pith. sign in

arxiv: 2606.24207 · v1 · pith:MJOTMCKCnew · submitted 2026-06-23 · 🧮 math.NA · cs.NA

Quasi-Monte Carlo for SDE Simulation: Error Analysis and Dimensionality Reduction

Du Ouyang , Zexin Pan , Zhijian He This is my paper

Pith reviewed 2026-06-25 23:38 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords quasi-Monte Carlostochastic differential equationsEuler-Maruyama schememultilevel time griddimensionality reductionsampling errortruncation errorexact simulation
0
0 comments X

The pith

Quasi-Monte Carlo applied to the Euler-Maruyama scheme accelerates sampling error decay for SDEs beyond classical Monte Carlo, while the MSTG method achieves super-exponential truncation decay to cut integration dimension.

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

The paper shows that replacing standard Monte Carlo with quasi-Monte Carlo on the Euler-Maruyama discretization of general SDEs produces faster decay in sampling error and a better asymptotic rate. It then introduces the Multilevel Stochastic Time Grid construction, which relies on exact simulation to keep the sampling error at high order while driving the truncation error down super-exponentially. Because far fewer time steps suffice for a target accuracy, the effective dimension of the QMC integral drops sharply and practical run times improve. A reader would care because SDE simulation in finance, physics, and engineering is routinely limited by high-dimensional integration cost.

Core claim

The QMC method applied to the Euler-Maruyama scheme significantly accelerates the decay of the sampling error and achieves an asymptotically superior convergence rate over classical Monte Carlo; the MSTG method preserves high-order sampling convergence while its truncation error exhibits super-exponential decay, drastically reducing integration dimension.

What carries the argument

The Multilevel Stochastic Time Grid (MSTG) method, which builds a multilevel time grid from exact simulation techniques to replace the standard Euler-Maruyama time-stepping.

If this is right

  • Sampling error under QMC on the Euler-Maruyama scheme converges faster than under classical Monte Carlo.
  • MSTG maintains the high-order convergence rate of randomized QMC sampling.
  • The truncation error of MSTG decays super-exponentially, so a prescribed accuracy requires far fewer discretization steps than the Euler-Maruyama scheme.
  • The resulting lower integration dimension improves the practical efficiency of the overall QMC algorithm.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same multilevel construction could be tested on SDEs where exact simulation is feasible but the dynamics are more complex than those used in the paper's experiments.
  • If the super-exponential truncation holds, the method would allow QMC to reach fixed accuracy targets on path-dependent functionals that previously required prohibitive dimensions.
  • Direct comparison of wall-clock time versus classical Monte Carlo on the same SDE at equal error tolerance would quantify the dimension-reduction gain.

Load-bearing premise

Exact simulation techniques must be available for the underlying SDE so that the multilevel time grid can be realized without adding extra discretization bias.

What would settle it

Numerical runs on an SDE admitting exact simulation that show only polynomial rather than super-exponential decay in the truncation error of the MSTG method would refute the claimed rate.

read the original abstract

We investigate the numerical simulation of general stochastic differential equations (SDEs) using Quasi-Monte Carlo (QMC) methods. First, we provide a rigorous theoretical analysis of the QMC method applied to the Euler-Maruyama (EM) scheme, establishing that it significantly accelerates the decay of the sampling error and achieves an asymptotically superior convergence rate over the classical Monte Carlo method. Second, the traditional EM scheme exhibits a slow polynomial decay of the discretization error, which necessitates a large number of time steps and leads to a significantly high integration dimension. To address this issue, we propose a Multilevel Stochastic Time Grid (MSTG) method based on Exact Simulation techniques, and we rigorously establish its convergence rate under randomized QMC sampling, proving that it preserves the high-order convergence of the sampling error. In terms of the overall error, the truncation error of the proposed MSTG method exhibits a remarkably fast super-exponential decay. Consequently, to achieve a given accuracy level, our approach requires significantly fewer discretization steps than the EM scheme, thereby drastically reducing the actual integration dimension of the QMC method. This substantial dimensionality reduction strategy greatly enhances the practical efficiency of the QMC algorithm. Numerical experiments fully corroborate the superiority of the proposed approach.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript claims that Quasi-Monte Carlo (QMC) applied to the Euler-Maruyama discretization of general SDEs yields asymptotically superior sampling-error convergence compared with classical Monte Carlo. It further proposes a Multilevel Stochastic Time Grid (MSTG) construction based on exact simulation techniques whose truncation error decays super-exponentially while preserving the high-order QMC sampling rate, thereby drastically lowering the effective integration dimension.

