Introduced nested parallelization for better usage of HPC resources
Introduced chunking of tasks for better performance in parallel environments
Updated vignette on HPC environment
Update documentation of performAnalyses()
Recommended doFuture and future over doParallel and parallel when no parallel backend is detected
Updated Suggests and SystemRequirements in DESCRIPTION
Added a WORDLIST
Minor changes in code
Fixed a bug in continueRecruitment() that could result in additional subjects to be recruited when the overall decision for a trial realization is NoGo but some cohorts of that trial realization have Go decisions.
Fixed a bug in performAnalyses() that would occur if all trial realizations of a scenario had a previous overall NoGo decision and would result in performAnalyses() to return an empty list for that scenario’s posterior quantiles.
Specified R2jags package version requirement in DESCRIPTION to prevent ‘unused argument’ bug in performAnalyses()
Fixed warning message not showing when specifying deprecated arguments ‘seed’ and ‘n_cores’ in performAnalyses()
Usage of doRNG package for fully reproducible results in parallel execution
Usage of hash tables for mapping unique trial realizations to scenario trial realizations resulting in performance improvement
Added a vignette that provides a short example on how to use bhmbasket in a high performance computing environment
JAGS model files stored in package instead of writing to temporary files
Updated Imports in DESCRIPTION
Updated CITATION
Minor changes in code
Updated documentation
Removed superfluous files
Fixed a bug in continueRecruitment() that could result in an error message and prevent the function from running although all conditions were met.
Fixed a bug leading to an error messages in getEstimates(). It would occur if in a previous call of performAnalyses() differences between cohorts were calculated, but not in the current call of performAnalyses().
Registration of parallel backend is now the responsibility of the user to allow for flexibility. A respective message is displayed in performAnalyses() if no parallel backend is registered.
Functions simulateScenarios() and performAnalyses() look up for their arguments ‘n_trials’ and ‘n_mcmc_iterations’, respectively, in the global environment if not provided by user.
Arguments ‘n_subjects_list’ and ‘n_subjects_add_list’ of the functions simulateScenarios() and continueRecruitment(), respectively, can be provided as a single vector for the case when all scenarios recruit the same number of subjects.
Arguments ‘seed’ and ‘n_cores’ were deprecated in performAnalyses(). The argument ‘seed’ was not used in the call of R2jags::jags(), see also the the documentation of ‘jags.seed’ in ?R2jags::jags. The argument ‘n_cores’ is no longer needed as the registration of the parallel backend is now the responsibility of the user.
Added the argument ‘overall_min_nogos’ to the function negateGoDecisions().
Added an input check in function getEstimates() regarding the argument ‘add_parameters’.
Changed date format in message from performAnalyses()
Updated documentation
Updated Imports and Suggests in DESCRIPTION
Minor changes in code
Fixed a rare bug that could result in wrong overall decision in getGoDecisions() in the presence of previous go decisions
Updated documentation
Updated Description in DESCRIPTION
Minor changes in code