Thermodynamic Properties

JTHERGAS: Thermodynamic estimation from 2D graphical representations of molecules, Edward S. Blurock, , Valérie Warth, Xavier Grandmougin, Roda Bounaceur, Pierre-Alexandre Glaude, Frédérique Battin-Leclerc, Energy, 43, 161–171 (2012).

This paper is exemplifies itself by exclusively using an external database of graphical representations of molecules and substructures to compute temperature dependent thermodynamic quantities of molecules and radicals for combustion. Traditional implementations rely on hard-coded base information and are thus not easily extensible. In addition, quantities, such as symmetry, which are normally determined by hand by the modeler is determined automatically.

Databases: The key to this system, as opposed to other approaches in this field, to have all data represented and accessible through an external database as opposed to being hard coded. This presented unique representational challenges. The database system and its management uses standard technologies and it decoupled from the actual system which uses the data to assemble and calculate the thermodynamic quantities.

Graph Theory: Once again, as in mechanism generation, graph representation is the key to the thermodynamic estimations. Using the “2D-graphical” representation (typical for how organic chemists, for example, reason with pencil and paper), to estimate three dimensional properties such as different types of symmetry, vibrational activity and even electron structure.

Algorithms and data structures: This paper represents a synergy of symbolic (primarily graphs and tables) and numeric data. To compute the different modes associated with the total calculation a large variety of different generalized data structures are used to organize the data and specialized algorithms to compute with these structures. For example, a general key structure in the estimations are a graphical representation of a molecular substructure associated with a table of corresponding information. The database information is in a MySQL database. Graphical browser interfaces are provided for the management of data structures. In addition, all major functionality is in a web-service.

Artificial Intelligence: The basic methods used in these calculations are to a large part mimicking human manipulation. This is especially true of the symmetry calculations needed for the estimations. The system as a whole, written in JAVA, can be considered an expert database system.