diff --git a/package/doc/sphinx/source/documentation_pages/analysis/parallelization.rst b/package/doc/sphinx/source/documentation_pages/analysis/parallelization.rst index 91ae05fceca..3070614b5a3 100644 --- a/package/doc/sphinx/source/documentation_pages/analysis/parallelization.rst +++ b/package/doc/sphinx/source/documentation_pages/analysis/parallelization.rst @@ -41,7 +41,8 @@ impossible with any but the ``serial`` backend. Parallelization is getting added to existing analysis classes. Initially, only :class:`MDAnalysis.analysis.rms.RMSD` supports parallel analysis, but - we aim to increase support in future releases. + we aim to increase support in future releases. Please check issues labeled + `parallelization` on the `MDAnalysis issues tracker `_. How does parallelization work @@ -106,6 +107,10 @@ If you want to write your own *parallel* analysis class, you have to implement denote if your analysis can run in parallel by following the steps under :ref:`adding-parallelization`. +.. Note:: + + Attributes that are the same for the whole trajectory, should be defined + in `__init__` method because they need to be consistent across all workers. For MDAnalysis developers ~~~~~~~~~~~~~~~~~~~~~~~~~ diff --git a/package/doc/sphinx/source/images/AnalysisBase_parallel.png b/package/doc/sphinx/source/images/AnalysisBase_parallel.png index 960e7cc257b..9a9b1ea4d78 100644 Binary files a/package/doc/sphinx/source/images/AnalysisBase_parallel.png and b/package/doc/sphinx/source/images/AnalysisBase_parallel.png differ