Code Base Development

The following discussion is based on my experience with the oscillation subgroup averages.

Our code base consists of the programs and scripts we use to calculate averages. The existing code works, but could be upgraded to:

improve useability, especially for non-expert users,
ease maintenance and upgrades,
support new averaging techniques.

Useability

The steps involved in reproducing the EPS 2003 Delta(m_d) average are outlined here and involve downloading about 30 input files in various formats, and running multiple passes of a master shell scripts that invokes 3 executables (COMBOS, BFRACTIONS, PAW). This level of complexity effectively prevents a casual user from understanding the inputs and methodology used in a particular average. Some amount of complexity is unavoidable given the underlying problem, but the following changes might improve the overall useability:

more structured set of input files with consistent input format (eg, XML-based with XSD schema definitions),
tools to browse input and output files (eg, XML-based with XSLT translation to HTML or SVG),
merging of executables.

Maintenance and Upgrades

The COMBOS fortran code started as a simple program and then gradually accumulated more and more features. The result is difficult to maintain and upgrade. The following changes might help:

use standard input format (eg, XML based) to eliminate complex input-parsing code,
re-examine the overall program design in light of the actual main use cases today,
switch to an object-oriented language (also makes it easier to find students willing to help).

New Averaging Techniques

There are already several recent results that we cannot fully exploit in COMBOS averages, for example, a simultaneous measurements of (tau(Bd), Delta(m_d)). There are also new oscillation quantities that have been measured that we cannot yet average, for example, Delta(Gamma_d) and |q/p|. Should we try to implement these within the existing COMBOS framework, or make other code changes first?

Page created by David Kirkby and last update 9 Sep 2003