Dennis L. Young

Application of statistical design and response surface methods to computer-aided VLSI device design

A statistically oriented methodology for optimization and sensitivity analysis of VLSI process, device, and circuit design through computer simulation has been developed. Emphasis has been placed on maintaining a clear distinction between design synthesis and design analysis. Design analysis is viewed as a multiple input-output system resulting in a multiple-constraints-optimization problem. It is shown how simple graphic techniques or rigorous mathematical optimization can be performed within a constrained desirability space to determine optimal operating conditions. This leads directly to the concept of global input factors. Which affect a large number of the response variables, and specific input factors, which can be used to adjust the operating level of a small number of response variables. By using the derived empirical equations to desensitize the responses to variations in input factors, the proposed methodology can play a key role in designing for manufacturability. As proof of concept, the methodology has been applied to the optimization of a VLSI BIMOS technology, with satisfactory results. >

Application of statistical design and response surface methods to computer-aided VLSI device design II. Desirability functions and Taguchi methods

Methodology which incorporates the philosophy and methods of Taguchi is developed to address the issues of designing for manufacturability and sensitivity analysis. Specifically, the goal is to obtain responses (outputs) at or near required targets and to minimize output variability when the inputs are subject to manufacturing tolerances. The desirability functions, which generalize the notion of yield, are used to assess how close responses are to their targets. Manufacturing tolerances in the inputs are incorporated in the analysis by means of the expected loss of the desirability function which is estimated by a Taguchi outer array approach. The desirability expected loss is used to determine an optimum nominal input point to give responses close to target with small variability. The sensitivity of the various responses to the inputs is determined by an analytical method and an analysis of variance approach using a Taguchi outer array. The methods are applied to the optimization of a BIMOS NPN transistor. >

Knot selection for least-squares and penalized splines

Two new stochastic search methods are proposed for optimizing the knot locations and/or smoothing parameters for least-squares or penalized splines. One of the methods is a golden-section-augmented blind search, while the other is a continuous genetic algorithm. Monte Carlo experiments indicate that the algorithms are very successful at producing knot locations and/or smoothing parameters that are near optimal in a squared error sense. Both algorithms are amenable to parallelization and have been implemented in OpenMP and MPI. An adjusted GCV criterion is also considered for selecting both the number and location of knots. The method performed well relative to MARS in a small empirical comparison.

Comparative analysis of weight gain, hand/wrist maturation; and dental emergence rates in chimpanzees aged 0–24 months from varying captive environments

Variability during the first 2 years of growth and development is examined in captive chimpanzees. The mixed longitudinal study of 175 animals compares curves of weight, hand/wrist maturation, and dental emergence for groups within the sample which differ in sex, rearing circumstances (mother-reared versus hand-reared), and colony (Primate Foundation of Arizona, White Sands Research Center, and The University of Texas M.D. Anderson Cancer Center Department of Veterinary Sciences in Bastrop, Texas). Comparison of LOWESS fits of the curves, using a conservative jack-knife approach, reveals trends toward significant differences between colonies for weight (with 4 comparisons reaching significance) and between rearing groups for maturation (1 reaching significance). Results of a full versus reduced model approach show the same trends, for which significance is reached in a higher number of comparisons. The latter approach also indicates possible effects of sex and environmental differences on dental emergence rate. Difficulties with both approaches are discussed. It is concluded that the results are suggestive of significant sex and environmental effects on the variables monitored, justifying further analysis and continuation of the study. The study is significant in 1) providing norms specific to sex and rearing and colony environments with which individual colony animals may be compared in the assessment of their development and in 2) providing a standard, based upon data from a larger and more varied captive chimpanzee population than previously available, with which the dental emergence status and hand/wrist maturation of fossil apes and hominids may be compared.

Linear Nearest Neighbor Analysis

Linear nearest neighbor analysis is reconsidered and revised. This statistical method facilitates decisions about whether points along a line are clustered, random, or dispersed. One of its possible archaeological applications is illustrated using a set of probable housemounds along the Cabeza de la Vaca Arroyo in Veracruz, Mexico. The mounds are in a dispersed pattern. Ecological and subsistence factors offer one promising explanation for the pattern.

