Tutorial 5 Problem Set
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Tutorial 5 Problem Set
1. Consider we have a set of weights w = (w1 , w2 , . . . , wn) generated by the importance sampling. Suppose wi ≥ 0 for each i = 1, . . . , n and let ∑i wi < ∞ . Show the effective sample size (ESS) in the form of
ESS1(w) := ( ∑ wi)2
satisfies 1 ≤ ESS1(w) ≤ n.
Then, given the sample mean and sample variance of the weights, n = ∑ wi and n(2) = ∑(wi −
n)2 , respectively. Derive ESS1(w) using n and n(2) .
2. Suppose we want to compute Ep[h(X)] < ∞ using importance sampling with a importance density g(x). Suppose the second moments Ep[h(X)2] is bounded. The so-called defensive importance sampling uses a mixture density in the form of
gα(x) = αgdef(x) + (1 − α)g(x), α ∈ (0, 1),
where g(x) is a “good” importance density but not necessarily satisfies Requirement 6.4.1 and gdef(x) is the defensive term which satisfies
p(x)
for some constant c > 0. Show that the new density gα(x) satisfies both Requirement 6.4.1 and Requirement
6.4.2.
3. Consider a pair of random variable X and Y . Suppose the random variable X follows a Laplace distribution
with density
fX (x) = exp(−|x|).
Suppose further the conditional random variable Y | X follows the logistic distribution with density
exp(x2 − y)
fY |X (y | x) =
• Define the joint probability density function of (X, Y ).
• Derive an inverse transform method that can draw random variables (X, Y ) using a pair of uniform random variables (U1, U2 ).
2023-05-31