Abstract:Despite the success of adaptive time-stepping in ODE simulation, it has so far seen few applications for Stochastic Differential Equations (SDEs). To simulate SDEs adaptively, methods such as the Virtual Brownian Tree (VBT) have been developed, which can generate Brownian motion (BM) non-chronologically. However, in most applications, knowing only the values of Brownian motion is not enough to achieve a high order of convergence; for that, we must compute time-integrals of BM such as $\int_s^t W_r \, dr$. With the aim of using high order SDE solvers adaptively, we extend the VBT to generate these integrals of BM in addition to the Brownian increments. A JAX-based implementation of our construction is included in the popular Diffrax library (this https URL). Since the entire Brownian path produced by VBT is uniquely determined by a single PRNG seed, previously generated samples need not be stored, which results in a constant memory footprint and enables experiment repeatability and strong error estimation. Based on binary search, the VBT's time complexity is logarithmic in the tolerance parameter $\varepsilon$. Unlike the original VBT algorithm, which was only precise at some dyadic times, we prove that our construction exactly matches the joint distribution of the Brownian motion and its time integrals at any query times, provided they are at least $\varepsilon$ apart. We present two applications of adaptive high order solvers enabled by our new VBT. Using adaptive solvers to simulate a high-volatility CIR model, we achieve more than twice the convergence order of constant stepping. We apply an adaptive third order underdamped or kinetic Langevin solver to an MCMC problem, where our approach outperforms the No U-Turn Sampler, while using only a tenth of its function evaluations.

Approximating the signature of Brownian motion for high order SDE simulation

A representation for the Expected Signature of Brownian motion up to the first exit time of the planar unit disc

Path-dependent processes from signatures

Exact Weak Convergence Rates for Signatures of Brownian Motion

A PDE approach for solving the characteristic function of the generalised signature process

Small mass limit of expected signature for physical Brownian motion

Stratonovich's Signatures of Brownian Motion Determine Brownian Sample Paths

Concentration and Exact Convergence Rates for Expected Brownian Signatures

Single-seed generation of Brownian paths and integrals for adaptive and high order SDE solvers

Modelling pathwise uncertainty of Stochastic Differential Equations samplers via Probabilistic Numerics

Stochastic Differential Equation Driven by Countably Many Brownian Motions with Non-Lipschitzian Coefficients

Asymptotic behaviors for distribution dependent SDEs driven by fractional Brownian motions

Approximation of SDEs: a stochastic sewing approach

Almost sure approximations and laws of iterated logarithm for signatures

Numerical approximation of SDEs with fractional noise and distributional drift

Stochastic analysis of fractional brownian motions

Pathwise Unique Solutions and Stochastic Averaging for Mixed Stochastic Partial Differential Equations Driven by Fractional Brownian Motion and Brownian Motion

Deviation frequencies of Brownian path property approximations

Stochastic analysis, rough path analysis and fractional Brownian motions

A High Order Solver for Signature Kernels

The rate of convergence of Euler approximations for solutions of stochastic differential equations driven by fractional Brownian motion