Isaac Siwale

Isaac Siwale
Zenith Genetica Ltd

About

13
Publications
10,477
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
30
Citations
Introduction
I am the founder and Chief Executive Officer of Zenith Genetica Ltd. The company’s research is ‘application-driven’ and is focussed on ideating, developing and implementing optimization and machine learning methods using evolutionary computation technology. The current product-line comprises an optimization and equation solver called GENO, and a machine learning algorithm called EAGLE—the second generation of the GENO solver is available at www.tomopt.com. The launch of EAGLE is imminent.

Publications

Publications (13)
Data
Documents published by Zenith Genetica Ltd are “live” entities that evolve over time and in this regard, the company endeavours to make corrections and issue new editions of said documents in a timely manner. Each document includes the following metadata in the header: ‘Document Name’, ‘Document ID’, ‘Revision Number’ and ‘Issue Date and Time’. Sem...
Technical Report
Full-text available
This paper offers a new proof of Fisher’s invariance theorem. The proof is based on the definition of a likelihood function and it employs known results from probability and optimization, coupled with some elementary geometry of quadratic functions.
Technical Report
Full-text available
This paper comprises a collection of some basic concepts of statistical estimation theory obtained from various sources and presented in summary fashion; it makes no original contribution to the subject—its purpose is merely to act as a supplement to the Decentralized Parameter Estimation paper [15] that presents a proprietary numerical estimator c...
Technical Report
Full-text available
This paper presents a reformulation that is a natural by-product of the 'variable endogenization' process for equality-constrained programs. The method results in a partial relaxation of the constraints which in turn confers some computational advantages. A fully-annotated example illustrates the technique and presents some comparative numerical re...
Technical Report
Full-text available
This paper presents a method for finding global optima to constrained nonlinear programs via slack variables. The method only applies if all functions involved are of class C 1 but without any further qualification on the types of constraints allowed; it proceeds by reformulating the given program into a bi-objective program that is then solved for...
Technical Report
Full-text available
This paper presents a relatively “unfettered” practical method for finding global optima to constrained nonlinear programs. The method reformulates the given program into a bi-objective mixed-integer program that is then solved for the Nash equilibrium solution. Two numerical examples are provided to showcase the efficacy of the method—the solution...
Technical Report
Full-text available
This paper presents a statistical modelling procedure that applies to several scenarios. The regression model is based on the Box-Cox transform, and the associated numerical estimator is a decentralized computational scheme in which the estimation task is shared between an evolutionary algorithm and an ordinary least-squares (OLS) program—the evolu...
Technical Report
Full-text available
This paper presents four optimization models for solving nonlinear equation systems. The models accommodate both over-specified and under-specified systems. A variable endogenization technique that improves efficiency is introduced, and a basic comparative study shows that the optimization methods presented are very effective.
Technical Report
Full-text available
This paper is on practical solutions to the multi-objective optimization problem; it advocates for single-point solutions either of the Nash equilibrium or the Tchebycheff compromise type, depending on whether one can reasonably ascribe competition or cooperation to the problem at hand. A transform method that greatly simplifies implementation of t...
Technical Report
Full-text available
This paper is on the portfolio optimization problem for which two generic models are presented in the context of a proprietary solver called GENO: the first is a pseudo-dynamic model with a single state variable that is meant for the single holding-period case; the second is a truly dynamic model with several state variables that applies to both si...
Technical Report
Full-text available
This document present the results of some numerical experiments whose aim was to assess the efficacy of GENO on equation systems that may generically be stated as C(x) = 0, where C is vector comprising numeric functions ci(x) which may or may not be nonlinear. The examples are practical problems emanating from a wide variety of applied science and...
Technical Report
Full-text available
This document presents the results of some numerical experiments whose aim was to assess the efficacy of GENO on static and dynamic programs. The numerical examples cover a wide range of problem-types, and a variety of solutions. Most of the examples have been attempted previously by others using various techniques. The emphasis in this study is on...

Projects

Projects (3)
Project
The company is undergoing a rebranding process. The name has been changed to 'Zenith Genetica Ltd' and subsequently, all documents will be re-issued to reflect this change.
Project
A hard-copy printed version of the GENO user manual for GAUSS programmers is available from Amazon.com as well as other online retailers; the aim of this new project is provide the equivalent book product for TOMLAB programmers.
Project
The aim of the project is to publish a book version of the user manual (for GAUSS programmers) of a solver called GENO—a software product that has been successfully tested on real-life and artificial problems, and against well-known methods embedded in popular scientific computing packages. GENO—an acronym for ‘General Evolutionary Numerical Optimizer’—is a versatile solver that may be used on an exceptionally wide range of numerical problems: one may use it to solve simultaneous equation systems of all types, or to solve various classes of static or dynamic optimization problems. The method is relatively unfettered because it does not require the problem at hand to have a special mathematical structure; it may also be set to generate real or integer-valued solutions or a mixture of the two as required. The book begins with a generic statement of the mathematical model that underlies GENO’s design together with brief statements on how to specialize the model (by mere choice of a few program parameters) so as to address specific problem types; it include a set of fully-worked examples that span the ‘applicability spectrum’ of GENO; and it ends with a set of appendices comprising: (i) an essay on the main numerical method employed by GENO—namely evolutionary search; (ii) a classification and index of the example programs that ship with the product; (iii) an ‘applicability map’ that vividly illustrates the relative advantage of GENO when compared to twenty other numerical solution methods that are currently available.