Significance. If the stated rates can be established under verifiable assumptions, the dimensionality-reduction strategy would materially improve the practicality of QMC for SDE simulation. The explicit separation of sampling and truncation errors, together with the super-exponential truncation claim, would constitute a concrete advance over standard multilevel Monte Carlo approaches.

major comments (2)
  1. [MSTG construction and convergence analysis] MSTG construction (abstract and § on MSTG): the super-exponential truncation-error decay and the dimension-reduction claim are derived under the premise that unbiased exact simulation of the SDE transition is available. For the general SDEs targeted by the paper, such unbiased exact simulators do not exist; any practical implementation therefore re-introduces a discretization bias that the analysis assumes away. This assumption is load-bearing for the headline dimensionality-reduction result.
  2. [Theoretical analysis sections] Theoretical analysis (abstract): the manuscript asserts that it 'rigorously establish[es] its convergence rate' for both the EM-QMC scheme and the randomized-QMC MSTG method, yet supplies neither the precise regularity hypotheses on the drift and diffusion coefficients nor any proof outline or reference to the key lemmas. Without these, the claimed rates cannot be verified.
minor comments (1)
  1. The numerical experiments are stated to 'fully corroborate' the claims, but the abstract gives no information on the test SDEs, the range of dimensions examined, or the precise error metrics used.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the scope of our results. We address each major comment below and indicate the revisions planned for the next version of the manuscript.

read point-by-point responses

  1. Referee: [MSTG construction and convergence analysis] MSTG construction (abstract and § on MSTG): the super-exponential truncation-error decay and the dimension-reduction claim are derived under the premise that unbiased exact simulation of the SDE transition is available. For the general SDEs targeted by the paper, such unbiased exact simulators do not exist; any practical implementation therefore re-introduces a discretization bias that the analysis assumes away. This assumption is load-bearing for the headline dimensionality-reduction result.

    Authors: The manuscript distinguishes the EM-QMC analysis, which applies to general SDEs under standard Lipschitz and growth conditions, from the MSTG construction, which is explicitly based on the availability of unbiased exact simulation of transitions. Such exact simulation is feasible for specific SDE classes (e.g., geometric Brownian motion, certain affine diffusions, or processes with known transition densities). The super-exponential truncation decay and associated dimension reduction hold under that premise. We agree that the abstract and MSTG section should more explicitly delimit the applicable SDE class and note that, without exact simulation, the approach reverts to a multilevel discretization scheme. We will revise accordingly to avoid any implication that the MSTG rates apply to arbitrary general SDEs. revision: yes

  2. Referee: [Theoretical analysis sections] Theoretical analysis (abstract): the manuscript asserts that it 'rigorously establish[es] its convergence rate' for both the EM-QMC scheme and the randomized-QMC MSTG method, yet supplies neither the precise regularity hypotheses on the drift and diffusion coefficients nor any proof outline or reference to the key lemmas. Without these, the claimed rates cannot be verified.

    Authors: Sections 3 and 4 of the full manuscript contain the precise regularity assumptions (Lipschitz continuity and bounded derivatives up to order two for the EM-QMC rates; additional smoothness and moment conditions for the exact-simulation MSTG rates) together with the complete proofs. To address the concern about verifiability, we will add a concise summary of the key hypotheses and a high-level proof outline (with references to the main lemmas) either in the introduction or in a new preliminary section. This will not alter the stated rates but will make them easier to check. revision: yes

Circularity Check

0 steps flagged

No circularity detected; derivation is self-contained

full rationale

The paper's claims rest on standard QMC error bounds applied to the Euler-Maruyama discretization and a proposed MSTG construction that invokes external exact simulation techniques as a premise. No equations, fitted parameters, or self-citations are exhibited in the abstract or described claims that reduce the stated convergence rates or dimensionality reduction to a quantity defined by the authors' own prior work. The analysis is presented as independent theoretical derivation under the stated assumptions, qualifying as self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The abstract invokes standard existence/uniqueness theory for SDEs and the availability of exact simulation methods without listing any fitted constants or new postulated entities; the only background assumptions are those required for the Euler-Maruyama scheme and for QMC discrepancy bounds to apply.

axioms (2)
  • domain assumption The SDE coefficients satisfy conditions that guarantee existence and uniqueness of strong solutions and permit application of the Euler-Maruyama scheme.
    Implicit in the statement that the analysis applies to 'general stochastic differential equations' and the use of the EM scheme.
  • domain assumption Exact simulation techniques exist for the SDE under consideration.
    Explicitly required by the sentence 'we propose a Multilevel Stochastic Time Grid (MSTG) method based on Exact Simulation techniques'.

