Pith. sign in

REVIEW 3 major objections 4 minor 2 cited by

Strong Stability Preservation for Stochastic Partial Differential Equations

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A contractive Euler-Maruyama step is enough for stochastic Runge-Kutta methods to preserve nonlinear stability, and bounded noise keeps mean-square order 1/2.

desk verdict The pathwise stochastic SSP idea is genuinely new, but Theorem 2.1's radius transfer uses the wrong increment scaling and fails for R(A)>1; the paper needs major repair before the SSP104 claims stand. read the letter →

arxiv 2411.11172 v2 pith:QUKI2SZJ submitted 2024-11-17 math.NA cs.NAmath.PR

James Woodfield This is my paper · ORCID
classification math.NAcs.NAmath.PR MSC 65L0665C3060H1565M20
keywords strongstabilitypreservationstochasticRunge-KuttamethodsEuler-Maruyamaschemeboundedincrementsmean-squareconvergencepartialdifferentialequationsslopelimitersoperatorsplitting
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

Deterministic strong stability preservation (SSP) is the art of proving that a timestepping scheme preserves nonlinear bounds such as positivity, total-variation bounds, or local maxima. This paper argues that the same logic works for stochastic differential and partial differential equations, provided three conditions hold: an Euler-Maruyama spatial discretisation that is itself contractive under a convex functional, noise increments that are bounded, and an SSP Runge-Kutta method with a positive radius of monotonicity. The payoff is a class of schemes whose pathwise solutions stay inside the same range as the exact solution, which matters for data assimilation and for any SPDE whose signal is known to satisfy a monotonicity property. The paper also proves that bounding the noise increments does not destroy convergence: schemes using the truncated-Gaussian increments retain mean-square order $1/2$.

What carries the argument

The central identity is the Euler-Maruyama/forward-Euler equivalence $FE(q,\Delta t,F)=EM(q,\Delta t,f,G,\Delta S)$ with $F(q)=f(q)+G(q)\Delta S/\Delta t$. It allows a stochastic Runge-Kutta stage to be read as a deterministic RK stage applied to $F$, so the Shu-Osher representation of the method becomes a convex combination of Euler-Maruyama maps. The algebraic conditions that make the convex combination valid are the Kraaijevanger conditions, encoded in the radius of monotonicity $R(A)$; Theorem 2.1 merely verifies those conditions and inherits contraction. The other half of the machinery is the bounded-increment construction of Definition 2.5, a symmetrically truncated Gaussian whose moment estimates control the distance between bounded- and normal-increment paths.

What would settle it

For the stochastic Burgers example, run the stochastic LLF Euler-Maruyama map from Definition 3.1 with the $N^2(K)\cup N(K)$ limiter and bounded increments, and search over paths and $\Delta t\le \tau_0$ for a cell value whose local maximum-principle range exceeds the previous range; any such realisation falsifies the monotone-EM premise on which Theorem 2.1 relies for that example.

Watch

Extended reading notes

Core claim

The central claim is Theorem 2.1: writing the combined field $F(q)=f(q)+G(q)\Delta S/\Delta t$ turns the Euler-Maruyama map into the forward Euler map, so the whole deterministic SSP theorem transfers unchanged. A stochastic Runge-Kutta method is contractive under any convex functional whenever the Euler-Maruyama map is, with the timestep condition $\Delta t \le R(A)\tau_0$, where $R(A)$ is the standard radius of monotonicity of the tableau. Additive methods inherit the property from separate forward-Euler and diffusion-only Euler-Maruyama maps, which is useful when a full monotone EM map cannot be built. Theorem 2.4 supplies the convergence half: with bounded increments truncated at $\sqrt{2k|\ln \Delta t|}$, $k\ge 1$, the local difference from normal-increment paths is small enough that the global mean-square order stays $1/2$ for stochastic additive Runge-Kutta methods, and the argument extends to generalised additive (GARK) schemes.

Load-bearing premise

The load-bearing assumption is that an Euler-Maruyama spatial discretisation that is contractive under some convex functional can actually be constructed for the target SPDE; for the paper's flagship Burgers example this contractivity rests on a stochastic LLF flux plus a limiter from the author's cited preprints rather than on a standalone proof from the SPDE.

