Running the MCMC

Last updated: 2020-02-20

Checks: 4 3

Knit directory: lieb/

This reproducible R Markdown analysis was created with workflowr (version 1.6.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

R Markdown file: uncommitted changes

The R Markdown file has unstaged changes. To know which version of the R Markdown file created these results, you’ll want to first commit it to the Git repo. If you’re still working on the analysis, you can ignore this warning. When you’re finished, you can run wflow_publish to commit the R Markdown file and build the HTML.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(20190717)

The command set.seed(20190717) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: unavailable

Recording the operating system, R version, and package versions is critical for reproducibility. To record the session information, add sessioninfo: “sessionInfo()” to _workflowr.yml. Alternatively, you could use devtools::session_info() or sessioninfo::session_info(). Lastly, you can manually add a code chunk to this file to run any one of these commands and then disable to automatic insertion by changing the workflowr setting to sessioninfo: “”.

Cache: detected

The following chunks had caches available:

unnamed-chunk-3

To ensure reproducibility of the results, delete the cache directory running_mcmc_cache and re-run the analysis. To have workflowr automatically delete the cache directory prior to building the file, set delete_cache = TRUE when running wflow_build() or wflow_publish().

File paths: relative

Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.

Repository version: 3a9bf9d

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:


Ignored files:
    Ignored:    .DS_Store
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    analysis/.DS_Store
    Ignored:    analysis/.Rhistory
    Ignored:    analysis/pairwise_fitting_cache/
    Ignored:    analysis/running_mcmc_cache/
    Ignored:    output/.DS_Store
    Ignored:    output/chip_shiny/.DS_Store
    Ignored:    output/chip_shiny/Data/.DS_Store
    Ignored:    output/vision_shiny/.DS_Store
    Ignored:    output/vision_shiny/.Rhistory
    Ignored:    output/vision_shiny/Data/.DS_Store

Unstaged changes:
    Modified:   analysis/running_mcmc.Rmd

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

These are the previous versions of the R Markdown and HTML files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view them.

File	Version	Author	Date	Message
html	3a9bf9d	hillarykoch	2020-02-20	Build site.
html	72e6fec	hillarykoch	2020-02-20	Build site.
Rmd	1c27b8e	hillarykoch	2020-02-20	add description of mcmc output
html	1c27b8e	hillarykoch	2020-02-20	add description of mcmc output
Rmd	862bc02	hillarykoch	2020-02-07	big updates for mcmc processors
html	862bc02	hillarykoch	2020-02-07	big updates for mcmc processors
html	8b3e556	hillarykoch	2019-12-03	Build site.
html	39383ac	hillarykoch	2019-12-03	Build site.
Rmd	645d408	hillarykoch	2019-12-03	workflowr::wflow_publish(files = “*“)
html	e467a51	hillarykoch	2019-08-16	different shiny location
html	d58e1a6	hillarykoch	2019-08-16	resource data
html	38fb1c0	hillarykoch	2019-08-16	edit shiny
html	d6d897c	hillarykoch	2019-07-18	reorder mcmc section
Rmd	330ef60	hillarykoch	2019-07-18	reorder mcmc section
html	330ef60	hillarykoch	2019-07-18	reorder mcmc section
Rmd	a8d87b4	hillarykoch	2019-07-18	updat mcmc section
html	a8d87b4	hillarykoch	2019-07-18	updat mcmc section
Rmd	4991356	hillarykoch	2019-07-18	updat mcmc section
html	4991356	hillarykoch	2019-07-18	updat mcmc section
html	228123d	hillarykoch	2019-07-17	update up to obtaining the hyperparameters
html	c1dc0c1	hillarykoch	2019-07-17	update up to obtaining the hyperparameters
html	674120c	hillarykoch	2019-07-17	update about page
html	e67a3a2	hillarykoch	2019-07-17	update about page
html	2faf446	hillarykoch	2019-07-17	update about page
html	2177dfa	hillarykoch	2019-07-17	update about page
html	8b30694	hillarykoch	2019-07-17	update about page
html	e36b267	hillarykoch	2019-07-17	update about page
html	a991668	hillarykoch	2019-07-17	update about page
html	a36d893	hillarykoch	2019-07-17	update about page
html	e8e54b7	hillarykoch	2019-07-17	update about page
html	f47c013	hillarykoch	2019-07-17	update about page
Rmd	50cf23e	hillarykoch	2019-07-17	make skeleton
html	50cf23e	hillarykoch	2019-07-17	make skeleton

—This is likely the most time-consuming step of CLIMB analysis—

The final step of CLIMB involves doing inference on the parsimonious Gaussian mixture using MCMC. MCMC is an iterative method, and thus the user needs to specify how many iterations to use. We recommend running a quick pilot analysis–say, for 10 iterations. This pilot analysis will give a good idea of how long an analysis will need to run for a given larger number of iterations (say, 20,000 iterations).

You can run an mcmc simply with the function run_mcmc(). This function calls a script written in Julia, and executes everything at the default settings in the CLIMB methodology. The user needs to provide 4 arguments:

dat: the input data you’ve been using throughout the analysis
hyp: the hyperparameter values estimated in the previous step
nstep: number of MCMC iterations to run
retained_classes: the parsimonious list of candidate latent classes, after finally filtering out by prior weights as done in the previous step

First, we load in our data, list of candidate latent classes, and estimated hyperparameters.

data("sim")
load("output/hyperparameters.Rdata")
retained_classes <- readr::read_tsv("output/retained_classes.txt", col_names = FALSE)

Now we are ready to launch an MCMC:

results <- run_mcmc(sim$data, hyp =  hyp, nstep = 100, retained_classes = retained_classes)
out <- extract_chains(results)

Running the MCMC...100%|████████████████████████████████| Time: 0:00:33

The object results contains 3 objects:

chain: the estimate parameters over the course of nstep iterations
acceptane_rate_chain: an \(M\times\)nstep matrix of the acceptance rates for each cluster covariance. The proposals for each cluster are adaptively tuned such that the acceptance rates converge to about 0.3
tune_df_chain: the tuning degrees of freedom across the chain, adjusted to yield optimal acceptance rates

results is effectively a Julia object, so the first thing you should do with this object is to extract the data for R’s use:

out will contain all the different chains from the MCMC. For example, you can check the MCMC trace plots. Here is the trace plot of the mean from the first cluster in the third dimension:

plot(out$mu_chains[[1]][,3], type = "l", xlab = "iteration", ylab = expression(mu[3]))

Version	Author	Date
862bc02	hillarykoch	2020-02-07

More specifically, extract_chains() returns a list with 4 elements. First, recall that M is the number of input classes, D is the dimension of the data, and let iterations be nstep+1.

mu_chains: a list with M elements, each element a matrix of dimension iterations x D. mu_chains[[i]] is the MCMC samples for the mean vector of cluster i.
Sigma_chains: a list with M elements, each element an array of dimension D x D x iterations. Sigma_chains[[i]] is the MCMC samples for the covariance matrix of cluster i.
prop_chain: A matrix of dimension M x iterations, containing the MCMC samples for the mixing proportions of each class.
z_chain: A matrix of dimension n x iterations, containing the MCMC samples for the class labels of each observation. These labels correspond to the row indices of retained_classes (above).