What is OPTIONS ?
OPTIONS is a design exploration and optimization
package that may be used to study and compare a large range of optimization
methods when applied to design problems.
Such processes are set out in our
book on design.
This system is now marketed world-wide by
DeZineForce.
The original software is described in the user manual which is
available at
http://www.soton.ac.uk/~ajk/options.ps.
Search Methods
The package contains, or has interfaces to, the following search methods:
- NAg Methods (E04UCF and E04UAF)
- Bit Climbing (BC)
- Dynamic Hill Climbing (DHC)
- Population Based Incremental Learning (PBIL)
- Design of Experiment methods (DOE)
- Numerical Recipes Methods (Num_Rcp)
- Method of Successive Linear Approximation (APPROX)
- Random Exploration with Shrinkage (RANDOM)
- Adaptive Random Search (ADRANS)
- Davidon, Fletcher, Powell Strategy (DAVID)
- Fletcher's 1972 Method (FLETCH)
- Jacobson and Oksman Method (JO)
- Powell's Direct Search (PDS - Siddall's implementation)
- Hooke and Jeeves Direct Search (SEEK - Siddall's implementation)
- Simplex Method (SIMPLX - Siddall's implementation)
- Rosenbrock's Rotating Co-ordinate Search (ROSE)
- Complex Strategy of M.J. Box (COMP)
- Two-membered Evolution Strategy (EVOL)
- Multi-membered Evolution Strategy (KORR)
- Repeated Fibonacci Search (FIBO)
- Repeated Golden Section Search (GOLD)
- Repeated Lagrangian Interpolation (LAGR)
- Hooke and Jeeves Direct Search (HOJE - Schwefel's implementation)
- Davies, Swann and Campey Search Method with Gram-Schmidt Orthogonalization (DSCP)
- Davies, Swann and Campey Search Method with Palmer Orthogonalization (DSCP)
- Powell's Direct Search (POWE - Schwefel's implementation)
- Davidon, Fletcher, Powell Strategy (DFPS)
- Simplex Method (SIMP - Schwefel's implementation)
- Genetic Algorithm (GA)
- Simulated Annealing (SA)
- Evolutionary Programming (EP)
- Evolution Strategy (ES)
The package can also be used to construct a range of response surface
models including:
- Sheperd weighting
- linear, quadratic and cubic polynomial regression
- linear, thin plate and cubic spline radial basis functions
- stochastic process (kriging) models with and without regression
It can also be used to construct Pareto fronts for multi-objective searches.
Application Areas and Examples
We have applied/are applying OPTIONS to:
Full references for papers describing this work and a number of related subjects
may be found
on the home page of Prof. A.J. Keane (http://www.soton.ac.uk/~ajk/welcome.html ).
This article may be found at
http://www.soton.ac.uk/~ajk/options.html
Correspondence to Andy.Keane@soton.ac.uk, or Prof. A.J. Keane, School of
Engineering Sciences, University of Southampton, Highfield, Southampton, SO17 1BJ, U.K.
Tel +44-2380-592944, FAX +44-2380-593230.