Editorial extensions

If this is right

  • Any SSP Runge-Kutta method with $R(A)>0$ inherits the contractivity of a monotone Euler-Maruyama map under $\Delta t\le R(A)\tau_0$.
  • Stochastic additive Runge-Kutta methods inherit monotonicity from separate forward-Euler and diffusion-only Euler-Maruyama maps, so a full monotone EM scheme is not required.
  • Bounded increments with truncation parameter $k\ge 1$ keep mean-square convergence order $1/2$, and the two-point and three-point bounded distributions keep weak order 1.
  • Range-bounded numerical solutions are attainable for stochastic Burgers, advection, and 2D Euler with transport noise when all three sufficient conditions are used, as demonstrated numerically.
  • The three conditions are sufficient but not strictly necessary: one numerical example with non-SSP timestepping still stayed range-bounded, while another did not.

Reading between the lines

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

  • A testable extension is to use higher-moment bounded distributions beyond the two- and three-point examples; the proof structure suggests that any symmetric bounded noise with matching moments and small odd cross-moments should preserve both the SSP property and the convergence order.
  • For data assimilation, an implication left implicit in the paper is that every ensemble member of a monotone SPDE forecast remains in range pathwise, so no post-hoc clipping or projection step is needed to keep states physical.
  • If the monotone-EM premise can be proven for more SPDEs, such as monotone stochastic conservation laws with spatially varying fluxes, the same SSP-RK transfer would yield bounded schemes without designing a new stochastic Riemann solver, by using the additive or operator-split route.
  • The analysis does not quantify how the bounded-increment threshold scales with the number of RK stages; since the substage increment is scaled by $\Delta S/\sqrt{r}$, high-stage SSP methods may require tighter noise truncation, which is an open practical question.
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

3 major / 4 minor

Summary. The paper proposes a stochastic extension of strong stability preservation (SSP) for Runge-Kutta methods applied to SDEs and SPDEs. It defines an SSP property for the Euler-Maruyama map using bounded increments, and claims that stochastic Runge-Kutta, additive Runge-Kutta, and GARK methods inherit contractivity from the Euler-Maruyama scheme when their Butcher tableaus satisfy the usual Kraaijevanger conditions. It also proves (or attempts to prove) mean-square convergence order 1/2 for bounded-increment SARK methods, following the Milstein-Tretyakov truncation framework. The numerical sections demonstrate range-bounded solutions for stochastic Burgers, advection, and incompressible Euler equations, using slope limiters and bounded increments.

Significance. If the main theorems were correct, this would be a novel and potentially useful framework for nonlinear stability preservation in stochastic PDEs, with plausible applications in data assimilation and uncertainty quantification. The paper appears to be the first to extend deterministic SSP theory to the stochastic pathwise setting, and the convergence analysis for bounded increments extends known Milstein-Tretyakov results to additive Runge-Kutta methods. However, the central transfer theorem (Theorem 2.1) contains a scaling error that invalidates the claimed SSP radius transfer for every method with R(A)>1, including the SSP104 demonstration. The numerical experiments are suggestive but rely on spatial discretizations whose monotonicity is asserted via the author's own preprints rather than proven here. Thus the significance is currently conditional on a substantial revision.

