Automated Composition of Module Chains

We introduce a problem of module composition. Modules are seen as “black boxes” with input and output ports. A compatibility relation models which input ports can connected to which output ports. We are given a set of available modules and a target module. We want to connect available modules into a chain that implements the target module. Constraints may be given on how many copies of each module can or should appear in the solution chain. Costs may be given on modules or ports or connections, so that an optimal solution with respect to these costs is found. We derive an algorithm to solve the above problem automatically. The algorithm transforms the problem into a shortest-path problem in a graph.     

Input Feature Extraction for Multilayered Perceptrons Using Supervised Principal Component Analysis

A method is proposed for constructing salient features from a set of features that are given as input to a feedforward neural network used for supervised learning. Combinations of the original features are formed that maximize the sensitivity of the network's outputs with respect to variations of its inputs. The method exhibits some similarity to Principal Component Analysis, but also takes into account supervised character of the learning task. It is applied to classification problems leading to improved generalization ability originating from the alleviation of the curse of dimensionality problem.     

Sticky-free and overhang-free DNA languages

An essential step of any DNA computation is encoding the input data on single or double DNA strands. Due to the biochemical properties of DNA, complementary single strands can bind to one another forming double-stranded DNA. Consequently, data-encoding DNA strands can sometimes interact in undesirable ways when used in computations. It is crucial thus to analyze properties that guard against such phenomena and study sets of sequences that ensure that no unwanted bindings occur during any computation. This paper formalizes and investigates properties of DNA languages that guarantee their robusteness during computations. After defining and investigating several types of DNA languages possessing good encoding properties, such as sticky-free and overhang-free languages, we give algorithms for deciding whether regular DNA languages are invariant under bio-operations. We also give a method for constructing DNA languages that, in addition to being invariant and sticky-free, possess error-detecting properties. Finally, we present the results of running tests that check whether several known gene languages (the set of genes of a given organism) as well as the input DNA languages used in Adlemans DNA computing experiment, have the defined properties.Received: 6 February 2003, Published online: 2 September 2003Research partially supported by Grants R2824A01 and R220259 of the Natural Sciences and Engineering Research Council of Canada.     

Macroexothermic phenomena in exothermic additions: mathematical and physical modelling

In this paper a mathematical model is presented to predict the macroexothermic phenomena occurring when exothermic additions in lump form are assimilated in ferrous metals. The macroexothermic phenomena take place during the free assimilation period of exothermic additions in ferrous metals. These phenomena are characterized by unique coupled heat, mass and momentum transport phenomena. The presence of a moving boundary complicates further these phenomena. The model uses the Simpler algorithm to solve numerically the pertinent partial differential equations. The extensive verification of the model was carried out in two contexts. The first was, in a low temperature physical model consisting of ice immersion in different sulfuric acid solutions. The melting of ice in these solutions is extremely exothermic. In this physical model, both temperature and velocity measurements were carried out. The model results were compared with experimental measurements and they were found to be in excellent agreement. The second context employed high temperatures, involving the assimilation of silicon in high carbon liquid iron. The model was also applied to predict the position of the moving boundary for these high temperature experiments and a good agreement was obtained. In addition new dimensionless convective heat transfer correlations that quantify these complex phenomena are presented.     

Finding boundary conditions: A coupling strategy for the modeling of metal casting processes: Part I. Experimental study and correlation development

A generalized temperature boundary condition coupling strategy for the modeling of conventional casting processes was implemented via experiments and numerical simulations with commercial purity aluminum, aluminum alloy, and tin specimens in copper, graphite, and sand molds. This novel strategy related the heat transfer coefficient at the metal-mold interface to the following process variables: the size of the air gap that forms at the metal-mold interface, the roughness of the mold surface, the conductivity of the gas in the gap, and the thermophysical properties of both the metal and mold. The objective of this study was to obtain, apply, and evaluate the effect of incorporating an experimentally derived relationship for specifying transient heat transfer coefficients in a general conventional casting process. The results are presented in two parts. Part I details the implementation of a systematic experimental approach not limited to a specific process to determine the heat transfer coefficient and characterize the formation of the air gap at the metal-mold interface. The heat transfer mechanisms at the interface were identified, and seen to vary in magnitude during four distinct stages, as the air gap formed and grew. A semiempirical inverse equation was used to characterize the heat transfer coefficient-air gap relationship, across the various stages, for experimental data from the literature and this study.     

Analysis of supercritical carbon dioxide extracts from cones and leaves of a Humulus lupulus L cultivar

Extracts obtained by supercritical carbon dioxide extraction of cones and leaves of hop (Humulus lupulus L) at different combinations of temperature and pressure were analysed for their α- and β-acids and volatiles, using HPLC and GC respectively. The yield and composition of the mixtures of bitter compounds from the cones were largely influenced by the temperature and pressure applied during the extraction. Bitter compounds could not be detected in the extracts from the leaves. The most important volatile components identified were β-myrcene, β-caryophyllene and α-humulene. The extraction parameters also influenced the composition of the mixture of volatiles from the cones and leaves, 40°C and 20.0 MPa being optimum for the extraction of both bitter compounds and volatiles, which are thought to be important for the aroma of beer.     

Characterisation of Conform? and conventionally extruded Al–4Mg–1Zr. Effect of extrusion route on superplasticity

Continuous extrusion (Conform™) is a well-established technique for the production of profiles from both solid and particulate feed stock. To the first approximation it is considered to be analogous to conventional extrusion, although there are significant differences in the metal flow during both processes. Metal flow during conventional extrusion is characterised by relatively low redundant work, whereas the Conform™ process requires significant redundant work to be successful. Most of the available scientific literature to date is concerned with the simulation of the Conform™ process and not its effect on the resultant microstructure of the product. In this paper, a detailed comparison of the microstructure, texture and superplastic properties developed during Conform™ and conventional extrusion for a particulate Al–4Mg–1Zr alloy are presented.