Speaker
C. Jones
(CORNELL UNIVERSITY)
Description
Generic programming as exemplified by the C++ standard library makes
use of functions or function objects (objects that accept function
syntax) to specialize generic algorithms for particular uses. Such
separation improves code reuse without sacrificing efficiency. We
employed this same technique in our combinatoric engine: DChain. In
DChain, physicists combine lists of child particles to form a
list of parent hypotheses. E.g., d0 = pi.plus() * K.minus(). The
selection criteria for the hypothesis is defined in a function or
function object that is passed to the list's constructor.
However, C++ requires that functions and class declarations be defined
outside the scope of a function.
Therefore physicists are forced to separate the code that defines the
combinatorics from the code that sets the selection criteria. We will
discuss a technique using C++ expression templates to allow users to
define function objects using a mathematical expression directly in
their main function, e.g.,
func = (sqrt( beamEnergy*beamEnergy - vPMag*vPMag) >= 5.1*k_GeV).
Use of such techniques can greatly decrease the coding 'excess' needed
to perform an analysis.
Primary author
C. Jones
(CORNELL UNIVERSITY)
