Pith. sign in

REVIEW 4 major objections 5 minor 66 references

A transfer principle for computing the adapted Wasserstein distance between stochastic processes

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A transfer principle turns the adapted Wasserstein distance between Gaussian processes into a kernel computation, yielding an explicit formula for fractional Brownian motions.

desk verdict Real fBm distance formula, but the general Gaussian causal-factorization theorem has a measure mismatch that breaks Corollary 3.6 and the abstract's main claim. read the letter →

arxiv 2505.21337 v2 pith:ECNSVVBD submitted 2025-05-27 math.PR

Yifan Jiang , Fang Rui Lim This is my paper
classification math.PR MSC 60G1560G2249Q2247L35
keywords adaptedWassersteindistanceGaussianprocessfractionalBrownianmotionstochasticdifferentialequationnestalgebrapath-dependentHJBcausalfactorizationcanonicalrepresentation
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 aims to make the adapted 2-Wasserstein distance — the transport distance between stochastic processes that also respects the information flow of their natural filtrations — explicitly computable for Gaussian processes and fractional stochastic differential equations, a setting where almost nothing was known beyond semimartingales. The engine is a transfer principle: if each process $X^i$ is a filtration-preserving function $T^i(Y^i)$ of a simpler process $Y^i$, then the bicausal couplings of $X^1,X^2$ are exactly those of $Y^1,Y^2$, so one may switch to canonical coordinates in which the couplings are easy to describe. In those coordinates the paper derives a closed formula for the adapted distance between two unit-multiplicity mean-square continuous Gaussian processes in terms of their canonical kernels; for fractional Brownian motions this reduces to the squared $L^2$ difference of their Molchan–Golosov kernels $k_H$, with the synchronous coupling optimal. The same approach shows that for one-dimensional fractional SDEs with monotone structure the synchronous coupling between the driving fractional noises attains the distance, and it identifies the best martingale approximation of a fractional Brownian motion in this distance. These are the first such formulas for processes that are neither Markov nor semimartingales, and they make adapted distances usable in robust finance and model risk.

What carries the argument

Three objects carry the argument. The first is the transfer principle, equation (1.2): if $T^i$ pushes $Y^i$ onto $X^i$ and each pair generates the same natural filtration, then $\Pi_{bc}(X^1,X^2)=\Pi_{bc}(Y^1,Y^2)$, so an intractable transport problem can be solved in simpler coordinates. The second is the causal factorization (Definition 3.1), an infinite-dimensional analogue of the Cholesky decomposition: a Hilbert–Schmidt operator $K:H_\mu\to H$ with $\Sigma=KK^*$ that is causal, mapping each future subspace $H_{\mu,t}^\perp$ into $H_t^\perp$; a canonical causal factorization further requires every compression $K_t=P_{H_t}K|_{H_{\mu,t}}$ to be injective, which corresponds exactly to unit multiplicity in the canonical representation of Gaussian processes. The third is the canonical-kernel calculus: under any bicausal coupling the canonical Gaussian martingales $M_1,M_2$ remain martingales, the Kunita–Watanabe inequality forces their covariation density $\rho(s)$ against $\sqrt{\mu_1\mu_2}$ to lie in $[-1,1]$, and the optimal choice is $\rho(s)=\operatorname{sign}\langle k_1(\cdot,s),k_2(\cdot,s)\rangle$, which is why the absolute value of the pointwise kernel inner product enters formula (4.1). For the fractional SDE half of the paper, the companion machinery is the two-parameter auxiliary system $\Theta_i(s;t)$ that restores the flow property lost by Volterra drivers, making possible a functional Itô formula and a verification theorem for the path-dependent HJB equation whose control is the correlation between the driving noises.

