The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

Parallelize 'glmmTMB' functions

The CRAN 'glmmTMB' package + The 'futurize' hexlogo = The 'future' logo

The futurize package allows you to easily turn sequential code into parallel code by piping the sequential code to the futurize() function. Easy!

TL;DR

library(futurize)
plan(multisession)
library(glmmTMB)

m <- glmmTMB(count ~ mined + (1 | site), data = Salamanders, family = nbinom2)
pr <- profile(m) |> futurize()

Introduction

This vignette demonstrates how to parallelize glmmTMB functions such as profile() through futurize().

The glmmTMB package fits generalized linear mixed models (GLMMs) using Template Model Builder (TMB). Its profile() function computes likelihood profiles for model parameters. These computations are performed independently for each parameter, making them candidates for parallelization.

Example: Likelihood profile

The profile() function computes the likelihood profile for each model parameter. For example, using the built-in Salamanders dataset to model salamander counts:

library(glmmTMB)

## Fit a negative binomial GLMM
m <- glmmTMB(count ~ mined + (1 | site), data = Salamanders, family = nbinom2)

## Compute likelihood profile
pr <- profile(m)

Here profile() is calculated sequentially. To calculate in parallel, we can pipe to futurize():

library(futurize)
library(glmmTMB)

m <- glmmTMB(count ~ mined + (1 | site), data = Salamanders, family = nbinom2)
pr <- profile(m) |> futurize()

This will distribute the per-parameter profile computations across the available parallel workers, given that we have set up parallel workers, e.g.

plan(multisession)

The built-in multisession backend parallelizes on your local computer and works on all operating systems. There are other parallel backends to choose from, including alternatives to parallelize locally as well as distributed across remote machines, e.g.

plan(future.mirai::mirai_multisession)

and

plan(future.batchtools::batchtools_slurm)

Supported Functions

The following glmmTMB functions are supported by futurize():

Without futurize: Manual PSOCK cluster setup

For comparison, here is what it takes to parallelize profile() using the parallel package directly, without futurize:

library(glmmTMB)
library(parallel)

## Fit a negative binomial GLMM
m <- glmmTMB(count ~ mined + (1 | site), data = Salamanders, family = nbinom2)

## Set up a PSOCK cluster
ncpus <- 4L
cl <- makeCluster(ncpus)

## Compute likelihood profile in parallel
pr <- profile(m, parallel = "snow", ncpus = ncpus, cl = cl)

## Tear down the cluster
stopCluster(cl)

This requires you to manually create and manage the cluster lifecycle. If you forget to call stopCluster(), or if your code errors out before reaching it, you leak background R processes. You also have to decide upfront how many CPUs to use and what cluster type to use. Switching to another parallel backend, e.g. a Slurm cluster, would require a completely different setup. With futurize, all of this is handled for you - just pipe to futurize() and control the backend with plan().

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.