Feasible solution algorithms for high breakdown estimation
Version of September 1997, incorporating additional necessary
condition.  Minor further modifications are from 2003.

This directory contains an IBM PC ZIP archive  fsa_v2.zip

When unzipped, the ZIP archive generates a README
file, a collection of FORTRAN 77 source and data files as listed
in the README file, and 7 executables
FSALMSV2.EXE
FSALTSV2.EXE
FSAMCDV2.EXE
FSAMVEV2.EXE
FSALTAV2.EXE
HBEDA.EXE
EXACTMVE.EXE

These IBM PC executables are compiled versions of the corresponding
source files, and may be executed directly to perform high
breakdown estimation on user data files.

The LMS code is effectively identical to that in the original
FSA codes.  The LTS and MCD codes use the additional weak necessary
condition to screen out poor initial random subsets, and run
much faster than the original FSA's, particularly on large data sets.

The MVE code uses only the weaker condition.  It will produce
approximations (albeit much inferior) in data sets so large that
the original FSA took an intolerable length of time to produce any
feasible solutions at all.  As it does not use the strong case-swap
necessary condition at all, it should be used only on data sets
where the original FSA code could not run in a tolerable amount of time.

Current thinking seems to avoid use of LMS and MVE in favor of
LTS and MCD respectively.

The LTA code performs multiple regression using the least trimmed
sum of absolute deviations criterion.  It provides an alternative
to LTS and LMS.

HBEDA is the multivariate analysis of variance / discriminant
analysis procedure of Hawkins and McLachlan.

EXACTMVE uses the exact algorithm for MVE described in
Cook, Hawkins and Weisberg.

Hawkins, D. M., and Olive, D. J., (1999) `Applications and
Algorithms for Least Trimmed Sum of Absolute Deviations Regression',
Computational Statistics and Data Analysis, 32, 119-134.

Hawkins, D. M. and Olive, D. J. (1999), `Improved Feasible
Solution Algorithms for High Breakdown Estimation',
Computational Statistics and Data Analysis, 30, 1-11.

Hawkins, D. M., and McLachlan, G. J., (1997), `High Breakdown
Linear Discriminant Analysis', Journal of the American
Statistical Association, 92, 136-143.

Cook, R. D., Hawkins, D. M., and Weisberg, S., (1993),
`Exact Computation of the Robust Multivariate Minimum Volume
Ellipsoid Estimator', Statistics and Probability Letters,
16, 213-218.

Hawkins, D. M., (1993), `The Feasible Set Algorithm for Least
Median of Squares Regression', Computational Statistics and
Data Analysis, 16, 81-101.