What would settle it

The decisive check is the multiplicity-2 Gaussian process $X(t)=B_1(t)+F(t)B_2(t)$ described in the paper's Remark 4.12, with $B_1,B_2$ independent Brownian motions, $F'$ integrable, and $F$ nowhere locally square-integrable. Corollary 3.6 predicts that $X$ has the law of a scalar Gaussian Volterra process $\int_0^t k(t,s)\,dM(s)$ with a single Gaussian martingale $M$, which forces unit multiplicity of the filtration; counting the independent innovation martingales in the canonical decomposition of $X$ — the cited construction gives two — settles the corollary. Independently, Theorem 1.1 can be tested numerically: evaluate $\int_0^T\int_0^T (k_{H_1}(t,s)-k_{H_2}(t,s))^2\,dt\,ds$ for two Hurst parameters and compare it with a fine time-discretization of the optimal bicausal transport problem, which the discrete-time Gaussian formula discussed in Remark 4.6 approximates; the two values must converge as the mesh size goes to zero.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 4.3: for two centered, mean-square continuous, purely nondeterministic Gaussian processes of unit multiplicity with canonical representations $X_i(t)=\int_0^t k_i(t,s)\,dM_i(s)$, the adapted 2-Wasserstein distance is $$AW_2(X_1,X_2)^2=\int_0^T\|k_1(\cdot,s)\|^2\,\mu_1(ds)+\int_0^T\|k_2(\cdot,s)\|^2\,\mu_2(ds)-2\int_0^T|\langle k_1(\cdot,s),k_2(\cdot,s)\rangle|\,\sqrt{\mu_1\mu_2}(ds),$$ where $\mu_i=[M_i]$, equivalently $\operatorname{tr}(\Sigma_1+\Sigma_2)$ minus twice a triangular integral of the operator $K_1^*K_2$ over the nest of past subspaces. For fractional Brownian motions this specializes to Theorem 1.1: $AW_2(B^{H_1},B^{H_2})^2=\int_0^T\int_0^T (k_{H_1}(t,s)-k_{H_2}(t,s))^2\,dt\,ds$, and the synchronous coupling, in which both fractional noises are driven by the same Brownian motion in their Molchan–Golosov representations, is the optimal coupling. For one-dimensional fractional SDEs with $H\in(1/2,1)$, smooth positive coefficients, and either $b_i/\sigma_i$ non-decreasing or both Volterra kernels non-decreasing, Theorem 5.5 claims the synchronous coupling between the driving noises attains the distance, proved by a path-dependent HJB verification argument that solves the additive-noise case and then applies a Lamperti transform. Theorem 4.14 gives the best $F^{B^H}$-martingale approximation of a fractional Brownian motion: the martingale whose volatility at $r$ averages the prediction-process volatility over $[r,T]$. The paper intends all of this to hold for processes that are neither Markov nor semimartingales.

Load-bearing premise

The load-bearing premise is that every purely nondeterministic Gaussian process is driven by a single scalar noise — the causal factorization of Theorem 3.5 — whose proof inserts a Radon–Nikodym factor that shifts the measure against which the covariance kernel is integrated, and which must be reconciled with the higher-multiplicity Gaussian processes the paper itself cites before the blanket version of the Gaussian formula can be trusted.

Editorial extensions

If this is right

  • For fractional Brownian motions with Hurst parameters $H_1,H_2$, the distance takes the closed form $AW_2(B^{H_1},B^{H_2})^2=\int_0^T\!\int_0^T (k_{H_1}(t,s)-k_{H_2}(t,s))^2\,dt\,ds$, and the synchronous coupling is the unique optimal coupling (Theorem 1.1, Example 4.8).
  • For any pair of unit-multiplicity mean-square continuous Gaussian processes, the distance is given by formula (4.1) and does not depend on which canonical representation is chosen (Theorem 4.3, Remark 4.4); fractional Ornstein–Uhlenbeck processes are an explicit instance (Example 4.9).
  • For higher-multiplicity Gaussian processes the formula becomes a trace-norm identity involving the correlation matrices of the canonical martingales (Theorem 4.10), recovering the known discrete-time multi-dimensional Gaussian formula as a special case (Remark 4.6).
  • For one-dimensional fractional SDEs with $H\in(1/2,1)$, positive bounded coefficients, and either $b_i/\sigma_i$ non-decreasing or both Volterra kernels non-decreasing, the synchronous coupling between the driving fractional noises attains the adapted Wasserstein distance (Theorems 1.2 and 5.5).
  • The closest $F^{B^H}$-martingale to a fractional Brownian motion in adapted Wasserstein distance is explicitly identified: it is the martingale whose volatility at time $r$ averages the prediction-process volatility $(T-r)^{-1}\int_r^T k_H(s,r)\,ds$ (Theorem 4.14).

