Thesis (Ph. D.)--University of Texas at Austin, 2003. / Vita. Includes bibliographical references. Available also from UMI Company.

While research into improving data quality from analytical instrumentation has gone on for decades, only recently has research been done to improve information extraction methods. One of these methods, correlation analysis, is based upon the shifting of one function relative to another and determining a correlation value for each displacement. The cross correlation algorithm allows one to compare two files and find the similarities that exist, the convolution operation combines two functions two dimensionally (e.g. any input into an analytical instrument convolves with that instrument response to give the output) and deconvolution separates functions that have convolved together. In correlation chromatography, multiple injections are made into a chromatograph at a rate which overlaps the instrument response to each injection. Injection intervals must be set to be as random as possible within limits set by peak widths and number. When the input pattern representation is deconvolved from the resulting output, the effect of that input is removed to give the instrument response to one injection. Since the operation averages all the information in the output, random noise is diminished and signal-to-noise ratios are enhanced. The most obvious application of correlation chromatography is in trace analysis. Signal-to-noise enhancements may be maximized by treating the output data (for example, with a baseline subtraction) before the deconvolution operation. System nonstationarities such as injector nonreproducibility and detector drift cause baseline or "correlation" noise, which limit attainable signal-to-noise enhancements to about half of what is theoretically possible. Correlation noise has been used to provide information about changes in system conditions. For example, a given concentration change that occurs over the course of a multiple injection sequence causes a reproducible correlation noise pattern; doubling the concentration change will double the amplitude of each point in the noise pattern. This correlation noise is much more amenable to computer analysis and, since it is still the result of signal averaging, the effect of random fluctuations and noise is reduced. A method for simulating conventional coupled column separations by means of time domain convolution of chromatograms from single column separations is presented.

Not available / text

A Fourier Transform Spectrometer (FTS) is an interferometric instrument configured so that the optical spectrum of the input to the instrument can be determined from the Fourier transform of the output. Spectral resolution is proportional to the maximum path differences in the interferometer. The Kitt Peak 1-Meter FTS is a high resolution instrument featuring a one-meter maximum path difference. The instrument is of the continuous motion type and utilizes equal time sampling of the interferogram. Realization of the potential of a one-meter path difference instrument requires very precise path difference positioning. Static and dynamic position accuracies of approximately seven Angstrom units are achieved in this instrument. Path difference command and control is accomplished through use of phase-lock techniques. A key element in the implementation is a Zeeman frequency stabilized Zeeman laser. Stabilization is accomplished by phase-locking the Zeeman frequency to a quartz crystal controlled system reference frequency. Path position information is derived by a control position interferometer which utilizes the Zeeman laser so that one cycle in phase at the Zeeman frequency corresponds to one wavelength in path difference at the laser optical frequency. Position control is achieved by phase-locking the output of the position control interferometer to a command signal. The command signal is synthesized from the laser Zeeman frequency by an additional phase-lock loop. Two approaches to minimizing spectral ghosts resulting from velocity errors are developed. One technique is applicable to band-pass designs. The second optimizes response of low-pass designs that include a single high-pass pole to reject the average value of the interferogram. Both techniques result in ghost responses improved by factors of at least ten to one-hundred compared with those of conventional data filters. The instrument includes an autoranging analog-to-digital converter with fifteen bits of resolution and total range of twenty-two bits developed to cope with the large dynamic range of the interferogram. These efforts together with careful design and construction have resulted in a powerful instrument which is in active use.

No description available.

No description available.

No description available.

No description available.

Electronic solid state circuitry controlled by an SDK-85 microcomputer forms the basis of a digitally controlled Fourier synthesizer module. The hardware involved in this module is the major thrust of this particular research work. Ultimately, once properly machine programmed, the microcomputer will control the base frequency, the number of harmonics present in the output waveshape, and the relative amplitudes and phases of these harmonics.The first phase of the hardware interface entails the conversion of square waves (from 8253 timers via the microcomputer) to triangle waveforms (via LM741 integrators). The triangle waves are then sine shaped by special clipping circuitry. It is the resultant sine waves that are then digitally controlled by machine software. The first stage of the above process eliminates the problem of frequency dependence and gain dropoff of the LM741 operational amplifiers. The relative amplitude of the triangle wave output of the LM741 is compensated for such that the sine shaping circuitry receives a constant amplitude waveform. The second phase of the hardware interface uses digital control of the final sine waveform by means of a Voltage Controlled Amplifier (VCA) and a Digital to Analog Converter (DAC). The VCA acts as the actual interface between the computer and the module.Finally, the output of the VCA'a are mixed and sent through an audio amplifier or to an oscilloscope for user inspection.

We say that a restriction theorem holds for a curve (gamma) (t) in (//R)('n) if for all f(epsilon) ((//R)('n)) and for some p and q, there is a constant C(,p,q) such that / (VBAR)(VBAR) f (VBAR) (,(gamma)) (VBAR)(VBAR) (,L('q)(du)) (LESSTHEQ) C(,p,q) (VBAR)(VBAR) f (VBAR)(VBAR) (,L('P)((//R)('n))). / In Chapter 1, we prove restriction theorems for non-compact plane curves with non-negative affine curvature when 1 (LESSTHEQ) p < 4/3 and / (DIAGRAM, TABLE OR GRAPHIC OMITTED...PLEASE SEE DAI) / There is an analogous result for space curves in the same chapter. / The Hilbert transform along the curve (gamma) is defined by / (DIAGRAM, TABLE OR GRAPHIC OMITTED...PLEASE SEE DAI) / In Chapter 2, it is shown that when (gamma) has the rapidly decreasing positive affine curvature, H(,(gamma)) is a L('P)-bounded operator for / (DIAGRAM, TABLE OR GRAPHIC OMITTED...PLEASE SEE DAI)