Unifying Optimization Algorithms to Aid Software System Users: optimx for R

By John C. Nash, R. Varadhan. 2011

418 citations recorded by the source corpus.

R users can often solve optimization tasks easily using the tools in the optim function in the stats package provided by default on R installations. However, there are many other optimization and nonlinear modelling tools in R or in easily installed add-on packages. These present users with a bewildering array of choices. optimx is a wrapper to consolidate many of these choices for the optimization of functions that are mostly smooth with parameters at most bounds-constrained. We attempt to provide some diagnostic information about the function, its scaling and parameter bounds, and the solution characteristics. optimx runs a battery of methods on a given problem, thus facilitating comparative studies of optimization algorithms for the problem at hand. optimx can also be a useful pedagogical tool for demonstrating the strengths and pitfalls of different classes of optimization approaches including Newton, gradient, and derivative-free methods.

Subfields: Artificial Intelligence, Computational Theory and Mathematics

View Paper: https://www.semanticscholar.org/paper/1e00a660e236aa21e992400553fa35b15f9b3ace

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computational Theory and Mathematics

Citations in the corpus

A Toolbox for Nonlinear Regression in R: The Package nlstools (2015, background): “Unfortunately, minimization, or optimization in general, is not always a straightforward exercise for such models due to the nonlinear relationships between parameters [Nash and Varadhan, 2011].”
A New Weibull Class of Distributions: Theory, Characterizations and Applications (2018, methodology): “MLEs are computed based on this data using R function optimx (see John and Ravi, 2011).”
Model selection in historical biogeography reveals that founder-event speciation is a crucial process in Island Clades. (2014, methodology): “By default, BioGeoBEARS uses the R package optimx (Nash and Varadhan 2011; 2012) to perform ML estimation of the free parameters, using the quasi-Newton method with box constraints (Byrd et al.”
Fitting Linear Mixed-Effects Models Using lme4 (2014, methodology): “An internal wrapper allows the use of the optimx package (Nash and Varadhan 2011), although the only optimizers provided via optimx that satisfy the constraints above are the nlminb and L-BFGS-B algorithms that are themselves wrappers around the versions provided in base R.”
Probabilistic historical biogeography: new models for founder-event speciation, imperfect detection, and fossils allow improved accuracy and model-testing (2013, neutral): “ ML optimization was conducted with optimx (Nash & Varadhan 2011; Nash & Varadhan 2012).”