Reading between the lines

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

  • If Theorem 1.1 is correct, the map $H\mapsto k_H$ embeds the family of fractional Brownian motions isometrically into the Hilbert–Schmidt space of Volterra kernels, so all pairwise adapted distances are Hilbert distances that could be tabulated once the kernel map is known; the paper states the two-point formula but not this geometric reading of it.
  • The broad Gaussian statement rests on the scalar causal factorization of Theorem 3.5, which yields Corollary 3.6 that every purely nondeterministic Gaussian process is a scalar Gaussian Volterra process; the paper itself cites (Remark 4.12) higher-multiplicity Gaussian processes that escape a scalar Volterra representation, so a reader should check whether a unit-multiplicity hypothesis is actuall
  • In the paper's own spirit, any Gaussian family with a known canonical representation — Gaussian Markov processes, integrated fractional Brownian motions, Volterra processes with monotone kernels — should receive an explicit adapted distance as an $L^2$ kernel gap; this is a testable programme consistent with the announced extension to stochastic Volterra equations.
  • The recurring conclusion that the synchronous coupling is optimal in the canonical coordinates suggests a practical heuristic for non-Gaussian Volterra models: align the driving noises whenever the kernels are co-monotone, and use the resulting cost as a lower bound against which numerical adapted-transport algorithms can be checked.
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

4 major / 5 minor

Summary. The paper proposes a transfer principle for computing the adapted 2-Wasserstein distance between stochastic processes. It introduces a notion of causal factorization of covariance operators, claims existence of such a factorization for every purely nondeterministic mean-square continuous Gaussian process, and derives explicit formulas: Theorem 4.3 for unit-multiplicity Gaussian processes, Theorem 1.1 for fractional Brownian motions, Theorem 4.10 for finite higher multiplicity, and Theorem 5.5 for fractional SDEs. The fractional Brownian motion and fractional SDE results appear plausible, but the paper's central structural theorem on causal factorization is invalid, and the characterization of Gaussian Volterra processes in Corollary 3.6 is false.

Significance. If the unit-multiplicity and fractional SDE results are correct, they are significant: they provide some of the first explicit continuous-time adapted Wasserstein computations beyond the semimartingale setting, with the synchronous coupling identified as optimal. The paper also usefully connects Hida's canonical representation theory with adapted transport, and the formulas are explicit and falsifiable. However, because Theorem 3.5 and Corollary 3.6 fail, the advertised universal treatment of all real-valued mean-square continuous Gaussian processes is not supported. The credit that can be given is therefore to the unit-multiplicity and finite-multiplicity formulas, not to the claimed general existence of scalar causal factorizations.