major comments (3)
  1. [Section 2.3, Theorem 2.1 proof, Eqs. (2.20)-(2.21)] The proof rewrites the stochastic RK stage equations using EM(k_j, Δt/r, f, G, ΔS/r^{1/2}). Substituting the definition of EM (Method 2.1) gives r∑ a_ij EM(k_j, Δt/r, f, G, ΔS/r^{1/2}) = r∑ a_ij k_j + Δt∑ a_ij f(k_j) + √r ∑ a_ij G(k_j)ΔS, whose diffusion term differs from the original stage equation (2.11) by a factor √r. Equality with the original equation holds only for r=1. The correct algebraic identity requires EM(k_j, Δt/r, f, G, ΔS/r), since then the diffusion contribution becomes ∑ a_ij G(k_j)ΔS. As written, the claimed inheritance of contractivity with radius R(A)>1 is unproved, and the SSP104 demonstration with radius 6 (Method 3.3) is not covered by Theorem 2.1. Moreover, the practical methods in Section 3.1 use the Brownian-consistent scaling a^{1/2}G(q)ΔS (e.g., Method 3.3), which is not the scaling required by the corrected convex-combination representation; the implemented schemes must therefore be re-examined against any corrected theorem.
  2. [Section 2.4, Theorem 2.4 proof, Eq. (2.85)] Equation (2.85) contains a notational error involving E[Δ\tilde Z^p]. Since the bounded increments Δ\tilde Z have zero mean, the factor (1 - E[Δ\tilde Z^p]) would equal 1 and would not yield the claimed O(Δt^{k+1}) bound. The intended expression should involve E[(Δ\tilde Z)^2] (or the P-dimensional analogue), consistent with inequality (2.49) used immediately afterward in Eq. (2.86). The proof should be corrected to make the moment index explicit. Separately, the assertion that the underbraced term in Eq. (2.88) vanishes "due to Isserlis theorem and symmetry" is not demonstrated for the truncated normal distribution; the conditioning argument on the truncation events (|Z|<A, Z>A, Z<-A) is only sketched. Since Theorem 2.4 is the main convergence result for bounded-increment SARK methods, these points need to be resolved before the theorem can be accepted as proven.
  3. [Section 3.2.1, Method 3.6 and the Burgers example] The numerical demonstration of the flagship Burgers example relies on a spatial discretization (Method 3.6) whose monotonicity is asserted via a stochastic LLF flux and the N2(K)∪N(K) limiter. However, the paper itself states in the same subsection that "the numerical method described above is not proven to be convergent or even claimed to be sensible for this equation." Thus the example does not provide a verified instance of condition 1 (a provably monotone Euler-Maruyama scheme); it is a plausibility demonstration pending the author's preprints [43,44]. This caveat is mentioned locally but should be stated more prominently in the conclusions, since the numerical evidence is a central part of the paper's practical claims.
minor comments (4)
  1. [Global] There are numerous typographical and grammatical errors throughout, including "bouned", "Kraiijevanger" for Kraaijevanger, "Ruemelin" for Rümelin, inconsistent use of ΔW versus ΔS in equations (1.5) and (2.9), and equation numbering inconsistencies in the appendices. A careful proofreading pass is needed.
  2. [Theorem 2.2 proof, Eq. (2.39)] Equation (2.39) introduces a symbol S̃ that is never defined; in the additive RK augmented system it should be S. This appears to be a typographical error in the derivation of the additive RK convex combination.
  3. [Theorem 2.4 proof, Eq. (2.87)] Equation (2.87) uses (∑_{i=1}^s b_i)^2 for the leading-order squared term, but the corresponding leading term in Eq. (2.81) is proportional to ∑ \tilde b_i, not ∑ b_i. This should be corrected to (∑ \tilde b_i)^2.
  4. [Section 2.4, Definition 2.5] The notation Δ\tilde Z_Δt is defined only for a scalar increment, while the paper later applies it componentwise to P-dimensional increments. It should be clarified that the definition and the estimates (2.49)-(2.50) are applied componentwise, with the P-dimensional extension being straightforward.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: The core theorems are conditional transfers of deterministic SSP theory and Milstein–Tretyakov convergence results; the only self-citations support illustrative limiter numerics, not the derivation chain.

full rationale

The paper's main theorems are conditional transfers of deterministic SSP theory, not self-referential predictions. Theorem 2.1 explicitly assumes the EM contractivity stated in Definition 2.1 ('Assuming there exists a critical timestep tau0 such that EM is assumed contractive with respect to a convex functional'), then rewrites Method 2.2 using the identification F(q) = f(q) + G(q)DeltaS/Delta t and invokes the external Kraaijevanger/Higueras conditions, so the RK contractivity conclusion follows from genuine deterministic results [28,14] rather than from itself. Theorem 2.4 is anchored to the independent Milstein–Tretyakov fundamental theorem on mean-square convergence [35] and their bounded-increment construction [34,35]; the paper supplies Taylor-expansion estimates and does not rename its output as an input. The self-references [43,44] appear only in the numerical section to support the slope-limiter local maximum principle, and the paper explicitly disclaims convergence of that discretization, so these citations support illustrative numerics rather than the core proof chain. The scaling mismatch in eqs. (2.20)–(2.21) flagged in review is a correctness issue (the convex-combination identity fails for the implemented DeltaS/sqrt(r) scaling), not a case of a conclusion being equivalent to its hypotheses, so it is outside the circularity scope of this pass.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the assumption that a monotone Euler-Maruyama scheme exists for the target SPDE, which is asserted for the numerical tests but not proven there; it also rests on standard well-posedness, Taylor remainders, and the Milstein-Tretyakov truncation estimates. No free parameters are fitted to data in the theory.

