%\VignetteIndexEntry{Introduction to BiocParallel}
%\VignetteEngine{knitr::knitr}

\documentclass{article}

<<style, eval=TRUE, echo=FALSE, results="asis">>=
BiocStyle::latex()
@ 

\newcommand{\BiocParallel}{\Biocpkg{BiocParallel}}

\title{Introduction to \BiocParallel} 
\author{Vincent Carey, Michael Lawrence, Martin
  Morgan\footnote{\url{mtmorgan@fhcrc.org}}} 
\date{Edited: February 16, 2014; Compiled: \today}

\begin{document}

\maketitle

\section{Introduction}

The \BiocParallel{} package provides a consistent way of specifying
parallel evaluation within \Bioconductor. Enable its use by attaching
the package
%% 
<<BiocParallel>>=
library(BiocParallel)
@ 
%% 
To use, invoke a \BiocParallel{} function like \Rcode{bplapply}, or
use a \BiocParallel-enabled function provided by another package.

\subsection{Why use \BiocParallel?}

The Task View document at
\url{http://cran.r-project.org/web/views/HighPerformanceComputing.html}
shows that numerous approaches to parallel computing are available in
\R.  Most applications cited in the task view identify one or more of
\CRANpkg{snow}, \CRANpkg{Rmpi} or \CRANpkg{foreach} as relevant
parallelization infrastructure.

A basic objective of \BiocParallel{} is reduction of complexity faced
by developers and users in creating and using software that benefits
from performing computations in parallel.  This is accomplished by
defining abstractions of the key components of parallel computing
environments.  Information on the parallel environment can be stored
in formally structured ``parameter'' arguments that are used at run
time to define the approach to parallel execution.  This allows
developers to focus on \textit{what} is to be computed, leaving the
\textit{how} to the infrastructure.

Advantages for developers and power users of \BiocParallel{} over
\textit{ad hoc} \R{} programming for parallel computation include the
following.
\begin{itemize}
\item A uniform idiom (using \Rcode{BiocParallelParam} instances) is
  available for defining parallel computing resources; sensible
  default definitions are generated when \BiocParallel{} is loaded.
\item \Rcode{bplapply} and \Rcode{bpvec} address iteration in parallel
  and parallel evaluation of vectorized functions respectively.
\item When the parallel environment is managed by a cluster scheduler
  through \CRANpkg{BatchJobs}, job management and result retrieval are
  considerably simplified.
\item \Rcode{foreach} and programming with the \CRANpkg{iterators}
  package are fully supported, but registration of the parallel back
  end uses \Rcode{BiocParallelParam} instances.
\end{itemize}

\section{The \BiocParallel{} Interface}

The \BiocParallel{} work flow is simple:
\begin{enumerate}
\item Invoke \BiocParallel-enabled functions. The functions use the
  registered back-ends for evaluation.
\end{enumerate}

An optional step is to register appropriate back-ends for your
particular configuration by
\begin{enumerate}
\item Creating a \Rcode{BiocParallelParam} instance to describe how
  parallel evaluation is to be implemented.
\item Registering the \Rcode{BiocParallelParam} instance for use in
  your \R{} session.
\end{enumerate}
%% 
The registry is a `stack', with the last entry added to the stack used
first, so your own back-ends generally take precedence over the
back-ends established when the \BiocParallel{} package is loaded.

\subsection{\Rclass{*Param} objects to describe parallel evaluation environments}

Different types of parallel computation are supported by creating and
\Rcode{register()}ing a `\Rcode{Param}'. Supported \Rcode{Param}
objects are:

\begin{description}
\item[\Rcode{SerialParam}] Evaluate \BiocParallel-enabled code with
  parallel evaluation disabled. This is very useful when writing new
  scripts and trying to debug code.
\item[\Rcode{MulticoreParam}] Evaluate \BiocParallel-enabled code
  using multiple cores on a single computer. When available, this is
  the most efficient and least troublesome way to parallelize
  code. Unfortunately, Windows does not support multi-core evaluation
  (the \Rcode{MulticoreParam} object can be used, but evaluation is
  serial). On other operating systems, the default number of workers
  equals the value of the global option \Rcode{mc.cores} (e.g.,
  \Rcode{getOption("mc.cores")}) or, if that is not set, the number
  of cores returned by \Rcode{parallel::detectCores()}.
\item[\Rcode{SnowParam}] Evaluate \BiocParallel-enabled code across
  several distinct \R{} instances, on one or several computers. This
  can be an easy way to parallelize code when working with one or
  several computers, and is based on facilities originally implemented
  in the \CRANpkg{snow} package. Different types of \CRANpkg{snow}
  `back-ends' are supported, including socket and MPI clusters.
\item[\Rcode{BatchJobsParam}] Evaluate \BiocParallel-enabled code by
  submitting to a cluster scheduler like SGE. 
\item[\Rcode{DoparParam}] Register a parallel back-end supported by
  the \CRANpkg{foreach} package for use with \BiocParallel.
\end{description}

The simplest illustration of creating \Rcode{BiocParallelParam} is
<<BiocParallelParam-SerialParam>>=
serialParam <- SerialParam()
serialParam
@ 
%% 
Most parameters have additional arguments influencing behavior, e.g.,
specifying the number of `cores' to use when creating a
\Rcode{MulticoreParam} instance
<<BiocParallelParam-MulticoreParam>>=
multicoreParam <- MulticoreParam(workers=8)
multicoreParam
@ 
%% 
Arguments are detailed on the corresponding help page, e.g.,
\Rcode{?MulticoreParam}.