major comments (4)
  1. [Section 3.1, proof of Theorem 3.5] The final step of the proof does not produce a factorization on the Lebesgue space H=L^2(dt). From Sigma_mu=K_mu K_mu^* on H_mu the kernel satisfies R(t,u)=integral_0^{t^u} k_mu(t,s)k_mu(u,s) mu(ds). The operator K defined by Kf(t)=integral_0^t k_mu(t,s) sqrt(dlambda/dmu)(s) f(s) mu(ds) has adjoint K^*g(s)=sqrt(dlambda/dmu)(s) integral_s^T k_mu(u,s)g(u)du, so that KK^*g(t)=integral_0^T g(u)du integral_0^{t^u} k_mu(t,s)k_mu(u,s) lambda(ds). This equals the covariance operator of X only when lambda=mu. Thus the proposed construction does not establish existence of a scalar causal factorization for a general Sigma in C(H); the density sqrt(dlambda/dmu) does not repair the measure mismatch.
  2. [Section 3.1, Corollary 3.6] Corollary 3.6 is false as stated. A scalar Gaussian Volterra process has a canonical representation driven by a single Gaussian martingale and hence unit multiplicity. The manuscript's own Remark 4.12 cites Hida's purely nondeterministic Gaussian process X(t)=B_1(t)+F(t)B_2(t) of multiplicity 2. By Theorem 2.5 such a process has a canonical representation with two martingale components, so it cannot be a scalar Gaussian Volterra process. Therefore the implication (ii) implies (iii) fails, and the corollary contradicts the higher-multiplicity theory that the paper itself cites.
  3. [Abstract and Section 1] Because Theorem 3.5 and Corollary 3.6 fail, the claim that the paper obtains an explicit formula for the adapted Wasserstein distance between arbitrary real-valued mean-square continuous Gaussian processes via causal factorization is unsupported. The formulas that remain supported are Theorem 4.3 and Theorem 1.1 for unit-multiplicity processes with a canonical representation given a priori, and Theorem 4.10 for finite multiplicity. The universal existence of a scalar causal factorization is no longer available, so the abstract and introduction need substantial qualification.
  4. [Section 4.2, Theorem 4.10] Theorem 4.10 only states a formula for finite multiplicities m,n. The surrounding text, especially Remark 4.11, suggests that arbitrary multiplicity can be handled by the same argument, but no statement or proof for N=infinity is provided. If the paper intends to cover all mean-square continuous Gaussian processes, an explicit infinite-multiplicity version of the formula and its proof are required; otherwise the universal claim in the abstract is not justified.
minor comments (5)
  1. [Section 4.2, Theorem 4.10] The displayed formula in Theorem 4.10 has malformed symbols in the cross term, and the geometric mean notation sqrt(mu_1^1 mu_1^2)(ds) should be defined consistently with Definition 2.1; please clean up the typesetting.
  2. [Section 4.2, Lemma 4.13] The equality-attaining choice Gamma=A^{1/2}UVB^{1/2} requires the thin SVD convention; otherwise the matrix dimensions in the product UV are unclear.
  3. [Section 5.1, Assumption 5.2] Assumption 5.2 uses a single Hurst parameter H in the bounds for both kernels k_i. If the two fractional noises have different Hurst parameters H_1 and H_2, the bounds should be stated separately for each kernel.
  4. [Section 5.1, proof of Theorem 5.10] The sentence 'u(r) equivalent pi/2 gives an optimal control' is notationally confusing; it should say that the correlation rho(r) equivalent 1, i.e. the synchronous coupling, attains the optimum.
  5. [Section 4.1, Example 4.7] In Example 4.7 the statement that every bicausal coupling attains the adapted Wasserstein distance is not immediate from the displayed formula and would benefit from a short justification.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the adapted-Wasserstein formulas are forward deductions from Hida's canonical representation theory, the Molchan-Golosov kernel, and standard martingale coupling estimates; self-citations are contextual only.

full rationale

The main derivation chain is not circular. The transfer principle in equation (1.2) is a filtration-preservation identity: if X_i = T_i(Y_i) with equal natural filtrations, then the sets of bicausal couplings coincide, so the adapted-Wasserstein problem is exactly reformulated in the Y-coordinates. No target distance is assumed as an input. Theorem 4.3 is then derived by optimizing the cross term over bicausal couplings: the paper represents E[<X1,X2>] as an integral of k1 k2 against the bracket [M1,M2], uses that M1 and M2 remain martingales under bicausal couplings, applies Kunita-Watanabe to bound the bracket by sqrt(mu1 mu2), and maximizes pointwise over correlation densities rho in [-1,1]. This is a genuine optimization, not a fit or a definitional identity. The fBM formula in Theorem 1.1 is a direct plug-in of the externally established Molchan-Golosov canonical representation (cited to Molchan-Golosov, Decreusefond-Ustunel, and Jost) into formula (4.1); since the Molchan-Golosov kernels are nonnegative, the optimal rho is identically 1, giving the synchronous coupling. There are no fitted parameters and no benchmark subset whose output is renamed a prediction. The paper's self-citations (Cont-Lim 2024, Jiang 2024, Jiang-Obloj 2024) are contextual references for causality density results and applications, and they do not carry the weight of any theorem needed for Theorem 1.1. The one substantive concern is a non-circular mathematical gap in the proof of Theorem 3.5: the operator K defined with the factor sqrt(d lambda/d mu) produces a KK* whose kernel is integrated against lambda, whereas the covariance kernel R is expanded in mu, so the asserted scalar causal factorization for every purely nondeterministic Gaussian process is not established, and Corollary 3.6 conflicts with the higher-multiplicity example in Remark 4.12. This affects the abstract's universal Gaussian-process claim, but it is a correctness issue, not circular reasoning, and it does not undermine the independently conditioned fBM calculation. Therefore the circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper adds no free parameters. It relies on deep background results: Hida's canonical representation, nest algebra factorization, and the Viens-Zhang functional Itô calculus. The key additional assumption is the existence of a scalar causal factorization, which the paper's proof appears to get wrong.