free parameters (1)
  • k = k ≥ 1 (e.g., k=1)
    Exponent in the Milstein-Tretyakov truncated increment bound A_Δt = sqrt(2k|ln Δt|); chosen by the user, not fitted, and the convergence result holds for any k ≥ 1.
assumptions (5)
  • standard math Lipschitz continuity and linear growth bounds on drift and diffusion (eqs. 2.6-2.7) ensure well-posedness of the SDE.
    Invoked in Section 2.2 and used for growth bounds in convergence proofs.
  • domain assumption A monotone Euler-Maruyama flow map exists for the SPDE under consideration (Definition 2.1).
    Central premise of Theorem 2.1; the paper does not prove this for its numerical examples and defers to self-cited limiter papers [43,44].
  • domain assumption Bounded random increments whose moments match the normal distribution preserve weak and strong convergence (Definitions 2.5 and Examples 2.1-2.2).
    Needed for the SSP property and for the convergence analysis in Section 2.4; the bound in Definition 2.5 grows like sqrt(ln Δt) as Δt tends to 0.
  • standard math Local error estimates (Assumption 2.1) and the Fundamental Theorem on Mean Square Convergence [35] hold.
    Used to convert one-step estimates into global mean-square order 1/2 in Theorem 2.4.
  • domain assumption The N2(K)∪N(K) limiter from [43] and the one-dimensional limiter from [44] enforce discrete local maximum principles.
    Used in Section 3 to claim monotonicity of the EM map; these are prior preprints by the same author, not independently verified here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Strong Stability Preservation for Stochastic Partial Differential Equations." pith.science (2026). https://pith.science/paper/QUKI2SZJ

@misc{pith2026241111172,
  author       = {Pith},
  title        = {Pith review of: Strong Stability Preservation for Stochastic Partial Differential Equations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QUKI2SZJ}},
  note         = {Machine review of arXiv:2411.11172}
}
read the original abstract

This paper extends deterministic notions of Strong Stability Preservation (SSP) to the stochastic setting, enabling nonlinearly stable numerical solutions to stochastic differential equations (SDEs) and stochastic partial differential equations (SPDEs) with pathwise solutions that remain unconditionally bounded. This approach may offer modelling advantages in data assimilation, particularly when the signal or data is a realization of an SPDE or PDE with a monotonicity property.

Figures

Figures reproduced from arXiv: 2411.11172 by the authors.

Figure 3.1
Figure 3.1. We plot the final time plot, of the stochastic Burgers equation in row 1. Red indicates maxima over 1, blue indicates [PITH_FULL_IMAGE:figures/full_fig_p019_3_1.png] view at source ↗
Figure 3.2
Figure 3.2. In figs. 3.2a and 3.2b we plot the final time solution of the γ = 1/4, γ = 3/4 respectively. In figs. 3.2c and 3.2d the maximum value is plotted with time for the γ = 1/4, γ = 3/4 respectively. In figs. 3.1g, 3.1k and 3.2b, one finds that for one particular scheme, in one particular example, running beyond the radius of monotonicity (using a Non-SSP timestepper) was not a strictly necessary condition for the range b… view at source ↗
Figure 3.3
Figure 3.3. We plot 16 ensemble member solutions of the final timestep on a perceptually uniform grey colour scale between [0,1] [PITH_FULL_IMAGE:figures/full_fig_p021_3_3.png] view at source ↗
Figures from the paper (3 more)
Figure 3.4
Figure 3.4. Figure 3.4: Stochastic-2D Euler. We plot 8 ensemble member solutions of the final timestep on a perceptually uniform grey colorscale [PITH_FULL_IMAGE:figures/full_fig_p022_3_4.png]
Figure 3.5
Figure 3.5. Figure 3.5: Deterministic 2D Euler. We plot a 1 ensemble member solution of the deterministic incompressible Euler equation at [PITH_FULL_IMAGE:figures/full_fig_p023_3_5.png]
Figure 3.6
Figure 3.6. Figure 3.6: Burgers equation with deformational transport noise. We plot 16 ensemble member solutions of the final timestep on a [PITH_FULL_IMAGE:figures/full_fig_p024_3_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Discontinuous Galerkin methods for the complete stochastic Euler equations

