wepy.analysis package¶
Analysis tools for wepy datasets.
Analysis of weighted ensemble (WE) datasets are complicated by their nonlinear branching structure, and thus in its raw form cannot be analyzed by the tools available for linear trajectories. See the documentation (TODO: link) for a more in depth explanation of the general structure of weighted ensemble data.
These analysis modules do not aim to redo any particular analysis calculation but rather to aid in accessing, summarizing, and manipulating branched simulation data.
The starting point for this is the ContigTree class in the contig_tree module. This is a general purpose container that combines the data specifying the branching structure with the underlying linear data structures. It is necessary for branched data but can also be used for collections of linear data as well.
The main anlaysis routines we attempt to facilitate are:
Generating probability histograms over calculated observables. (TODO)
Generating contiguous trajectories of walkers that have reached target boundary conditions (TODO).
Visualizing the cloning and merging history of simulations (i.e. resampling trees) using the parents module and network_layouts.
Generating ‘state networks’ from frame labels and computing transition matrices with variable lag times using the network and transitions modules.
To calculate observables on the dataset you need to work directly with the data storage API in the wepy.hdf5.WepyHDF5 module using the compute_observable method.
See also
Notes
References
Examples
