Giray Ökten

A central limit theorem and improved error bounds for a hybrid-Monte Carlo sequence with applications in computational finance

In problems of moderate dimensions, the quasi-Monte Carlo method usually provides better estimates than the Monte Carlo method. However, as the dimension of the problem increases, the advantages of the quasi-Monte Carlo method diminish quickly. A remedy for this problem is to use hybrid sequences; sequences that combine pseudorandom and low-discrepancy vectors. In this paper we discuss a particular hybrid sequence called the mixed sequence. We will provide improved discrepancy bounds for this sequence and prove a central limit theorem for the corresponding estimator. We will also provide numerical results that compare the mixed sequence with the Monte Carlo and randomized quasi-Monte Carlo methods.

Parallel Quasi-Monte Carlo Methods on a Heterogeneous Cluster

We introduce a new parallelization scheme for low-discrepancy sequences. Our approach is similar to the parameterization approach used for pseudorandom number generators. We present a theoretical analysis of this scheme and compare it numerically with the conventional blocking and leap-frogging parallelization strategies, when they are applied to problems from option pricing and transport theory. The numerical results suggest that our scheme might be very useful especially in distributed computing environments with unreliable and heterogeneous clusters.

Simulation Estimation of Mixed Discrete Choice Models with the Use of Randomized Quasi-Monte Carlo Sequences: A Comparative Study

The overall performance of the quasi-Monte Carlo (QMC) sequences proposed by Halton and Faure, as well as their scrambled versions, are numerically compared against each other and against the Latin hypercube sampling sequence in the context of the simulated likelihood estimation of a mixed multinomial logit model of choice. In addition, the efficiency of the QMC sequences generated with and without scrambling is compared across observations, and the performance of the Box-Muller and inverse normal transform procedures is tested. Numerical experiments were performed in five dimensions with 25, 125, and 625 draws and in 10 dimensions with 100 draws. Results indicate that the Faure sequence consistently outperforms the Halton sequence and that the scrambled versions of the Faure sequence perform best overall.

Generalized von Neumann-Kakutani transformation and random-start scrambled Halton sequences

It is a well-known fact that the Halton sequence exhibits poor uniformity in high dimensions. Starting with Braaten and Weller in 1979, several researchers introduced permutations to scramble the digits of the van der Corput sequences that make up the Halton sequence, in order to improve the uniformity of the Halton sequence. These sequences are called scrambled Halton, or generalized Halton sequences. Another significant result on the Halton sequence was the fact that it could be represented as the orbit of the von Neumann-Kakutani transformation, as observed by Lambert in 1982. In this paper, I will show that a scrambled Halton sequence can be represented as the orbit of an appropriately generalized von Neumann-Kakutani transformation. A practical implication of this result is that it gives a new family of randomized quasi-Monte Carlo sequences: random-start scrambled Halton sequences. This work generalizes random-start Halton sequences of Wang and Hickernell. Numerical results show that random-start scrambled Halton sequences can improve on the sample variance of random-start Halton sequences by factors as high as 7000.

Random and Deterministic Digit Permutations of the Halton Sequence

The Halton sequence is one of the classical low-discrepancy sequences. It is effectively used in numerical integration when the dimension is small, however, for larger dimensions, the uniformity of the sequence quickly degrades. As a remedy, generalized (scrambled) Halton sequences have been introduced by several researchers since the 1970s. In a generalized Halton sequence, the digits of the original Halton sequence are permuted using a carefully selected permutation. Some of the permutations in the literature are designed to minimize some measure of discrepancy, and some are obtained heuristically.In this paper, we investigate how these carefully selected permutations differ from a permutation simply generated at random. We use a recent genetic algorithm, test problems from numerical integration, and a recent randomized quasi-Monte Carlo method, to compare generalized Halton sequences with randomly chosen permutations, with the traditional generalized Halton sequences. Numerical results suggest that the random permutation approach is as good as, or better than, the “best” deterministic permutations.

Parametric uncertainty quantification in the Rothermel model with randomised quasi-Monte Carlo methods

Rothermels wildland surface fire model is a popular model used in wildland fire management. The original model has a large number of parameters, making uncertainty quantification challenging. In this paper, we use variance-based global sensitivity analysis to reduce the number of model parameters, and apply randomised quasi-Monte Carlo methods to quantify parametric uncertainties for the reduced model. The Monte Carlo estimator used in these calculations is based on a control variate approach applied to the sensitivity derivative enhanced sampling. The chaparral fuel model, selected from Rothermels 11 original fuel models, is studied as an example. We obtain numerical results that improve the crude Monte Carlo sampling by factors as high as three orders of magnitude.

Generating low-discrepancy sequences from the normal distribution: Box-Muller or inverse transform?

Quasi-Monte Carlo simulation is a popular numerical method in applications, in particular, economics and finance. Since the normal distribution occurs frequently in economic and financial modeling, one often needs a method to transform low-discrepancy sequences from the uniform distribution to the normal distribution. Two well known methods used with pseudorandom numbers are the Box-Muller and the inverse transformation methods. Some researchers and financial engineers have claimed that it is incorrect to use the Box-Muller method with low-discrepancy sequences, and instead, the inverse transformation method should be used. In this paper we prove that the Box-Muller method can be used with low-discrepancy sequences, and discuss when its use could actually be advantageous. We also present numerical results that compare Box-Muller and inverse transformation methods.

Parameterization based on randomized quasi-Monte Carlo methods

We present a theoretical framework where any randomized quasi-Monte Carlo method can be viewed and analyzed as a parameterization method for parallel quasi-Monte Carlo. We present deterministic and stochastic error bounds when different processors of the computing environment run at different speeds. We implement two parameterization methods, both based on randomized quasi-Monte Carlo, and apply them to pricing digital options and collateralized mortgage obligations. Numerical results are used to compare the parameterization methods by their parallel performance as well as their Monte Carlo efficiency.

On pricing discrete barrier options using conditional expectation and importance sampling Monte Carlo

Estimators for the price of a discrete barrier option based on conditional expectation and importance sampling variance reduction techniques are given. There are erroneous formulas for the conditional expectation estimator published in the literature: we derive the correct expression for the estimator. We use a simulated annealing algorithm to estimate the optimal parameters of exponential twisting in importance sampling, and compare it with a heuristic used in the literature. Randomized quasi-Monte Carlo methods are used to further increase the accuracy of the estimators.

Computation of the endogenous mortgage rates with randomized quasi-Monte Carlo simulations

The problem of computing the mortgage rate implied by a prepayment and interest rate model is considered. A Monte Carlo algorithm that uses a correlated sampling approach is introduced to simulate the model. Numerical results are used to compare Monte Carlo and randomized quasi-Monte Carlo methods with a numerical PDE solution. A particular randomized quasi-Monte Carlo method, random-start scrambled Halton sequence, gives superior performance, especially in high dimensions.