pith-pipeline@v0.9.1-grok · 5751 in / 1655 out tokens · 21775 ms · 2026-06-25T23:38:09.082605+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

39 extracted references · 1 canonical work pages · 1 internal anchor

  1. [1]

    Andersson and A

    P. Andersson and A. Kohatsu-Higa, Unbiased simulation of stochastic differential equations using parametrix expansions, Bernoulli, 23 (2017). 30 DU OUYANG, ZEXIN PAN, AND ZHIJIAN HE

    2017

  2. [2]

    Beskos, O

    A. Beskos, O. Papaspiliopoulos, and G. O. Roberts , Retrospective exact simulation of diffusion sample paths with applications , Bernoulli, 12 (2006)

    2006

  3. [3]

    Beskos and G

    A. Beskos and G. O. Roberts , Exact simulation of diffusions , The Annals of Applied Prob- ability, 15 (2005), https://arxiv.org/abs/math/0602523

    Pith/arXiv arXiv 2005

  4. [4]

    R. E. Caflisch, W. Morokoff, and A. Owen , Valuation of mortgage-backed securities us- ing Brownian bridges to reduce effective dimension , Journal of Computational Finance, 1 (1997), pp. 27–46

    1997

  5. [5]

    H. A. David and H. N. Nagaraja , Order Statistics, Wiley, Hoboken, NJ, 3 ed., 2003

    2003

  6. [6]

    Dick and F

    J. Dick and F. Pillichshammer , Digital Nets and Sequences. Discrepancy Theory and Quasi–Monte Carlo Integration, Cambridge University Press, Cambridge, 2010

    2010

  7. [7]

    Doumbia, N

    M. Doumbia, N. Oudjane, and X. Warin , Unbiased Monte Carlo estimate of stochastic differential equations expectations, ESAIM: Probability and Statistics, 21 (2017), pp. 56– 87

    2017

  8. [8]

    R. G. Gallager, Discrete Stochastic Processes, Kluwer Academic Publishers, 1996

    1996

  9. [9]

    M. B. Giles and B. J. Waterhouse , Multilevel quasi-Monte Carlo path simulation , in Ad- vanced Financial Modelling, H. Albrecher, W. J. Runggaldier, and W. Schachermayer, eds., Walter de Gruyter, Oct. 2009, pp. 165–182

    2009

  10. [10]

    Glasserman, Monte Carlo Methods in Financial Engineering , Springer, New York, 2004

    P. Glasserman, Monte Carlo Methods in Financial Engineering , Springer, New York, 2004

    2004

  11. [11]

    He and X

    Z. He and X. Wang, Good path generation methods in quasi-Monte Carlo for pricing financial derivatives, SIAM Journal on Scientific Computing, 36 (2014), pp. B171–B197

    2014

  12. [13]

    He and X

    Z. He and X. Wang , On the convergence rate of randomized quasi–Monte Carlo for discon- tinuous functions, SIAM Journal on Numerical Analysis, 53 (2015), pp. 2488–2503

    2015

  13. [14]

    Henry-Labord`ere, X

    P. Henry-Labord`ere, X. Tan, and N. Touzi , Unbiased simulation of stochastic differential equations, The Annals of Applied Probability, 27 (2017)

    2017

  14. [15]

    Hlawka and R

    E. Hlawka and R. M¨uck, ¨Uber eine transformation von gleichverteilten folgen II, Computing, 9 (1972), pp. 127–138

    1972

  15. [16]

    P. E. Kloeden and E. Platen , Numerical Solution of Stochastic Differential Equations , Springer Berlin Heidelberg, Berlin, Heidelberg, 1992

    1992

  16. [17]

    F. Y. Kuo and D. Nuyens , Application of quasi-Monte Carlo Methods to elliptic PDEs with random diffusion coefficients: A survey of analysis and implementation , Foundations of Computational Mathematics, 16 (2016), pp. 1631–1696

    2016

  17. [18]

    F. Y. Kuo, I. H. Sloan, G. W. Wasilkowski, and B. J. Waterhouse , Randomly shifted lattice rules with the optimal rate of convergence for unbounded integrands , Journal of Complexity, 26 (2010), pp. 135–160

    2010

  18. [19]

    F. Y. Kuo, G. W. Wasilkowski, and B. J. Waterhouse , Randomly shifted lattice rules for unbounded integrands, Journal of Complexity, 22 (2006), pp. 630–651

    2006

  19. [20]

    McLeish, A general method for debiasing a Monte Carlo estimator , Monte Carlo Methods and Applications, 17 (2011)

    D. McLeish, A general method for debiasing a Monte Carlo estimator , Monte Carlo Methods and Applications, 17 (2011)

    2011

  20. [21]

