# Question: Suppose the random variable has an exponential distribution fX x

Suppose the random variable has an exponential distribution, fX (x) = exp (– x) u (x). We wish to estimate the probability of the event A = {X >xo} via simulation. We will compare the standard Monte Carlo estimate,

Where the random variables Xi are chosen according to the exponential distribution specifed by PDF fX (x), with an importance sampling estimate,

Where the random variables Yi are chosen from a suitable distribution specified by its PDF, fY (y). Note that both estimators are unbiased, so we will compare these estimators by examining their variances.

(a) Find the variance of the Monte Carlo estimate.

(b) Find the variance of the IS estimator assuming that the random variables Yi are chosen from a scaled exponential distribution, fY (y) = aexp (– ay) u (y).

(c) Assuming that xo = 20, find the value of that minimizes the variance of the IS estimator using the scale exponential distribution.

(d) How much faster do you expect the IS simulation to run as compared to the MC simulation?

Where the random variables Xi are chosen according to the exponential distribution specifed by PDF fX (x), with an importance sampling estimate,

Where the random variables Yi are chosen from a suitable distribution specified by its PDF, fY (y). Note that both estimators are unbiased, so we will compare these estimators by examining their variances.

(a) Find the variance of the Monte Carlo estimate.

(b) Find the variance of the IS estimator assuming that the random variables Yi are chosen from a scaled exponential distribution, fY (y) = aexp (– ay) u (y).

(c) Assuming that xo = 20, find the value of that minimizes the variance of the IS estimator using the scale exponential distribution.

(d) How much faster do you expect the IS simulation to run as compared to the MC simulation?

## Answer to relevant Questions

An experiment consists of tossing a coin twice and observing the sequence of coin tosses. The sample space consists of four outcomes ξ1 = (H, H), ξ2 (H, T), ξ3 (T, H), and ξ4 (T, T). Suppose the coin is not evenly ...(a) An experiment consists of rolling a die and flipping a coin. If the coin flip is heads, the value of the die is multiplied by - 1, otherwise it is left as is. What are the possible outcomes of this experiment? (b) Now, ...Prove that for any two events A and B, Pr (A∩B) ≤ Pr (A) ≤ Pr (AUB). Repeat Exercise 12.14 using random variables that follow a Gaussian distribution fX (x) = exp (– x2 / 2). Also, for parts (b)–(d) use a shifted distribution for the importance sampling estimator of the form fY (y) = exp ...Suppose a pseudorandom sequence is constructed using the power residue method. If q = 11, find a value of that leads to a sequence with maximum possible period. X k = ax k – 1 mod q, k = 1, 2, 3… Find the period and the ...Post your question