openfe/docs/guide/execution/execution_theory.rst

.. _userguide_execution_theory:

Protocols and the Execution Model Theory
========================================

Protocols in OpenFE are built on a flexible execution model. 
Result objects are shaped by this model, and therefore some basic
background on it can be useful when looking into the details of simulation
results. In general, most users don't need to work with the details of the
execution model, but the general ideas can be useful.

.. TODO figure showing an example dag

Each protocol involves a number of steps (called ``ProtocolUnit``\ s) which occur in
some order. Formally, this is described by a directed acyclic graph (DAG),
so the collection of steps to run is called a ``ProtocolDAG``. A
:class:`.Protocol` creates the ``ProtocolDAG``, and a single ``ProtocolDAG``
should give information necessary to obtain an estimate of the desired
thermodynamic observable. Over the course of a campaign, a single
:class:`.Protocol` may create multiple ``ProtocolDAG``\ s, e.g., to extend a
simulation. NB: While independent runs can be created as separate
``ProtocolDAG``\ s, the recommend way to do independent runs is as a
``repeats`` part of the settings for the protocol, which puts the
independent runs in a single ``ProtocolDAG``.

.. TODO review recommendation for repeats in context of NEQ protocol

There are results objects at each level of this: so the
:class:`.ProtocolResult` is associated with the :class:`.Protocol`. Just as
the :class:`.Protocol` may create one or more ``ProtocolDAG``\ s, the
:class:`.ProtocolResult` will be made from one or more
:class:`.ProtocolDAGResult`\ s.  Finally, each :class:`.ProtocolDAGResult`
may carry information about multiple :class:`.ProtocolUnitResult`\ s, just a
single ``ProtocolDAG`` may involve mutliple ``ProtocolUnit``\ s.

.. TODO FUTURE: figure showing the relations of protocol objects and result
   objects

.. TODO FUTURE: add information about scratch/shared/permanent storage
   once that becomes relevant