assumptions (5)
  • domain assumption Hida's canonical representation theorem (Theorem 2.5): every purely nondeterministic mean-square continuous Gaussian process has a canonical representation driven by a countable number of Gaussian martingales.
    This is the starting point for all Gaussian process results, used in Theorems 3.5, 4.3, and 4.10.
  • standard math Anoussis-Katsoulis factorization results in nest algebras (Proposition 3.4 and related theorems).
    Used in Theorem 3.5 to construct a causal factorization; the subsequent construction appears to contain an algebraic error.
  • domain assumption Jost's theorem that the Molchan-Golosov representation of a fractional Brownian motion is canonical (natural filtration equals the Brownian filtration).
    Needed for applying the transfer principle to fractional Brownian motions in Theorem 1.1.
  • domain assumption Viens-Zhang functional Itô formula and associated regularity estimates.
    Used in Section 5 to verify the path-dependent HJB equation and prove Theorem 5.5.
  • domain assumption Assumptions 5.1 and 5.2 ensuring well-posedness of the fractional SDEs and regularity of the kernels.
    These regularity conditions are used to apply the Viens-Zhang framework and the Lamperti transform in Section 5.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A transfer principle for computing the adapted Wasserstein distance between stochastic processes." pith.science (2026). https://pith.science/paper/ECNSVVBD

@misc{pith2026250521337,
  author       = {Pith},
  title        = {Pith review of: A transfer principle for computing the adapted Wasserstein distance between stochastic processes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ECNSVVBD}},
  note         = {Machine review of arXiv:2505.21337}
}
read the original abstract

We propose a transfer principle to study the adapted 2-Wasserstein distance between stochastic processes. First, we obtain an explicit formula for the distance between real-valued mean-square continuous Gaussian processes by introducing the causal factorization as an infinite-dimensional analogue of the Cholesky decomposition for operators on Hilbert spaces. We discuss the existence and uniqueness of this causal factorization and link it to the canonical representation of Gaussian processes. As a byproduct, we characterize mean-square continuous Gaussian Volterra processes in terms of their natural filtrations. Moreover, for real-valued fractional stochastic differential equations, we show that the synchronous coupling between the driving fractional noises attains the adapted Wasserstein distance under some monotonicity conditions. Our results cover a wide class of stochastic processes which are neither Markov processes nor semi-martingales, including fractional Brownian motions and fractional Ornstein--Uhlenbeck processes.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

66 extracted references · 57 canonical work pages

  1. [1]

    Acciaio and S

    B. Acciaio and S. Hou. Convergence of adapted empirical measures on R ^d . The Annals of Applied Probability, 34 0 (5): 0 4799--4835, Oct. 2024

    work page 2024

  2. [2]

    Acciaio, J

    B. Acciaio, J. Backhoff-Veraguas , and A. Zalashko. Causal optimal transport and its links to enlargement of filtrations and continuous-time stochastic optimization. Stochastic Processes and their Applications, 130 0 (5): 0 2918--2953, May 2020

    work page 2020

  3. [3]

    Acciaio, S

    B. Acciaio, S. Hou, and G. Pammer. Entropic adapted Wasserstein distance on Gaussians , Dec. 2024. arXiv:2412.18794

    arXiv 2024

  4. [4]

    D. Aldous. Weak convergence and the general theory of processes. July 1981. Incomplete monograph

    work page 1981

  5. [5]

    Anoussis and E

    M. Anoussis and E. G. Katsoulis. Factorisation in nest algebras. Proceedings of the American Mathematical Society, 125 0 (1): 0 87--92, 1997

    work page 1997

  6. [6]

    Anoussis and E

    M. Anoussis and E. G. Katsoulis. Factorisation in nest algebras. II . Transactions of the American Mathematical Society, 350 0 (1): 0 165--183, 1998

    work page 1998

  7. [7]

    W. Arveson. Operator algebras and invariant subspaces. Annals of Mathematics, 100 0 (3): 0 433--532, 1974

    work page 1974

  8. [8]