    math.NA 2024-12 conditional novelty 6.0 of 10

    An entropy-dissipative DG method for the stochastic full Euler system is shown to converge in law to a dissipative martingale solution, with error bounds up to a stopping time.

  2. Numerical comparison of energy- versus circulation-preserving stochastic vortex dynamics

    physics.flu-dyn 2026-06 unverdicted novelty 5.0 of 10

    SALT perturbations localize near vorticity gradients while SFLT produces diffuse variance across the domain in stochastic 2D Euler vortex dynamics.

Reference graph

Works this paper leans on

48 extracted references · 45 canonical work pages · cited by 2 Pith papers

  1. [1]

    W.-J. Beyn, E. Isaak, and R. Kruse. Stochastic C-stability and B-consistency of explicit and implicit Milstein-type schemes. Journal of Scientific Computing , 70:1042–1077, 2017

    work page 2017

  2. [2]

    Bolley and M

    C. Bolley and M. Crouzeix. Conservation de la positivit´ e lors de la discr´ etisation des probl` emes d’´ evolution paraboliques. RAIRO. Analyse num´ erique, 12(3):237–245, 1978

    work page 1978

  3. [3]

    Burrage, P

    K. Burrage, P. Burrage, and T. Tian. Numerical methods for strong solutions of stochastic differential equations: an overview. Proceedings of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences, 460(2041):373–402, 2004

    work page 2004

  4. [4]

    Conde, S

    S. Conde, S. Gottlieb, Z. J. Grant, and J. N. Shadid. Implicit and implicit–explicit strong stability preserving Runge–Kutta methods with high linear order. Journal of Scientific Computing , 73:667–690, 2017

    work page 2017

  5. [5]

    Cotter, D

    C. Cotter, D. Crisan, D. D. Holm, W. Pan, and I. Shevchenko. A particle filter for stochastic advection by lie transport: a case study for the damped and forced incompressible two-dimensional euler equation. SIAM/ASA Journal on Uncertainty Quantification , 8(4):1446–1492, 2020

    work page 2020

  6. [6]

    D´ avila, J

    J. D´ avila, J. F. n. Bonder, J. D. Rossi, P. Groisman, and M. Sued. Numerical analysis of stochastic differential equations with explosions. Stochastic Analysis and Applications , 23(4):809–825, 2005

    work page 2005

  7. [7]

    S. Fang, W. Zhao, and T. Zhou. Strong Stability Preserving Multistep Schemes for Forward Backward Stochastic Differential Equations. Journal of Scientific Computing , 94(3):53, 2023

    work page 2023

  8. [8]

    Ferracina and M

    L. Ferracina and M. Spijker. Stepsize restrictions for total-variation-boundedness in general Runge–Kutta procedures. Applied Numerical Mathematics , 53(2-4):265–279, 2005

    work page 2005

Show all 48 references
  1. [9]

    Ferracina and M

    L. Ferracina and M. N. Spijker. Stepsize restrictions for the total-variation-diminishing property in general runge–kutta methods. SIAM journal on numerical analysis , 42(3):1073–1093, 2004

    2004

  2. [10]

    Gottlieb

    S. Gottlieb. On high order strong stability preserving Runge-Kutta and multi step time discretizations. Journal of scientific computing , 25:105–128, 2005

    2005

  3. [11]

