Submission rejected on 21 September 2025 by Ldm1954 (talk). This submission is contrary to the purpose of Wikipedia. Rejected by Ldm1954 13 days ago. Last edited by Ldm1954 13 days ago. Ask for advice

• Comment: Wikipedia is not a place to advertise your recent math results, particularly when the work has not passed peer review. We only have established material which multiple, independent reputable sources have shown to be notable. If the paper is accepted and in 2 years has multiple citations (> 100) then revise and resubmit. Ldm1954 (talk) 14:58, 21 September 2025 (UTC)

This article has multiple issues. Please help improve it or discuss these issues on the talk page. (Learn how and when to remove these messages) This article contains promotional content. Please help improve it by removing promotional language and inappropriate external links, and by adding encyclopedic text written from a neutral point of view. (September 2025) (Learn how and when to remove this message) A major contributor to this article appears to have a close connection with its subject. It may require cleanup to comply with Wikipedia's content policies, particularly neutral point of view. Please discuss further on the talk page. (September 2025) (Learn how and when to remove this message) This article relies largely or entirely on a single source. Relevant discussion may be found on the talk page. Please help improve this article by introducing citations to additional sources. Find sources: "Duffield Discretization" – news · newspapers · books · scholar · JSTOR (September 2025) (Learn how and when to remove this message)

The Duffield discretization (also called lattice random walk discretization) is a method for approximately sampling trajectories from stochastic differential equations (SDEs). It is related to not to be confused with distinct from lattice random walk models of physical processes; in contrast the Duffield discretization samples to generic SDEs in the same way as the Euler-Maruyama discretization. The method was introduced by Samuel Duffield (and coauthors) at Normal Computing.^[1]

The method differs notably from Euler-Maruyama (and discretizations such as Runge-Kutta methods) through sampling increments in a binary ${\displaystyle \{-\delta _{x},\delta _{x}\}}$ or ternary ${\displaystyle \{-\delta _{x},0,\delta _{x}\}}$ space for some specified parameter ${\displaystyle \delta _{x}}$ rather than the full continuum ${\displaystyle [-\infty ,\infty ]}$. Yet has the same weak order 1 convergence as the Euler-Maruyama method.

Consider a multivariate SDE$${\displaystyle dX_{t}=a(X_{t},t)dt+b(X_{t},t)dW_{t},}$$with intial condition ${\displaystyle X_{0}=x_{0}}$, Weiner process ${\displaystyle W_{t}}$ and we want to solve the SDE on some time interval ${\displaystyle [0,T]}$. The Duffield discretisation requires the restriction that the diffusion matrix ${\displaystyle b(X_{t},t)}$ is diagonal.

The Duffield discretisation in its more general ternary definition has two hyperparameters. The first is a scalar temporal stepsize ${\displaystyle \delta _{t}}$ which is shared by all SDE discretizations and recovers the true SDE as ${\displaystyle \delta _{t}\to 0}$. The second is a spatial vector ${\displaystyle \delta _{x}(X,t)}$ which controls the size of the ternary steps.

The simplest form of the Duffield discretization sets ${\displaystyle \delta _{x}(X,t)={\sqrt {\delta _{t}}}b(X,t)}$. The binary Duffield approximation to the true solution ${\displaystyle X}$ is the Markov chain ${\displaystyle Y}$ defined as follows:

• Partition the interval ${\displaystyle [0,T]}$ into ${\displaystyle N}$ equal subintervals of width ${\displaystyle \delta _{t}=T/N}$:

