1: \begin{abstract}

2: This work is concerned with

3: approximating multivariate functions in unbounded domain by using

4: discrete least-squares projection with random points evaluations.

5: Particular attention are given to  functions with random Gaussian

6: or Gamma parameters. We first demonstrate that the traditional Hermite (Laguerre)

7: polynomials chaos expansion suffers from the \textit{instability}

8: in the sense that an \textit{unfeasible} number of points,

9: which is relevant to the dimension of the approximation space,

10: is needed to guarantee the stability in the least square framework.

11: We then  propose to use the Hermite/Laguerre {\em functions}

12:  (rather than polynomials) as bases in the expansion. The corresponding design points are obtained by

13: mapping the uniformly distributed random points in bounded intervals to the unbounded domain, which involved a mapping parameter $L$. By using the Hermite/Laguerre {\em functions}

14: and a proper mapping parameter,

15: the stability can be significantly improved even if

16: the number of design points scales \textit{linearly} (up to a

17: logarithmic factor) with the dimension of the approximation space.

18: Apart from the stability, another important issue is the rate of

19: convergence. To speed up the convergence, an effective scaling factor is introduced, and a principle for

20: choosing quasi-optimal scaling factor is discussed.

21: Applications to

22: parametric uncertainty quantification are illustrated by considering a

23: random ODE model together with an elliptic problem with lognormal random input.

24: \end{abstract}