    Gottlieb and C.-W

    S. Gottlieb and C.-W. Shu. Total variation diminishing Runge-Kutta schemes.Mathematics of computation, 67(221):73–85, 1998

    1998

  4. [12]

    D. J. Higham, X. Mao, and L. Szpruch. Convergence, non-negativity and stability of a new Milstein scheme with applications to finance. arXiv preprint arXiv:1204.1647 , 2012

    2012 arXiv

  5. [13]

    Higueras

    I. Higueras. On strong stability preserving time discretization methods. Journal of Scientific Computing , 21:193–223, 2004. 25

    2004

  6. [14]

    Higueras

    I. Higueras. Representations of Runge–Kutta methods and strong stability preserving methods. SIAM journal on numerical analysis , 43(3):924–948, 2005

    2005

  7. [15]

    Higueras

    I. Higueras. Strong stability for additive Runge–Kutta methods. SIAM journal on numerical analysis , 44(4):1735–1758, 2006

    2006

  8. [16]

    Higueras, D

    I. Higueras, D. I. Ketcheson, and T. A. Kocsis. Optimal monotonicity-preserving perturbations of a given Runge–Kutta method. Journal of Scientific Computing , 76:1337–1369, 2018

    2018

  9. [17]

    Higueras and T

    I. Higueras and T. Rold´ an. New third order low-storage SSP explicit Runge–Kutta methods. Journal of Scientific Computing , 79:1882–1906, 2019

    1906

  10. [18]

    D. D. Holm. Variational principles for stochastic fluid dynamics. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences , 471(2176):20140963, 2015

    2015

  11. [19]

    Hundsdorfer and M

    W. Hundsdorfer and M. Spijker. Boundedness and strong stability of runge-kutta methods. Mathematics of computation, 80(274):863–886, 2011

    2011

  12. [20]

    Karatzas and S

    I. Karatzas and S. Shreve. Brownian motion and stochastic calculus , volume 113. springer, 2014

    2014

  13. [21]

    Kelly, A

    C. Kelly, A. Rodkina, and E. M. Rapoo. Adaptive timestepping for pathwise stability and positivity of strongly discretised nonlinear stochastic differential equations. Journal of Computational and Applied Mathematics, 334:39–57, 2018

    2018

  14. [22]

    Ketcheson

    D. Ketcheson. An algebraic characterization of strong stability preserving Runge-Kutta schemes. Under- graduate Thesis, Brigham Young University, Provo, Utah, USA , 2004

    2004

  15. [23]

    D. I. Ketcheson. Highly efficient strong stability-preserving Runge–Kutta methods with low-storage im- plementations. SIAM Journal on Scientific Computing , 30(4):2113–2136, 2008

    2008

  16. [24]

    D. I. Ketcheson and A. C. Robinson. On the practical importance of the SSP property for Runge–Kutta time integrators for some common Godunov-type schemes. International Journal for Numerical Methods in Fluids , 48(3):271–303, 2005

    2005

  17. [25]

    Kiouvrekis and I

    Y. Kiouvrekis and I. S. Stamatiou. Domain preserving and strongly converging explicit scheme for the stochastic SIS epidemic model. Journal of Computational and Applied Mathematics , 456:116219, 2025

    2025

  18. [26]

    Kloeden and E

    P. Kloeden and E. Platen. Numerical Solution of Stochastic Differential Equation . Berlin; Springer-Verlag, 1992

    1992

  19. [27]

    P. E. Kloeden, E. Platen, P. E. Kloeden, and E. Platen. Stochastic differential equations. Springer, 1992

    1992

  20. [28]

    J. F. B. M. Kraaijevanger. Contractivity of runge-kutta methods. BIT Numerical Mathematics, 31(3):482– 528, 1991

    1991

  21. [29]

    Kurganov and E

    A. Kurganov and E. Tadmor. New high-resolution central schemes for nonlinear conservation laws and convection–diffusion equations. Journal of computational physics , 160(1):241–282, 2000

    2000

  22. [30]

    Z. Lei, S. Gan, and Z. Chen. Strong and weak convergence rates of logarithmic transformed truncated EM methods for SDEs with positive solutions. Journal of Computational and Applied Mathematics , 419:114758, 2023

    2023

  23. [31]

