Frederick H. Walters

Sequential Simplex Optimization - An Update

Sequential Simplex Optimization was popularized by Deming and coworkers at Emory University and the University of Houston in the 1980s. A text, Sequential Simplex Optimization was published by CRC Press in 1991.1 This article provides an updated bibliography from 1992 to present.

The Calibration Problem in Statistics and Its Application to Chemistry

Abstract Most linear regression problems involving prediction or estimation require determining the Y value corresponding to a given X. The calibration problem (or inverse estimation or discrimination) is the inverse of the above situation in that we have a given value of Y and we want to determine X. This paper will document the various statistical approaches to this problem and do a numerical example using the classical estimator.

The Use of a Box-Behnken Three Factor Design to Study the Paper Chromatographic Separation of Several Amino Acid Hydroxamates

Abstract A three factor Box Behnken design was used to study the effect of methanol, propanol and amyl alcohol on the paper chromatographic separation of the hydroxamates of arginine, leucine, threonine, histidine and Tryptophan. Box and Behnken 1 proposed some three level designs for fitting response surfaces. These designs are formed by conbining 2k factorials with incomplete block designs. The resulting designs are usually very efficient in terms of the number of required runs and they are rotatable (or nearly rotatable) The results here will be compared to a previous paper dealing with a three factor central composite design.

A Two-Factor Simplex Optimization of a Programmed Temperature Gas Chromatographic Separation

Abstract Simplex optimization was used to optimize the temperature programmed gas chromatographic separation of octane, 2-methyltetrahydrofuran, butylcyclohexane, and ethylbenzene. The initial temperature and the programming rate were the factors studied and a 5 minute time constraint was placed on the system. The Chromatographic Response Function (CRF) used by Deming and Morgan1 was used as an operational measure of performance. Further experiments in the region of the optimum and regression analysis of the data obtained were carried out to further understand the factor effects.

Window diagrams versus the sequential simplex method: which is correct?

Abstract The use of the sequential simplex technique may yield false local optima if the simplex is not large enough and collapses rapidly. Multiple simplexes started from different areas of factor space should be used. If they all arrive at the same optimum, then there is some assurance that one is looking at a global and not a local optimum. Window diagrams can be used if relatively few factors are to be investigated and time is available to do the necessary experimentation.

Indirect Fluorescent Visualization Of Aliphatic Amines

Abstract 2-Amino-5-methylbenzenesulfonic acid has been investigated as an indirect fluorescent visualization reagent in reversed phase ion pair chromatography. Aliphatic amines coelute as a negative peak which can be quantitated. Using a solvent of 1 mM 2-amino-5-methyl-benzenesulfonic acid, 0.05 M acetic acid, pH = 5.0 buffer and a C18 column, butylamine was quantitated. The calibration curve was linear from 1.0 M to 2.5 × 10−3M with a minimum detectable concentration of 1.01 × 10−3M with a 20 μl injection loop.

Multiple Linear Regression Analysis of the Retention Data for Several Neutral Metal Complexes

Abstract Multiple linear regression analysis for the retention data of neutral metal complexes of nickel, copper and palladium was carried out. Several columns (Microbondapack C18, Partisil-10-ODS, Alltech RP-8) and two ternary (water-methanol-acetonitrile-and water-methanol-tetrahydrofuran) and a quaternary (water-methanol-acetonitrile-tetrahydrofuran) solvent were studied. It was found that the palladium complex behaved differently than the nickel and copper complexes and needed a 21% methanol, 21% acetonitrile, 58% water solvent system in order to elute halfway between the nickel and copper complex.

The Use of a 3-Factor Statistical Mixiure Design and a 2-Factor Chromatograhic Response Function to the RP-HPLC Separation of Several Free α-Amino Acids

Abstract A 3-factor statistical mixture design and a 2-factor chromatographic response function were used in the solvent optimization of the RP-HPLC separation of a mixture containing free histidine, methionine, tyrosine, phenylalanine and tryptophan. The influences of phosphoric acid, methanol and acetonitrile on the separation were compared. A solvent condition resulting in a satisfactory separation of the amino acid mixture was predicted.

A Two Factor Simplex Optimization of An Ion Pair Liquid Chromatographic Separation

Abstract A simple two factor simplex is used to optimize the separation of several ionic naphthylamines by varying the pH and percent methanol in the eluting solvent.

Studies On The Reaction Of O-Phthaldialdehyde With Triphenylmethanethiol And Amino Acids

Abstract The reaction of o-phthaldialdehyde with alkyl thiols and amino acids is well known. This method has been used to determine the concentration of taurine4 and pico mole levels of amino acids5by liquid chromatography of the substituted isoindole formed. However, alkyl thiols are volatile and have noticeable odors. It was not known if this reaction works with arylthiols. Several aromatic thiols were investigated and triphenylmethanethiol which is a solid was found to react to give a fluorescent product which is most likely a substituted indole. The indole formed was studied with respect to stability as a function of time ans simplex optimization was used to study the factors involved in the derivatization reaction.