PeakLab v1 Documentation Contents AIST Software Home AIST Software Support

Genetic Algorithm

Refine Estimates for this Data Set with Genetic Algorithm...

This right click menu option in the three fitting options will use a genetic algorithm to attempt to refine the initial estimates. If you are having trouble placing peaks or if you have a local minima condition in a fit, the genetic algorithm may be of value. There are three strategies which may be available depending on peak count and the presence of an IRF:

·
**Reduced Data Genetic Prefit** - runs the genetic algorithm on this data set with a reduced
data density

·
**Reduced Data Genetic Prefit, Cycle Peaks** - similar to the general fitting option, this runs
multiple passes of the genetic algorithm, each with a focus that emphasizes a specific peak.

·
**Reduced Data Genetic Prefit, 2 Pass, Lock Shared Parameters on Pass 1** - similar to the general
fitting option, the reduced data GA prefit is done in two steps, the first pass locking all parameters
marked as shared.

The** Genetic Algorithm Preferences **option (and **Genetic** button in the Fit
Preferences dialog) is used to set the preference specifically applicable to this prefit. Given the
nature of peaks having such a local presence in an x-range of data, a valid a_{1} center estimate
and a_{1} GA constraint is essential. Note also that the default GA constraints are set to help
you escape a local minimum condition and refit. The constraints are quite narrow, but can be opened to
potentially find estimates for user-defined models where the higher order parameters may be largely unknown.

The GA (genetic algorithm) preferences dialog allows you to set the **Maximum Generations**, the **Maximum
Candidates per Generation**, the **Converge to Significant Digits**, and a **Genetic Strategy**.
The defaults are 100 generations, 40 candidates per generation, 6 significant digits convergence, and
the Advanced algorithm. The genetic algorithm should be considered experimental, and is furnished as an
alternative for difficult to place functions and for escaping settings which contain a local minimum state.
The algorithm mostly determines the pace of convergence. The constraints are far more important with respect
to realizing the global minimum.

The generic algorithm constraints are identical to the global LM algorithm constraints as there is no update. Unlike the LM algorithm, however, constraints are essential on all of the parameters. You will need to set these carefully. The defaults consist of small values specific to a refinement to take you out of a local minima position from a fit, not to find estimates when you have no idea of what the starting parameter values should be. The GA algorithm can be very effective in the latter, but with peak models you will need to constrain the estimates to some measure.