Home :: Science Division :: Products :: GLSA
Least Squares Analysis
(GLSA) library is a PC software library that performs simultaneous non-linear
least squares parameters optimization of several model functions. Each model
function may have its own set of points to be fitted. This type of analysis is
called Global Analysis especially in conjunction to Fluorescence Kinetics Analysis.
The standard, one model, one data set problem is just a particular case of GLSA problem.
GLSA problem is generally characterized by a large number of parameters being optimized and a large number of local minima to be caught in. GLSA library implements an Alango proprietary algorithm for non-linear least squares optimization. The algorithm was designed for complex non-linear least squares problems with many parameters, ill-conditioning and multiple local minima. It greatly reduces probability of falling into a local minimum when initial values of optimized parameters are far from optimal. Obligatory for GLSA problems, such robustness significantly simplifies the choice of initial values for the standard one model, one data set problems.
GLSA library is organized as a set of simple API functions written in an object-oriented manner. It is easily integrated into other software with just a few lines of code. There are APIs for creating/destroying a GLSA computational object, defining model functions, linking different parameters together and fixing specific parameters.
The library provides capabilities for callbacks during computational process so that the optimization may be easily and effectively visualized and interrupted if necessary. The library is written in ANSI C programming language and may be easily ported to other general purpose or DSP based computational platforms. The library may be licensed to commercial software developers for integration into proprietary software as well as to end-users writing their own data analysis software.
GLSA library is used in our Dynals and FAST products guaranteeing their stability and ease of use.