MCMC diagnostics

Explore the difference between MC and MCMC with a sample example.
A discrete variable δ∼{1,2,3}\delta\sim\{1,2,3\}δ∼{1,2,3}
 and a continuous variable θ∈I ⁣R\theta\in\mathrm{I\!R}θ∈IR
.
1
# discrete variable
2
delta = c(.45, .10, .45)
3
​
4
# continuous variable
5
mu = c(-3, 0, 3)
6
sigma2 = c(1/3, 1/3, 1/3)
7
​
8
# exact marginal density of theta
9
ext_margin_den <- function(x)
10
{
11
  dnorm(x, mu[1], sqrt(sigma2[1])) * delta[1] +
12
    dnorm(x, mu[2], sqrt(sigma2[2])) * delta[2] +
13
    dnorm(x, mu[3], sqrt(sigma2[3])) * delta[3]
14
}
15
​
16
theta = seq(-6, 6, length.out = 1000)
17
ptheta = ext_margin_den(theta)
18
plot(theta, ptheta, type = "l")
Copied!
The marginal density of θ\thetaθ
 would be
Firstly, generate 1000 Monte Carlo θ\thetaθ
-samples.
1
sample_delta <- function(x)
2
{
3
  denx = c(delta[1] * dnorm(x, mu[1], sqrt(sigma2[1])),
4
           delta[2] * dnorm(x, mu[2], sqrt(sigma2[2])),
5
           delta[3] * dnorm(x, mu[3], sqrt(sigma2[3])))
6
  condenx = denx / sum(denx)
7
  sample(1:3, 1, prob = condenx)
8
}
9
# gibbs sampler
10
gibbs <- function(m, delta = 3)
11
{
12
  DELTA = delta
13
  THETA = NULL
14
  for (i in 1:m)
15
  {
16
    # sample theta from full conditional distribution
17
    theta = sample_theta(delta)
18
    THETA = c(THETA, theta)
19
    delta = sample_delta(theta)
20
    DELTA = c(DELTA, delta)
21
  }
22
  return(THETA)
23
}
24
res = gibbs(1000)
25
plot(res)
26
hist(res, breaks = 20)
Copied!
For 1000 MCMC samples, we have
Actually, re-run the above code, we can get much different figure. Let's try 10000 MCMC samples,
It turns out to be much stable when you re-run the above code.
sample autocorrelation
Use R-function acf. If a Markov chain with high autocorrelation, then it will move around the parameter space slowly, taking a long time to achieve the correct balance among the different regions of the parameter space.
effective sample size
Use R command effectiveSize in the coda package, which can be interpreted as the number of independent Monte Carlo samples necessary to give the same precision as the MCMC samples.

MCMC

Metropolis-Hastings

Last modified 3yr ago

Copy link

Contents

sample autocorrelation

effective sample size