Using RngStreams for parallel random number generation in C++ and R

The RngStreams software package provides one viable solution to the problem of creating independent random number streams for simulations in parallel processing environments. Techniques are presented for effectively using RngStreams with C++ programs that are parallelized via OpenMP or MPI. Ways to access the backbone generator from RngStreams in R through the parallel and rstream packages are also described. The ideas in the paper are illustrated with both a simple running example and a Monte Carlo integration application.

Inference concerning the Mean Vector When the Covariance Matrix is Totally Reducible

Abstract Given a random sample from a p-variate normal distribution , the likelihood ratio criterion and an information theory criterion, which is an analogue of Hotellings T 2, are given for testing (known) under the condition that Σ is totally reducible, that is, diagonalizable by an orthogonal matrix which depends upon the pattern but not on the unknown elements of Σ. Exact, approximate and asymptotic distributions of the test criteria are considered and various confidence regions concerning are provided. Application to repeated measures experiments is discussed.

Effects of Early Rearing History on Growth and Behavioral Development in Captive Chimpanzees (Pan troglodytes)

Chimpanzees have one of the longest developmental periods of any mammal (Pereira and Fairbanks, 1993). Development includes a lengthy period of physical maturation (Riopelle, 1963; Doran, 1997), cognitive development (Plooij, 1987; Tomasello et al., 1993, 1994), and behavioral changes as the animal advances from infant (0–3.9 years) to juvenile (4–6.9 years), adolescent (7–9.9 years), and adult (10+ years). Weaning from their mothers or from their nursery environment is a long, gradual process (Horvat et al., 1980; Horvat and Kraemer, 1981; Fritz et al., 1991; Howell et al., 1993). Juveniles and adolescents develop social skills and relationships with others and gain new behaviors including foraging, grooming, and sexual behavior (Riopelle, 1963; Mason, 1967; Russon, 1990; Tomasello et al., 1994). While the purpose of the long developmental period is debated, it likely functions to promote physical growth, the acquisition of survival skills, and development into a socially and sexually competent adult (Walters, 1987; Pereira and Fairbanks, 1993).

Validating the probability of paternity

In cases of disputed parentage, it is commonly thought that the probability of paternity (P) represents the probability that the tested man is the biologic father of the child whose parentage is in doubt and that the paternity index (PI) is the likelihood ratio that enters into this probability. Using the results of HLA typing in a sample of disputed parentage cases, the distribution of P was examined within two groups: actual mother‐alleged father‐child trios, and simulated mother‐ unrelated man‐child trios. These distributions were found to be roughly consistent with the usual understanding of P.

Optimization of advanced bipolar ECL gates through statistical circuit simulation

Emitter-coupled logic (ECL) configurations that can be used in high-density gate arrays have been studied for a wide range of input conditions. CAD methodology which requires minimal user interaction is used. First, a general circuit schematic is created, allowing inputs to be varied. The basic schematic also incorporates estimates of layout-related parasitics. All the generic gates are tied together in the form of a ring oscillator. An internally written program then links the circuit configuration with other forced conditions such as voltage and temperature to the circuit simulator (MSPICE) with minimal designer interaction. Variations on all input variables are made through a Box-Benhken design so as to substantially reduce the number of simulations required in order to fit an approximating multi-dimensional quadratic output response surface to the inputs. The output of each circuit simulation (i.e. delay, power consumption, etc.) is automatically extracted and saved along with all its corresponding inputs. In the present experiment, a seven-factor Box-Benhken design was chosen using gate current, follower current, fan out, metal interconnect loading, voltage swing, and device size (Ae) as the input factors. The experiment required a total of 57 simulations, which were all run in batch mode. The output was analyzed using the statistical package SAS. For a given output (i.e. gate delay) an approximating quadratic polynomial, in terms of the seven inputs, is fit by the method of least squares. A ranking of the most important input variables as well as the correlation between two or more inputs can be obtained with this method.<<ETX>>