Go-Smart Simulation Algorithms
==============================

These are members of the `CDM <index>`__ that provide something
akin to lambda calculus to Go-Smart. They are database defined, have
textual body in an unspecified language, return a parameter as a
*Result* and take named arguments that must be supplied by the numerical
model as a condition of forming a valid *Combination*.

Definition
----------

They can be defined in `GSSA-XML <../gssa-xml>`__, within the
``<algorithms />`` node as follows:

.. code-block:: xml

        <algorithm result="RESULT">
            <arguments>
                <argument name="ARGNAME" />
            </arguments>
            <content>
                ALGORITHMDEFINITION
            </content>
        </algorithm>

The ``ALGORITHMDEFINITION`` may be defined in any language the
simulation family will understand (or, more accurately, the ultimate
third-party simulation tool). Examples include MATC (for
`GSSF <https://go-smart.github.io/goosefoot/overview/>`__) and Python (for
`FEniCS <../docker/fenics>`__). The ``RESULT`` is conventionally an
upper-case, underscore-spaced slug that can be used in place of a
conventional "constant" parameter. Arguments are specified solely by
name, not type (as we do not even know the applicable language). They
are expected to be supplied by the simulation procedure at run-time.

While this is no less secure than the numerical model running on the
Docker instance from an administrator perspective, it should be noted
that the provenance of the algorithm may well be different, and if the
author of the numerical model (or Docker instance) is concerned about
data leakage from their instance, they should take adequate precaution.
Within the Python Docker module, the algorithm will not be called except
as requested by the numerical module.