Alexander D. Gilbert,Frances Y. Kuo,Abirami Srikumar
Abstract:In this paper, we apply quasi-Monte Carlo (QMC) methods with an initial preintegration step to estimate cumulative distribution functions and probability density functions in uncertainty quantification (UQ). The distribution and density functions correspond to a quantity of interest involving the solution to an elliptic partial differential equation (PDE) with a lognormally distributed coefficient and a normally distributed source term. There is extensive previous work on using QMC to compute expected values in UQ, which have proven very successful in tackling a range of different PDE problems. However, the use of QMC for density estimation applied to UQ problems will be explored here for the first time. Density estimation presents a more difficult challenge compared to computing the expected value due to discontinuities present in the integral formulations of both the distribution and density. Our strategy is to use preintegration to eliminate the discontinuity by integrating out a carefully selected random parameter, so that QMC can be used to approximate the remaining integral. First, we establish regularity results for the PDE quantity of interest that are required for smoothing by preintegration to be effective. We then show that an $N$-point lattice rule can be constructed for the integrands corresponding to the distribution and density, such that after preintegration the QMC error is of order $\mathcal{O}(N^{-1+\epsilon})$ for arbitrarily small $\epsilon>0$. This is the same rate achieved for computing the expected value of the quantity of interest. Numerical results are presented to reaffirm our theory.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is: how to use pre - integration and the Quasi - Monte Carlo (QMC) method to estimate the probability density function (PDF) and cumulative distribution function (CDF) of the solutions of elliptic partial differential equations (PDEs) with random inputs. Specifically, this paper focuses on elliptic PDEs with log - normally distributed diffusion coefficients and normally distributed source terms.
### Background and Problem Description
In the field of Uncertainty Quantification (UQ), many studies have successfully used the QMC method to calculate expected values. However, for the estimation of probability density functions and cumulative distribution functions, due to the existence of discontinuities in the integral expressions, this brings greater challenges. Therefore, this paper first explores the application of the QMC method in density estimation.
### Main Methods
To solve this problem, the author proposes the following strategies:
1. **Preintegration**: Eliminate the discontinuities in the integral expressions through pre - integration. Specifically, choose an appropriate random parameter for integration so that the remaining integral becomes smooth and can be approximated using the QMC method.
2. **QMC Rules**: Construct an N - point lattice rule to approximate the integral. After pre - integration, the QMC error order is \(O(N^{- 1+\epsilon})\), where \(\epsilon>0\) is an arbitrarily small constant. This error order is the same as that when calculating the expected value.
### Mathematical Model
Consider the following form of the stochastic elliptic PDE:
\[-\nabla\cdot(a(x,\omega)\nabla u(x,\omega))=\ell(x,\omega),\quad x\in D,\]
\[u(x,\omega) = 0,\quad x\in\partial D,\]
where:
- \(D\subset\mathbb{R}^d\) is a bounded convex domain, \(d = 1,2,3\);
- \(a(x,\omega)\) is the log - normally distributed diffusion coefficient;
- \(\ell(x,\omega)\) is the normally distributed source term.
The quantity of interest \(X\) is a bounded linear functional of the PDE solution:
\[X = G(u(\cdot,\omega)).\]
The cumulative distribution function (CDF) and probability density function (PDF) are respectively:
\[F(t)=P[X\leq t]=E_{\omega}[\text{ind}(t - X)]=\int_{\Omega}\text{ind}(t - G(u(\cdot,\omega)))dP(\omega),\]
\[f(t)=E_{\omega}[\delta(t - X)]=\int_{\Omega}\delta(t - G(u(\cdot,\omega)))dP(\omega),\]
where \(\text{ind}\) is the indicator function and \(\delta\) is the Dirac delta function.
### Specific Steps of Pre - integration
To simplify the integral, first integrate the variable \(y_0 = w_0\) so that the integral of the CDF or PDF becomes an integral with respect to the remaining \(2s\) variables \(y_1,\ldots,y_{2s}\). This can eliminate the discontinuities caused by the indicator function and the Dirac delta function.
### Conclusion
Through pre - integration and the QMC method, this paper successfully solves the problem of estimating the probability density function and cumulative distribution function of the solutions of elliptic PDEs with random inputs, and proves the effectiveness and convergence of this method.