General usage

Running atomicrex

The syntax used to run atomicrex takes the following simple form:

$ ./atomicrex <jobfile>

The <jobfile> argument represents an XML job file that is processed by atomicrex.

The job file

A job file controls the operation of the atomicrex program. It defines the potential to be fitted and the atomic structures and their properties which are used for the fitting process.

The format of a job file is a custom extended markup language (XML) dialect, which is described in the following. The general structure of a job file is:

<?xml version="1.0" encoding="iso-8859-1"?>
  <!-- general settings -->
  <validate-potentials />

  <!-- specify the different atom types (to be referred to in the blocks that follow) -->

  <!-- specify optimization procedure -->

  <!-- specify potential form and parameters -->

  <!-- specify structures and associated properties -->

Following the XML header, the root element <job> contains several tags and sections.

First, there are several “general” settings.

  • The optional <name> block specifies the name of the job.

  • The verbosity of the output is controlled using the <verbosity> tag, which supports the settings none, minimum, medium, maximum, and debug.

  • The precision of the floating point input/output is controlled via the <real-precision> tag, which expects an integer argument.

  • The <validate-potentials> tag serves a special purpose and is usually only used during debugging and error tracing. It enforces the numerical evaluation of the forces and compares the outcome with the results from the force calculation implemented in the potential source code. This enables one to check the correctness of the latter implementation. As the numerical calculation is, however, much slower, this tag should not be activated deliberately.

Then follows the <atom-types> block, in which the user has to specify the different atom types using the following format:

  <species mass='69.723' atomic-number='31'>Ga</species>
  <species mass='22.989' atomic-number='11'>N</species>

The mass and atomic-number attributes are optional. They are currently only needed when writing a fitted embedded atom method (EAM) potential to a tabulated EAM potential file.

The bulk of the input parameters is then contained in the three final sections, which define the potentials that are subject to the fitting, the structures that are included in the fit, and general parameters that control the fitting process. The optimization procedure is specified in the <fitting> block, which is explained in this section. The <potentials> and <structures> blocks are described in the sections on potentials and structures. Note that the input file format supports XML Inclusions (XInclude), which provides an elegant mechanism for structuring the input file and reusing blocks.