Experimental evaluation of enhanced metropolis sampling

In this thesis a general sampling algorithm using Monte Carlo method is proposed. This sampling algorithm is applicable to all kinds of population distributions unlike other sampling algorithms. To date, Monte Carlo sampling algorithm is the only general, efficient, and accurate sampling method applicable to all probability distributions.;In this thesis the performance of Enhanced Metropolis sampling algorithm is evaluated by performing experiments to compute the sample means and variances. It is found that our algorithm gives accurate estimates of sample means and variances. The performance of our algorithm is also compared with the performance of other algorithms such as AR sampling and the original Monte Carlo sampling algorithm. It is observed that our algorithm gives accurate statistical estimates even for data in higher dimensions whereas the other algorithms under perform when it comes to data in higher dimensions. Though the AR Sampling has more accurate statistical estimates its high cost and poor efficiency make it unsuitable for data with higher dimensions. This shows that the performance of Enhanced Monte Carlo Sampling algorithm is dimension independent. It can therefore be applicable to any probability distribution and is a general method.