    Backhoff-Veraguas , M

    J. Backhoff-Veraguas , M. Beiglb \"o ck, Y. Lin, and A. Zalashko. Causal transport in discrete time and applications. SIAM Journal on Optimization, 27 0 (4): 0 2528--2562, Jan. 2017

    work page 2017

Show all 66 references
  1. [9]

    Backhoff-Veraguas , D

    J. Backhoff-Veraguas , D. Bartl, M. Beiglb \"o ck, and M. Eder. Adapted Wasserstein distances and stability in mathematical finance. Finance and Stochastics, 24 0 (3): 0 601--632, July 2020 a

    2020

  2. [10]

    Backhoff-Veraguas , D

    J. Backhoff-Veraguas , D. Bartl, M. Beiglb \"o ck, and M. Eder. All adapted topologies are equal. Probability Theory and Related Fields, 178 0 (3-4): 0 1125--1172, Dec. 2020 b

    2020

  3. [11]

    Backhoff-Veraguas , D

    J. Backhoff-Veraguas , D. Bartl, M. Beiglb \"o ck, and J. Wiesel. Estimating processes in adapted Wasserstein distance. The Annals of Applied Probability, 32 0 (1): 0 529--550, Feb. 2022 a

    2022

  4. [12]

    Backhoff-Veraguas , S

    J. Backhoff-Veraguas , S. K \"a llblad, and B. A. Robinson. Adapted Wasserstein distance between the laws of SDEs , Sept. 2022 b . arXiv:2209.03243

    2022

  5. [13]

    R. T. Baillie. Long memory processes and fractional integration in econometrics. Journal of Econometrics, 73 0 (1): 0 5--59, July 1996

    1996

  6. [14]

    Bartl and J

    D. Bartl and J. Wiesel. Sensitivity of multi-period optimization problems in adapted Wasserstein distance, Aug. 2022. arXiv:2208.05656

    2022 arXiv

  7. [15]

    Bartl, M

    D. Bartl, M. Beiglb \"o ck, and G. Pammer. The Wasserstein space of stochastic processes. Journal of the European Mathematical Society, Dec. 2024

    2024

  8. [16]

    Bartl, M

    D. Bartl, M. Beiglb \"o ck, G. Pammer, S. Schrott, and X. Zhang. The Wasserstein space of stochastic processes in continuous time, Jan. 2025. arXiv:2501.14135

    2025 arXiv

  9. [17]

    Beiglb \"o ck, G

    M. Beiglb \"o ck, G. Pammer, and S. Schrott. Denseness of biadapted Monge mappings. Annales de l'Institut Henri Poincar \'e , Probabilit \'e s et Statistiques , 61 0 (1), Feb. 2025

    2025

  10. [18]

    Bion--Nadal and D

    J. Bion--Nadal and D. Talay. On a Wasserstein-type distance between solutions to stochastic differential equations. The Annals of Applied Probability, 29 0 (3), June 2019

    2019

  11. [19]

    Cheridito

    P. Cheridito. Mixed fractional Brownian motion. Bernoulli, 7 0 (6): 0 913--934, 2001

    2001

  12. [20]

    Cheridito, H

    P. Cheridito, H. Kawaguchi, and M. Maejima. Fractional Ornstein-Uhlenbeck processes. Electronic Journal of Probability, 8: 0 1--14, Jan. 2003

    2003

  13. [21]

    R. Cont. Long range dependence in financial markets. In Fractals in Engineering , pages 159--179. Springer, 2005

    2005

  14. [22]

    Cont and D.-A

    R. Cont and D.-A. Fourni \'e . Change of variable formulas for non-anticipative functionals on path space. Journal of Functional Analysis, 259 0 (4): 0 1043--1072, Aug. 2010

    2010

  15. [23]

    Cont and F

    R. Cont and F. R. Lim. Causal transport on path space, Dec. 2024. arXiv:2412.02948

    2024 arXiv

  16. [24]

    Cram \'e r

    H. Cram \'e r. Structural and Statistical Problems for a Class of Stochastic Processes : The First Samuel Stanley Wilks Lecture at Princeton University , March 7, 1970 . Princeton University Press, 1971

    1970

  17. [25]

    K. R. Davidson. Nest Algebras : Triangular Forms for Operator Algebras on Hilbert Space . Pitman Research Notes in Mathematics Series. Longman Scientific & Technical and Wiley , 1988

