Bootstrapped likelihood ratio test for mixture models in R

Question

I'm using R (MixAll package) to conduct a mixture model analysis that includes continuous and categorical indicators. I want to calculate a bootstrapped likelihood ratio test (BLRT) to compare model fit between nested classes (K, K-1). The steps seem fairly straightforward (see below), except that I am struggling with the parametric bootstrap part (step 2). Anybody have any experience in R generating a parametric bootstrap sample under the distribution of a specific mixture model (in this case K-1).

Steps:

1. Initially estimate the k-1 and k class models to provide the likelihoods for calculating the 2 log likelihood difference.

2. Under the null k-1 class model, generate a bootstrap sample and calculate the 2 log likelihood difference between the k-1 and k class models.

3. Repeat this process independently many times and estimate the true distribution of the 2 log likelihood difference.

4. Estimate the p value by comparing the distribution obtained in Step 3 with the 2 log likelihood difference obtained in Step 1. This p value is then used to determine if the null k 1 class model should be rejected in favor of the k class model.

Thanks.