    Niederreiter , Random Number Generation and Quasi-Monte Carlo Methods , SIAM, Philadelphia, PA, 1992

    H. Niederreiter , Random Number Generation and Quasi-Monte Carlo Methods , SIAM, Philadelphia, PA, 1992

    1992

  21. [22]

    Ouyang, X

    D. Ouyang, X. Wang, and Z. He, Achieving high convergence rates by quasi-monte carlo and importance sampling for unbounded integrands , SIAM Journal on Numerical Analysis, 62 (2024), pp. 2393–2414

    2024

  22. [23]

    A. B. Owen, Randomly permuted (t,m,s)-nets and (t, s)-sequences, in Monte Carlo and Quasi- Monte Carlo Methods in Scientific Computing, H. Niederreiter and P. J.-S. Shiue, eds., New York, NY, 1995, Springer New York, pp. 299–317

    1995

  23. [24]

    A. B. Owen, Scrambled net variance for integrals of smooth functions, The Annals of Statistics, 25 (1997), pp. 1541–1562

    1997

  24. [25]

    A. B. Owen, Halton sequences avoid the origin , SIAM Review, 48 (2006), pp. 487–503

    2006

  25. [26]

    A. B. Owen, Local antithetic sampling with scrambled nets, The Annals of Statistics, 36 (2008), pp. 2319–2343

    2008

  26. [27]

    A. B. Owen, Monte Carlo Theory, Methods and Examples , Stanford, 2013

    2013

  27. [28]

    Pyke , Spacings, Journal of the Royal Statistical Society

    R. Pyke , Spacings, Journal of the Royal Statistical Society. Series B (Methodological), 27 (1965), pp. 395–449

    1965

  28. [29]

    Rhee and P

    C.-H. Rhee and P. W. Glynn , Unbiased estimation with square root convergence for SDE models, Operations Research, 63 (2015), pp. 1026–1043

    2015

  29. [30]

    I. H. Sloan and H. Wo´zniakowski, When are quasi-Monte Carlo algorithms efficient for high dimensional integrals?, Journal of Complexity, 14 (1998), pp. 1–33

    1998

  30. [31]

    Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole , Score- based generative modeling through stochastic differential equations, Feb. 2021, https://doi. QMC FOR SDE SIMULATION 31 org/10.48550/arXiv.2011.13456, https://arxiv.org/abs/2011.13456

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2011.13456 2021

  31. [32]

    Vihola, Unbiased estimators and multilevel monte carlo , Operations Research, 66 (2018), pp

    M. Vihola, Unbiased estimators and multilevel monte carlo , Operations Research, 66 (2018), pp. 448–462

    2018

  32. [33]

    H. Wang, T. Zariphopoulou, and X. Y. Zhou , Reinforcement learning in continuous time and space: A stochastic control approach, Journal of Machine Learning Research, 21 (2020), pp. 1–34

    2020

  33. [34]

    Wang, On the effects of dimension reduction techniques on some high-dimensional problems in finance, Operations Research, 54 (2006), pp

    X. Wang, On the effects of dimension reduction techniques on some high-dimensional problems in finance, Operations Research, 54 (2006), pp. 1063–1078

    2006

  34. [35]

    Wang, Dimension reduction techniques in quasi-Monte Carlo methods for option pricing , INFORMS Journal on Computing, (2009)

    X. Wang, Dimension reduction techniques in quasi-Monte Carlo methods for option pricing , INFORMS Journal on Computing, (2009)

    2009

  35. [36]

    Wang, Handling discontinuities in financial engineering: Good path simulation and smooth- ing, Operations Research, 64 (2016), pp

    X. Wang, Handling discontinuities in financial engineering: Good path simulation and smooth- ing, Operations Research, 64 (2016), pp. 297–314

    2016

  36. [37]

    Wang and K.-T

    X. Wang and K.-T. Fang, The effective dimension and quasi-Monte Carlo integration, Journal of Complexity, 19 (2003), pp. 101–124

    2003

  37. [38]

    Wang and I

    X. Wang and I. H. Sloan , Quasi-Monte Carlo methods in financial engineering: An equiva- lence principle and dimension reduction , Operations Research, (2010)

    2010

  38. [39]

    Wang and K

    X. Wang and K. S. Tan , Pricing and hedging with discontinuous functions: Quasi–Monte Carlo methods and dimension reduction , Management Science, (2012)

    2012

  39. [40]

    Yong and X

    J. Yong and X. Y. Zhou , Stochastic Controls: Hamiltonian Systems and HJB Equations , New York, NY: Spinger, 1999

    1999