Analysis of nested multilevel Monte Carlo using approximate Normal random variables

@article{Giles2022AnalysisON,
  title={Analysis of nested multilevel Monte Carlo using approximate Normal random variables},
  author={Michael B. Giles and Oliver Sheridan-Methven},
  journal={SIAM/ASA J. Uncertain. Quantification},
  year={2022},
  volume={10},
  pages={200-226}
}
The multilevel Monte Carlo (MLMC) method has been used for a wide variety of stochastic applications. In this paper we consider its use in situations in which input random variables can be replaced by similar approximate random variables which can be computed much more cheaply. A nested MLMC approach is adopted in which a twolevel treatment of the approximated random variables is embedded within a standard MLMC application. We analyse the resulting nested MLMC variance in the specific context… 

Figures from this paper

Multilevel Monte Carlo Using Approximate Distribution of the CIR Process

Giles and Sheridan-Methven propose Multi-level Monte Carlo (MLMC) with approximate normal random variables. In this article, we develop this approach by embedding the approximate CIR process, which

Rounding error using low precision approximate random variables

TLDR
It is found single precision offers the potential for approximate speed improvements by a factor of 7 across a wide span of discretisation levels, and half precision offers comparable improvements for several levels of coarse simulations, and even offers improvements for the very coarsest few levels.

Approximating inverse cumulative distribution functions to produce approximate random variables

For random variables produced through the inverse transform method, approximate random variables are introduced, which are produced by approximations to a distribution's inverse cumulative

References

SHOWING 1-10 OF 37 REFERENCES

Multilevel Monte Carlo Path Simulation

We show that multigrid ideas can be used to reduce the computational complexity of estimating an expected value arising from a stochastic differential equation using Monte Carlo path simulations. In

Multilevel Monte Carlo Methods

TLDR
Applications to stochastic solution of integral equations are given for the case where an approximation of the full solution function or a family of functionals of the solution depending on a parameter of a certain dimension is sought.

Monte Carlo Complexity of Global Solution of Integral Equations

TLDR
The Monte Carlo complexity, i.e., the complexity of the stochastic solution of this problem, is analyzed and the results show that even in the global case Monte Carlo algorithms can perform better than deterministic ones, although the difference is not as large as in the local case.

Approximating inverse cumulative distribution functions to produce approximate random variables

For random variables produced through the inverse transform method, approximate random variables are introduced, which are produced by approximations to a distribution's inverse cumulative

Numerical Solution of Stochastic Differential Equations

This paper provides an introduction to the main concepts and techniques necessary for someone who wishes to carryout numerical experiments involving Stochastic Differential Equation (SDEs). As SDEs

Multilevel Estimation of Rare Events

TLDR
A multilevel estimator is introduced which is based on and generalizes the idea of subset simulation and employs a hierarchy of approximations to the system response computed with different resolutions, leading to reduced computational costs compared to subset simulation.

Integral Inequalities for Convex Functions of Operators on Martingales

Let M be a family of martingales on a probability space (Ω, A, P) and let ф be a nonnegative function on [0, ∞]. The general question underlying both [2] and the present work may be stated as follows

Exploiting Mixed-Precision Arithmetics in a Multilevel Monte Carlo Approach on FPGAs

TLDR
A novel methodology to speed up Multilevel Monte Carlo (MLMC) simulations on reconfigurable architectures by developing a novel heuristic for selecting an appropriate precision at each stage of the simulation that can be executed with low costs at runtime.

Random bit multilevel algorithms for stochastic differential equations