${\displaystyle 0=\tau _{0}<\tau _{1}<\dots <\tau _{N}=T.}$

• Set ${\displaystyle Y_{0}=x_{0}}$.
• Recursively set ${\displaystyle Y_{n}}$ for ${\displaystyle 0\leq n\leq N-1}$ by$${\displaystyle Y_{n+1}=Y_{n}+\Delta (Y_{n},\tau _{n}),}$$where for each coordinate ${\displaystyle i}$ we have an independent binary random variable$${\displaystyle \mathbb {P} [\Delta _{i}(Y,\tau )=\Delta _{i}]={\begin{cases}1-p_{i}(Y,\tau ),&{\text{if }}\Delta _{i}=-{\sqrt {\delta _{t}}}b_{i}(Y,\tau ),\\p_{i}(Y,\tau ),&{\text{if }}\Delta _{i}={\sqrt {\delta _{t}}}b_{i}(Y,\tau ).\end{cases}}}$$Here the probability vector is defined as$${\displaystyle p(Y,\tau )={\frac {1}{2}}+{\frac {1}{2}}{\sqrt {\delta _{t}}}b(Y,\tau )^{-1}a(Y,\tau ).}$$

The more general ternary form allows the user to set ${\displaystyle \delta _{x}(X,t)}$ although weak order 1 convergence requires ${\displaystyle \delta _{x}(X,t)=\Theta ({\sqrt {\delta _{t}}})}$, that is ${\displaystyle \delta _{x}(X,t)}$ as a function of ${\displaystyle \delta _{t}}$ grows on the same order as ${\displaystyle {\sqrt {\delta _{t}}}}$.

The ternary Duffield approximation is the Markov chain defined as above but with more general random increments

• For each coordinate ${\displaystyle i}$ we have an independent ternary random variable$${\displaystyle \mathbb {P} [\Delta _{i}(Y,\tau )=\Delta _{i}]={\begin{cases}p_{-,i}(Y,\tau ),&{\text{if }}\Delta _{i}=-\delta _{x,i},\\1-p_{-,i}(Y,\tau )-p_{+,i}(Y,\tau ),&{\text{if }}\Delta _{i}=0,\\p_{+,i}(Y,\tau ),&{\text{if }}\Delta _{i}=\delta _{x,i}.\end{cases}}}$$With probability vectors defined as$${\displaystyle p_{\pm }(Y,\tau )={\frac {1}{2}}\delta _{t}\delta _{x}^{-1}[\pm a(Y,\tau )+\delta _{x}^{-1}b(Y,\tau )^{2}].}$$

The advantage of the ternary generalization is the decoupling of ${\displaystyle \delta _{x}(X,t)}$ from the SDE's diffusion coefficient ${\displaystyle b(X,t)}$. For example, in the ternary scheme one can set a constant ${\displaystyle \delta _{x}}$ and retain a valid discretization even if ${\displaystyle b(X,t)}$ varies as a function of space ${\displaystyle X}$ and/or time ${\displaystyle t}$. However, a general rule of thumb is to set ${\displaystyle \delta _{x}(X,t)={\sqrt {\delta _{t}}}b(X,t)}$ to recover the binary case or close to for performance and robustness.^[1]

The spatially discrete nature of the Duffield discretization is a marked departure from traditional discretizations such as Euler-Maruyama. This brings several advantages

• No requirement on Gaussian random variate generation. The binary or ternary random sampling is significantly simpler than Gaussian sampling which requires e.g. the Box-Muller transform or the Ziggurat algorithm. Gaussian samples are requried by majority of SDE discretisations (including Euler-Maruyama and Runge-Kutta).
• Inherent error mitigation due to the clamping of the drift and diffusion coefficients to only a discrete increment, in contrast to traditional methods that assume infinite precision. Errors in the computation drift and diffusion arise from e.g. numerical quantization.
• Robustness in the prescence of non-globally-Lipshitz drift coefficients, where methods such as Euler-Maruyama are known to peform poorly.^[2].
• Compatability with stochastic computing models of computation. For example, in the binary case the generation of the increment can be considered a vector of bipolar stochastic numbers. Stochastic computing can be used to generate such random variables with specified expectations using stochastic rather than floating point arithmetic which can be significantly more efficient. The Duffield discretisation offers the advantage over traditional stochastic computing of only requiring a single random output rather than long bitstreams that need to be accumulated, this can viewed as an example of dynamic stochastic computing^[3]

A significant disadvantage is the restriction of the diffusion matrix to be diagonal.