The Computational Platform for DT Models

This page presents a collection of different documents and code examples related to the computational platform for DT models.

Basic Ideas

Cumulative development of knowledge in Decision Theater work calls for a computational platform that allows to represent in a standardized way the different agents and entities involved in very different models. A key requirement for such a platform is that it should allow simulations to be run on machines with vastly different computing power. When used with high performance computers, the platform must allocate the available computational resources to different model components depending on the computational cost each element generates — a cost that typically varies in the course of a simulation run.

The basic ideas underlying the platform are explained in the document “A Computational Platform for DT Models”. The code for version 0 (written in R) is here.

Version 0: Tutorial

This text constitutes a tutorial for working with the GCF Computational Platform for DT Models, version 0, based on a simple example model. This version of the GCF Platform is written in the R programming language as its aim is not computational performance, but an illustration of the underlying ideas.

Source code for platform version 0 and tutorial: here
Tutorial text: here

Version 0: Technical Documentation

This document describes the current implementation of the platform. The current version of the platform is only a prototype to demonstrate the concepts, including the parallelization of the simulation. It is written in R, without consideration of the CPU performance. The source code of this section is included in the Tutorial source code above.

Technical documentation: here