    R. J. Leveque. High-resolution conservative algorithms for advection in incompressible flow. SIAM Journal on Numerical Analysis , 33(2):627–665, 1996

    1996

  24. [32]

    G. J. Lord, C. E. Powell, and T. Shardlow. An introduction to computational stochastic PDEs , volume 50. Cambridge University Press, 2014. 26

    2014

  25. [33]

    G. N. Milstein. Numerical integration of stochastic differential equations , volume 313. Springer Science & Business Media, 2013

    2013

  26. [34]

    G. N. Milstein, Y. M. Repin, and M. V. Tretyakov. Numerical methods for stochastic systems preserving symplectic structure. SIAM Journal on Numerical Analysis , 40(4):1583–1604, 2002

    2002

  27. [35]

    G. N. Milstein and M. V. Tretyakov. Stochastic numerics for mathematical physics , volume 39. Springer, 2004

    2004

  28. [36]

    Oksendal

    B. Oksendal. Stochastic differential equations: an introduction with applications . Springer Science & Business Media, 2013

    2013

  29. [37]

    R¨ oßler

    A. R¨ oßler. Runge–Kutta methods for the strong approximation of solutions of stochastic differential equations. SIAM Journal on Numerical Analysis , 48(3):922–952, 2010

    2010

  30. [38]

    Sandu and M

    A. Sandu and M. G¨ unther. A generalized-structure approach to additive Runge–Kutta methods. SIAM Journal on Numerical Analysis , 53(1):17–42, 2015

    2015

  31. [39]

    C. Scalone. Positivity preserving stochastic θ-methods for selected SDEs. Applied Numerical Mathematics, 172:351–358, 2022

    2022

  32. [40]

    Shu and S

    C.-W. Shu and S. Osher. Efficient implementation of essentially non-oscillatory shock-capturing schemes. Journal of computational physics , 77(2):439–471, 1988

    1988

  33. [41]

    M. Spijker. Contractivity in the numerical solution of initial value problems. Numerische Mathematik , 42:271–290, 1983

    1983

  34. [42]

    J. H. Williamson. Low-storage Runge-Kutta schemes. Journal of computational physics, 35(1):48–56, 1980

    1980

  35. [43]

    Woodfield

    J. Woodfield. Higher Order Multidimensional Slope Limiters with Local Maximum Principles. arXiv preprint arXiv:2407.06437, 2024

    2024 arXiv

  36. [44]

    Woodfield, H

    J. Woodfield, H. Weller, and C. J. Cotter. New limiter regions for multidimensional flows. arXiv preprint arXiv:2402.08395, 2024. Appendix A. Appendix Appendix A.1. Big O notation We understand the big-O(∆t)-notation in the limit ∆t → 0, more specifically, we say f (∆t) = O(g(...

    2024 arXiv

  37. [45]

    (A.1) Then in the vector-valued case, wheref is made of components f k, one can write down the following component- wise Multivariate Taylor theorem of the following form, with an error estimate on the remainder f (q + a) = f (q) + Df |qa + Rf (q, a), where ||Rf (q, a)||2 ≤ L||a||2

  38. [46]

    gp(q + a) = gp(q) + Dgp|q(a) + Rg(q, a), where ||Rg(q, a)||2 ≤ L||a||2

    (A.2) Similarly one can write this for the vector gp, the p-th component of G. gp(q + a) = gp(q) + Dgp|q(a) + Rg(q, a), where ||Rg(q, a)||2 ≤ L||a||2

  39. [47]

    (A.3) A matrix valued version can be written for G in Rd×P , G(q + a) = G(q) + DG|q(a) + RG(q, a), where ||RG(q, a)||2 ≤ L||a||2

  40. [48]

    Inequalities We tabulate some additional estimates on the moments of the bounded normal increments in [34, 35]

    (A.4) 27 Appendix A.3. Inequalities We tabulate some additional estimates on the moments of the bounded normal increments in [34, 35]. All odd moments are zero E[(∆ eZ)2m+1] = 0, ∀m = {1, ..., M}. The moment estimate in [34], can be attained by computing E[(∆Z − ∆ eZ)2] = 2√ 2...

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.