I-NoLLS: A program for interactive nonlinear least-squares fitting of the parameters of physical models

Mark M. Law*, Jeremy M. Hutson

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

57 Citations (Scopus)


The I-NoLLS program is a package for carrying out interactive nonlinear least-squares fits to determine the parameters of physical or mathematical models from experimental or other data, under circumstances where automated least-squares procedures are excessively computationally expensive and the user needs interactive control to apply physical insight to the fitting process. The program was developed to facilitate the fitting of molecular potential energy surfaces (PES) to spectroscopic and scattering data, but is also applicable to a variety of other optimization problems. A range of different algorithms adapted to highly nonlinear least-squares problems may be selected. The interactive nature of the code permits rapid and flexible control over the progress of the fit. I-NoLLS is written in a modular way that allows the easy incorporation of new modules for calculating observable quantities from model parameters. The structure of the program allows straightforward parallelisation of the time-consuming property calculations. In pilot applications, I-NoLLS has been interfaced with programs for calculating bound states of Van der Waals complexes, cross sections for molecular scattering processes, and second virial coefficients of gas mixtures. Parallelisation of the property calculations has been achieved using PVM running on a cluster of workstations.

Original languageEnglish
Pages (from-to)252-268
Number of pages17
JournalComputer Physics Communications
Issue number1-3
Publication statusPublished - 2 May 1997


  • Interactive
  • Model fitting
  • Molecular potential energy surface
  • Nonlinear least-squares
  • Optimization


Dive into the research topics of 'I-NoLLS: A program for interactive nonlinear least-squares fitting of the parameters of physical models'. Together they form a unique fingerprint.

Cite this