Physical Systems with Random Uncertainties: Chaos Representations with Arbitrary Probability Measure
read more
Citations
Global sensitivity analysis using polynomial chaos expansions
High-Order Collocation Methods for Differential Equations with Random Inputs
Adaptive sparse polynomial chaos expansion based on least angle regression
The stochastic finite element method: Past, present and future
References
Probability and Measure
Stochastic Finite Elements: A Spectral Approach
Stochastic Filtering Theory
Related Papers (5)
Frequently Asked Questions (14)
Q2. How can the representations presented in the paper be implemented into numerical codes?
The representations presented in the paper can be readily implemented into numerical codes, either using existing software via a Monte Carlo sampling scheme, or using stochastic codes that are adapted to the chaos decompositions.
Q3. What is the construction of the random solution of the elliptic boundary value problem?
The projection of the weak formulation of the elliptic boundary value problem with random uncertainties onto an m-dimensional subspace can be achieved through any one of a number of procedures, such as the finite element method (FEM), resulting in a random linear algebraic problem of the form(56) A(X1, . . . ,Xp) Y = F ,where F is a given element of Cm, andA is a randomm×m complex matrix depending on the basic vector-valued random variables X1, . . . ,Xp.
Q4. What is the construction of the chaos coefficients of Y r?
If this sequence is upper bounded by a positive finite constant, then all of the chaos coefficients of Y r converge to the chaos coefficients of the exact solution.
Q5. What is the construction of the chaos coefficients of Y?
This finite-dimensional deterministic algebraic system of equations yields the chaos coefficients of Y r. Unlike the construction via sampling introduced previously, the computed chaos coefficients now depend on the multi-index r = (r1, . . . , rp) used in the approximation.
Q6. What is the real Hilbert space associated with the probability measure?
It is noted here also that sk can be R or any bounded or compact subset thereof, and that, in general, Smj = s1 × · · · × smj .5.1.1. Hilbert spaces Hj,k and Kj.
Q7. What is the tensor product of Hilbert space?
Let { b1, . . . , bm } be the canonical basis of Rm, which is then also a basis for Cm. Then a Hilbertian basis of complex Hilbert space H(m) is given by (16) {( φ1α1 ⊗ · · · ⊗ φpαp ) ⊗ bj , α1 ∈ Nm1 , . . . ,αp ∈ Nmp , j = 1, . . . ,m } .4.3. Representation of f in H(m).
Q8. What is the recurrence relation of each polynomial?
Each of these polynomials is also associated with a onedimensional well-known probability density function, which is also indicated.
Q9. What is the construction of the coefficients Y1p?
An alternative construction [10] of the coefficients Yα1···αp consists in substituting the truncated chaos decomposition of Y , given in (59), into equation (56) and interpreting the resulting equality in the weak sense using the bilinear form in H(m) ×H defined in (21).
Q10. What is the probability of the random matrix A(X1,... ?
random matrix A(X1, . . . ,Xp)−1 exists almost surely, defining a nonlinear mapping f such that(57) Y = f(X1, . . . ,Xp).
Q11. What is the connection between classical orthogonal polynomials and chaos decompositions?
to each of the classical polynomials is associated a weight function that can be construed as a density of a measure on an appropriate space.
Q12. What is the tensor product of the real Hilbert space?
In addition, in order to simplify the notation, the tensor product H1 ⊗H2 has to be understood as the completion H1⊗̂H2 of the space H1 ⊗H2.Real Hilbert space Hj and complex Hilbert space H (m) are equipped with thefollowing inner products:〈u, v〉Hj = ∫R mju(xj)v(xj)
Q13. How can the p1 random process be discretized?
Each of the p1 random processes can itself be discretized in terms of a countable set of random variables via the Karhunen–Loeve expansion [17].
Q14. What is the Hilbert space of R-valued square-integrable functions on B?
Then let V = L2dζ(B,Rν) be the Hilbert space of Rν-valued square-integrable functions on B equipped with the inner product (44) 〈u,v〉V = ∫B〈u(ζ),v(ζ)〉