Carlos B. Lucasius

Multicriteria target vector optimization of analytical procedures using a genetic algorithm: Part I. Theory, numerical simulations and application to atomic emission spectroscopy

Abstract The target vector criterion was combined with a genetic algorithm as a parallel and evolutionary search technique. In this way a multicriteria optimization technique was obtained. In a number of numerical simulations the method was compared with classical optimization techniques used in analytical chemistry such as grid search, modified simplex, steepest ascent, overlapping resolution map, pattern search and simulated annealing according to their computational expenditure and speed, their ability to find global and local optima, their on-line applicability in analytical instrumentation and number of experiments needed. Then the technique was experimentally demonstrated for the simulataneous optimization of intensities of six atomic emission lines of trace elements in alumina powder as a function of spectroscopic excitation conditions.

Optimization of calibration data with the dynamic genetic algorithm

Abstract Genetic algorithms constitute a set of powerful search heuristics. A modified genetic algorithm was used to optimize calibration data sets. In order to construct an ideal genetic procedure, the diversity in a population is crucial. The idea proposed is to estimate the diversities along two directions, namely the diversity between the chromosomes in a population and the diversity between the alleles in all chromosomes. The newly defined diversity functions are able to describe the procedure of a genetic algorithm in detail and can be used as a feedback for dynamic control of the process in an almost ideal way. The optimization results show that for both short and long runs the dynamic genetic algorithm is superior to the “classical” genetic algorithms and that after optimization not only can the data sets be compacted and refined but also the predictive ability of the calibration model can be improved.

Sequential assignment of 2D-NMR spectra of proteins using genetic algorithms

The application of genetic algorithms to the problem of the sequential assignment of two-dimensional protein NMR spectra is discussed. The problem is heavily underconstrained since in most cases more patterns are available than amino acid positions, and uncertainties may exist in the preliminary assignments. The results indicate that relatively large amounts of errors may be present in the input data for the genetic algorithm while useful results may still be obtained.

Gates towards evolutionary large-scale optimization: A software-oriented approach to genetic algorithms—II. Toolbox description

Abstract Starting from Part I, Part II of this paper globally describes the toolbox GATES, elucidating how concepts of genetic algorithm methodology can be implemented for application. The prototype for numerical parameter estimation is treated in most detail. Additionally, an auxiliary utility for configuration, based on explorative walks in the search space, is outlined.

HIPS, a hybrid self-adapting expert system for nuclear magnetic resonance spectrum interpretation using genetic algorithms

An automatic system for the interpretation of two-dimensional NMR spectra of proteins, HIPS, is presented. Several artificial intelligence techniques are combined to form a flexible, hybrid system that has (limited) learning capabilities. Following the structure of the problem, the system is divided in modules with distinct functionalities. The first two modules are rule-based, and can be validated and refined semi-automatically using a set of already interpreted spectra. In this way, an optimized ruleset can be obtained to interpret unknown spectra. Results indicate a significant effect of training on performance. In the third module, a genetic algorithm is used to tackle a search problem of huge dimensions in which patterns found in the NMR spectra should be mapped to amino acids in the sequence.

Gates towards evolutionary large-scale optimization: A software-oriented approach to genetic algorithms—I. General perspective

Abstract Genetic algorithms comprise a novel methodology that has proven to be powerful in approaching complex, large-scale optimization problems in a wide variety of sciences, recently including computational chemistry. However, as it turns out, up to now the exploitation of this power is not at all a straightforward matter for many potential practitioners, among which are computational chemists. Both parts of this paper provide keys to this group of scientists that should enable them to open gates towards genetic algorithm applications on computers. After a general introduction, Part I presents a taxonomy for genetic algorithm software. The Discussion highlights important properties of different kinds of genetic algorithm software, and proposes a strategy to the applied scientist who needs an executable application without first having to become an expert in genetic algorithm science. The material presented is largely based on our past experience, which includes the insights that we gained during the development and use of our software library GATES. Applications built with GATES are spread across various fields of computational chemistry. GATES is described in Part II, the accompanying paper.

Conformational Analysis of DNA Using Genetic Algorithms

Among techniques for conformational analysis of DNA molecules, restrained molecular dynamics and distance geometry are, up to now, most widely used. Both techniques are essentially based on local search strategies which use evaluation criteria that are simplified for pragmatic reasons. In practice, this approach appears to be decreasingly adequate when increasingly complex conformational spaces are dealt with.

Multicriteria target vector optimization of analytical procedures using a genetic algorithm: PartII. Polyoptimization of the phothometric calibration graph of dry glucose sensors for quantitative clinical analysis

Abstract The target vector criterion as the vector distance between a set of desired responses and a set of really obtained responses of a multivariate function was combined with the genetic algorithm to improve simultaneously six properties of a biochemical test strip for human blood glucose determination as a function of twelve chemical and technological parameters. The advantage of the genetic algorithm in comparison with other search techniques became obvious for the search in this twelve-dimensional variables space with up to seven bit resolution, i.e., up to 1.93 25 search positions. The results obtained by target vector optimization on the basis of the genetic search technique were critically compared with the results obtained by a prediction with classical and non-linear partial least-squares regression and realized in laboratory and industrial verification experiments. In this way advantages and disadvantages of deductive and inductive polyoptimization strategies could be discussed theoretically and with respect to experimental results.