    1988

  18. [26]

    K. R. Davidson. The Russo-Dye theorem in nest algebras. Proceedings of the American Mathematical Society, 126 0 (10): 0 3055--3059, 1998

    1998

  19. [27]

    Decreusefond and A

    L. Decreusefond and A. \"U st \"u nel . Stochastic analysis of the fractional Brownian motion. Potential Analysis, 10 0 (2): 0 177--214, Mar. 1999

    1999

  20. [28]

    Decreusefond and A

    L. Decreusefond and A. S. \"U st \"u nel. Fractional Brownian motion: Theory and applications. ESAIM: Proceedings, 5: 0 75--86, 1998

    1998

  21. [29]

    R. G. Douglas. On majorization, factorization, and range inclusion of operators on Hilbert space. Proceedings of the American Mathematical Society, 17 0 (2): 0 413--415, 1966

    1966

  22. [30]

    D. C. Dowson and B. V. Landau. The Fr \'e chet distance between multivariate normal distributions. Journal of Multivariate Analysis, 12 0 (3): 0 450--455, Sept. 1982

    1982

  23. [31]

    B. Dupire. Functional It \^o Calculus , July 2009. Bloomberg Portfolio Research Paper No. 2009-04-FRONTIERS

    2009

  24. [32]

    Eckstein and G

    S. Eckstein and G. Pammer. Computational methods for adapted optimal transport. The Annals of Applied Probability, 34 0 (1A): 0 675--713, Feb. 2024

    2024

  25. [33]

    J. Erdos. Unitary invariants for nests. Pacific Journal of Mathematics, 23 0 (2): 0 229--256, Nov. 1967

    1967

  26. [34]

    J. A. Erdos. Operators of finite rank in nest algebras. Journal of the London Mathematical Society, s1-43 0 (1): 0 391--397, 1968

    1968

  27. [35]

    P. K. Friz and M. Hairer. A Course on Rough Paths : With an Introduction to Regularity Structures . Universitext. Springer International Publishing, 2020

    2020

  28. [36]

    Gelbrich

    M. Gelbrich. On a formula for the \( L ^2\) Wasserstein metric between measures on Euclidean and Hilbert spaces. Mathematische Nachrichten, 147 0 (1): 0 185--203, 1990

    1990

  29. [37]

    Gunasingam and T.-K

    M. Gunasingam and T.-K. L. Wong. Adapted optimal transport between Gaussian processes in discrete time. Electronic Communications in Probability, 30: 0 1--14, Jan. 2025

    2025

  30. [38]

    M. F. Hellwig. Sequential decisions under uncertainty and the maximum theorem. Journal of Mathematical Economics, 25 0 (4): 0 443--464, Jan. 1996

    1996

  31. [39]

    T. Hida. Canonical representations of Gaussian processes and their applications. Memoirs of the College of Science, University of Kyoto. Series A: Mathematics, 33 0 (1): 0 109--155, Jan. 1960

    1960

  32. [40]

    Hida and M

    T. Hida and M. Hitsuda. Gaussian Processes , volume 120. American Mathematical Soc., 1993

    1993

  33. [41]

    M. Hitsuda. Representation of Gaussian processes equivalent to Wiener process. Osaka Journal of Mathematics, 5 0 (2): 0 299--312, Jan. 1968

    1968

  34. [42]

    D. N. Hoover. A characterization of adapted distribution. The Annals of Probability, 15 0 (4), Oct. 1987

    1987

  35. [43]

    D. N. Hoover and H. J. Keisler. Adapted probability distributions. Transactions of the American Mathematical Society, 286 0 (1): 0 159--201, 1984

    1984

  36. [44]

    Y. Jiang. Duality of causal distributionally robust optimization: The discrete-time case, Jan. 2024. arXiv:2401.16556

    2024

  37. [45]

    Jiang and J

    Y. Jiang and J. Ob \'o j. Sensitivity of causal distributionally robust optimization, Aug. 2024. arXiv:2408.17109

    2024 arXiv

  38. [46]

    C. Jost. Transformation formulas for fractional Brownian motion. Stochastic Processes and their Applications, 116 0 (10): 0 1341--1357, Oct. 2006

