrogio,Optimal Transport for Applied Mathematicians, ser. PNLDE. Birkh¨ auser Boston, 2015, vol. 87

    2015

  45. [53]

    Villani,Topics in optimal transportation

    C. Villani,Topics in optimal transportation. American Mathematical Soc., 2003, no. 58

    2003

  46. [54]

    Hyt¨ onen, J

    T. Hyt¨ onen, J. Van Neerven, M. Veraar, and L. Weis,Analysis in Banach spaces. Springer, 2016, vol. 1

    2016

  47. [55]

    The shifted proper orthogonal decomposition: A mode decomposition for multiple transport phenomena,

    J. Reiss, P. Schulze, J. Sesterhenn, and V. Mehrmann, “The shifted proper orthogonal decomposition: A mode decomposition for multiple transport phenomena,”SIAM Journal on Scientific Computing, vol. 40, no. 3, pp. A1322–A1344, 2018. 58

    2018

  48. [56]

    A discrete cumulative distribution transform via optimal transport,

    H. Antil, G. Rohde, and A. Saxena, “A discrete cumulative distribution transform via optimal transport,”arXiv preprint arXiv:2606.12131, 2026

    2026 arXiv

  49. [57]

    Pinkus,N-widths in Approximation Theory

    A. Pinkus,N-widths in Approximation Theory. Springer Science & Business Media, 2012, vol. 7

    2012

  50. [58]

    A survey of projection-based model reduction methods for parametric dynamical systems,

    P. Benner, S. Gugercin, and K. Willcox, “A survey of projection-based model reduction methods for parametric dynamical systems,”SIAM Review, vol. 57, no. 4, pp. 483–531, 2015

    2015

  51. [59]

    E. A. Coddington and N. Levinson,Theory of Ordinary Differential Equations. New York: McGraw-Hill, 1955

    1955

  52. [60]

    Results concerning the nature of the continuity of solutions of parabolic equations and some of their applications,

    S. N. Kruˇ zkov, “Results concerning the nature of the continuity of solutions of parabolic equations and some of their applications,”Mathematical Notes of the Academy of Sciences of the USSR, vol. 6, no. 1, pp. 517–523, 1969

    1969

  53. [61]

    First order quasilinear equations in several independent variables,

    S. N. Kruˇ zkov, “First order quasilinear equations in several independent variables,”Math- ematics of the USSR-Sbornik, vol. 10, no. 2, pp. 217–243, 1970. 59

    1970

Pith tools

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