\subsection{\Rcode{register()}ing \Rcode{BiocParallelParam} instances}

The \Rcode{register()} function registers a \Rcode{BiocParallelParam}
instance for use in parallel evaluation.
<<register>>=
register(multicoreParam)
@ 
%% 
View registered parameters with \Rcode{registered()}
%% 
<<registered>>=
registered()
@ 
%% 
The list of registered \Rcode{BiocParallelParam} instances represents
the user's preferences for different types of back-ends. Individual
algorithms may specify a preferred back-end, and different back-ends
maybe chosen when parallel evaluation is nested.

\subsection{Functions for parallel computation}

There are facilities for querying and controlling parallel evaluation
environments.
\begin{description}
\item[\Rcode{bpisup(x)}] Query a \Rcode{BiocParallelParam} back-end
  \Rcode{x} for its status.
\item[\Rcode{bpworkers}] Query a \Rcode{BiocParallelParam} back-end
  for the number of workers available for parallel evaluation.
\item[\Rcode{bpstart(x)}] Start a parallel back end specified by
  \Rcode{BiocParallelParam} \Rcode{x}, if possible.
\item[\Rcode{bpstop(x)}] Stop a parallel back end specified by
  \Rcode{BiocParallelParam} \Rcode{x}.
\end{description}
%% 
These are used in common functions, implemented as much as possible
for all back-ends. The functions (see the help pages, e.g.,
\Rcode{?bplapply} for a full definition) include
\begin{description}
\item[\Rcode{bplapply(X, FUN, ...)}] Apply in parallel a function
  \Rcode{FUN} to each element of \Rcode{X}. \Rcode{bplapply} invokes
  \Rcode{FUN} \Rcode{length(X)} times, each time with a single element
  of \Rcode{X}.
\item[\Rcode{bpmapply(FUN, ...)}] Apply in parallel a function
  \Rcode{FUN} to the first, second, etc., elements of each argument in
  \ldots.
\item[\Rcode{bpvec(X, FUN, ...)}] Apply in parallel a function
  \Rcode{FUN} to subsets of \Rcode{X}. \Rcode{bpvec} invokes function
  \Rcode{FUN} as many times as there are cores or cluster nodes, with
  \Rcode{FUN} receiving a subset (typically more than 1 element, in
  contrast to \Rcode{bplapply}) of \Rcode{X}.
\item[\Rcode{bpaggregate(x, data, FUN, ...)}] Use the formula in
  \Rcode{x} to aggregate \Rcode{data} using \Rcode{FUN}.
\end{description}
%% 
There are facilities for recovering from errors
\begin{description}
\item[\Rcode{bplasterror}] Report the last error reported from a
  \BiocParallel{} evaluation.
\item[\Rcode{bpresume}] Attempt to resume computation after an error.
\end{description}

\section{Use cases}

\subsection{Single computer}

\subsection{\emph{Ad hoc} clusters}

\subsection{Clusters with schedulers}

\section{For developers}

Developers wishing to use \BiocParallel{} in their own packages should
include \BiocParallel{} in the \texttt{DESCRIPTION} file
\begin{verbatim}
    Imports: BiocParallel
\end{verbatim}
and import the functions they wish to use in the \texttt{NAMESPACE}
file, e.g.,
\begin{verbatim}
    importFrom(BiocParallel, bplapply)
\end{verbatim}
Then invoke the desired function in the code, e.g.,
<<devel-bplapply>>=
system.time(x <- bplapply(1:3, function(i) { Sys.sleep(i); i }))
unlist(x)
@ 
%%
This will use the back-end returned by \Rcode{bpparam()}, by default a
\Rcode{MulticoreParam()} instance or the user's preferred back-end if
they have used \Rcode{register()}.  The \Rcode{MulticoreParam} back-end
does not require any special configuration or set-up and is therefore
the safest option for developers. Unfortunately,
\Rcode{MulticoreParam} provides only serial evaluation on Windows.

Developers should document that their function uses \BiocParallel{}
functions on the man page, and should perhaps include in their
function signature an argument \Rcode{BPPARAM=bpparam()}.

Developers wishing to invoke back-ends other than
\Rcode{MulticoreParam} need to take special care to ensure that
required packages, data, and functions are available and loaded on the
remote nodes.

\end{document}