    2006

  39. [47]

    Lamperti

    J. Lamperti. A simple construction of certain diffusion processes. Journal of Mathematics of Kyoto University, 4 0 (1): 0 161--170, Jan. 1964

    1964

  40. [48]

    D. R. Larson. Nest algebras and similarity transformations. Annals of Mathematics, 121 0 (2): 0 409--427, 1985

    1985

  41. [49]

    Lassalle

    R. Lassalle. Causal transport plans and their Monge -- Kantorovich problems. Stochastic Analysis and Applications, 36 0 (3): 0 452--484, May 2018

    2018

  42. [50]

    L \'e vy

    P. L \'e vy. A special problem of Brownian motion, and a general theory of Gaussian random functions. In Proceedings of the Third Berkeley Symposium on Mathematical Statistics and Probability , Volume 2: Contributions to Probability Theory , volume 3.2, pages 133--176. Univers...

    1956

  43. [51]

    L \'e vy

    P. L \'e vy. Fonctions aléatoires à corrélation linéaire . Illinois Journal of Mathematics, 1 0 (2): 0 217 -- 258, 1957

    1957

  44. [52]

    L \'e vy-V \'e hel, E

    J. L \'e vy-V \'e hel, E. Lutton, and C. Tricot. Fractals in Engineering . Springer-Verlag, 2005

    2005

  45. [53]

    Metzler and J

    R. Metzler and J. Klafter. The random walk's guide to anomalous diffusion: A fractional dynamics approach. Physics Reports, 339 0 (1): 0 1--77, Dec. 2000

    2000

  46. [54]

    G. M. Molchan and Y. I. Golosov. Gaussian stationary processes with asymptotically a power spectrum . Doklady Akademii Nauk SSSR, 184 0 (3): 0 546--549, 1969

    1969

  47. [55]

    G. Pammer. A note on the adapted weak topology in discrete time. Electronic Communications in Probability, 29 0 (none): 0 1--13, Jan. 2024

    2024

  48. [56]

    G. C. Pflug and A. Pichler. A distance for multistage stochastic optimization models. SIAM Journal on Optimization, 22 0 (1): 0 1--23, Jan. 2012

    2012

  49. [57]

    G. C. Pflug and A. Pichler. From empirical observations to tree models for stochastic optimization: Convergence properties. SIAM Journal on Optimization, 26 0 (3): 0 1715--1740, Jan. 2016

    2016

  50. [58]

    D. R. Pitts. Factorization problems for nests: Factorization methods and characterizations of the universal factorization property. Journal of Functional Analysis, 79 0 (1): 0 57--90, July 1988

    1988

  51. [59]

    J. R. Ringrose. Super-diagonal forms for compact linear operators. Proceedings of the London Mathematical Society, s3-12 0 (1): 0 367--384, 1962

    1962

  52. [60]

    J. R. Ringrose. On some algebras of operators. Proceedings of the London Mathematical Society, s3-15 0 (1): 0 61--83, 1965

    1965

  53. [61]

    B. A. Robinson and M. Sz \"o lgyenyi. Bicausal optimal transport for SDEs with irregular coefficients, Sept. 2024. arXiv:2403.09941

    2024 arXiv

  54. [62]

    L. C. G. Rogers. Arbitrage with fractional Brownian motion . Mathematical Finance, 7 0 (1): 0 95--105, 1997

    1997

  55. [63]

    R \"u schendorf

    L. R \"u schendorf. The Wasserstein distance and approximation theorems. Probability Theory and Related Fields, 70 0 (1): 0 117--129, Mar. 1985

    1985

  56. [64]

    Sauldubois and N

    N. Sauldubois and N. Touzi. First order martingale model risk and semi-static hedging, Oct. 2024. arXiv:2410.06906

    2024

  57. [65]

    Viens and J

    F. Viens and J. Zhang. A martingale approach for fractional Brownian motions and related path dependent PDEs . The Annals of Applied Probability, 29 0 (6): 0 3489--3540, Dec. 2019

    2019

  58. [66]

    Vostrikova and M

    L. Vostrikova and M. Yor. Some invariance properties (of the laws) of Ocone’s martingales. S \'e minaire de Probabilit \'e s XXXIV , pages 417--431, 2007

    2007

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.