Generic systolic array for genetic algorithms

The authors present a systolic design for a simple GA mechanism which provides high throughput and unidirectional pipelining by exploiting the inherent parallelism in the genetic operators. The design computes in O(N+G) time steps using O(N2) cells where N is the population size and G is the chromosome length. The area of the device is independent of the chromosome length and so can be easily scaled by replicating the arrays or by employing fine-grain migration. The array is generic in the sense that it does not rely on the fitness function and can be used as an accelerator for any GA application using uniform crossover between pairs of chromosomes. The design can also be used in hybrid systems as an add-on to complement existing designs and methods for fitness function acceleration and island-style population management

Synthesis of a systolic array genetic algorithm

The paper presents a design for a hardware genetic algorithm which uses a pipeline of systolic arrays. These arrays have been designed using systolic synthesis techniques which involve expressing the algorithm as a set of uniform recurrence relations. The final design divorces the fitness function evaluation from the hardware and can process chromosomes of different lengths, giving the design a generic quality. The paper demonstrates the design methodology by progressively re-writing a simple genetic algorithm, expressed in C code, into a form from which systolic structures can be deduced. This paper extends previous work by introducing a simplification to a previous systolic design for the genetic algorithm. The simplification results in the removal of 2N 2 + 4N cells and reduces the time complexity by 3N + 1 cycles.

The systolic array genetic algorithm, an example of systolic arrays as a reconfigurable design methodology

We advocate the use of systolic design techniques to create custom hardware for Custom Computing Machines. We have developed a hardware genetic algorithm based on systolic arrays to illustrate the feasibility of the approach. The architecture is independent of the lengths of chromosomes used and can be scaled in size to accommodate different population sizes. An FPGA prototype design can process 16 million genes per second.

Implementing a generic systolic array for genetic algorithms

We have designed a highly parallel design for a simple genetic algorithm using a pipeline of systolic arrays. The systolic design provides high throughput and unidirectional pipelining by exploiting the implicit parallelism in the genetic operators. The design is significant because, unlike other hardware genetic algorithms, it is independent of both the fitness function and the particular chromosome length used in a problem. We have designed and simulated a version of the mutation array using Xilinix FPGA tools to investigate the feasibility of hardware implementation. A simple 5-chromosome mutation array occupies 195 CLBs and is capable of performing more than one million mutations per second. I. Introduction Genetic algorithms (GAs) are established search and optimization techniques which have been applied to a range of engineering and applied problems with considerable success [1]. They operate by maintaining a population of trial solutions encoded, using a suitable encoding scheme.

Nutritional and safety assessments of foods and feeds nutritionally improved through biotechnology: Case studies

During the last 2 decades, the public and private sectors have made substantial international research progress toward improving the nutritional value of a wide range of food and feed crops. Nevertheless, significant numbers of people still suffer from the effects of undernutrition. In addition, the nutritional quality of feed is often a limiting factor in livestock production systems, particularly those in developing countries. As newly developed crops with nutritionally improved traits come closer to being available to producers and consumers, we must ensure that scientifically sound and efficient processes are used to assess the safety and nutritional quality of these crops. Such processes will facilitate deploying these crops to those world areas with large numbers of people who need them. This document describes 5 case studies of crops with improved nutritional value. These case studies examine the principles and recommendations published by the Intl. Life Sciences Inst. (ILSI) in 2004 for the safety and nutritional assessment of foods and feeds derived from nutritionally improved crops (ILSI 2004). One overarching conclusion that spans all 5 case studies is that the comparative safety assessment process is a valid approach. Such a process has been endorsed by many publications and organizations, including the 2004 ILSI publication. The type and extent of data that are appropriate for a scientifically sound comparative safety assessment are presented on a case-by-case basis in a manner that takes into account scientific results published since the 2004 ILSI report. This report will appear in the January issue of Comprehensive Reviews in Food Science and Food Safety.

Evaluation of NRC, UC Davis and ADAS approaches to estimate the metabolizable energy values of feeds at maintenance energy intake from equations utilizing chemical assays and in vitro determinations

Feed samples received by commercial analytical laboratories are often undefined or mixed varieties of forages, originate from various agronomic or geographical areas of the world, are mixtures (e.g., total mixed rations) and are often described incompletely or not at all. Six unified single equation approaches to predict the metabolizable energy (ME) value of feeds determined in sheep fed at maintenance ME intake were evaluated utilizing 78 individual feeds representing 17 different forages, grains, protein meals and by-product feedstuffs. The predictive approaches evaluated were two each from National Research Council [National Research Council (NRC), Nutrient Requirements of Dairy Cattle, seventh revised ed. National Academy Press, Washington, DC, USA, 2001], University of California at Davis (UC Davis) and ADAS (Stratford, UK). Slopes and intercepts for the two ADAS approaches that utilized in vitro digestibility of organic matter and either measured gross energy (GE), or a prediction of GE from component assays, and one UC Davis approach, based upon in vitro gas production and some component assays, differed from both unity and zero, respectively, while this was not the case for the two NRC and one UC Davis approach. However, within these latter three approaches, the goodness of fit (r(2)) increased from the NRC approach utilizing lignin (0.61) to the NRC approach utilizing 48 h in vitro digestion of neutral detergent fibre (NDF:0.72) and to the UC Davis approach utilizing a 30 h in vitro digestion of NDF (0.84). The reason for the difference between the precision of the NRC procedures was the failure of assayed lignin values to accurately predict 48 h in vitro digestion of NDF. However, differences among the six predictive approaches in the number of supporting assays, and their costs, as well as that the NRC approach is actually three related equations requiring categorical description of feeds (making them unsuitable for mixed feeds) while the ADAS and UC Davis approaches are single equations, suggests that the procedure of choice will vary dependent Upon local conditions, specific objectives and the feedstuffs to be evaluated. In contrast to the evaluation of the procedures among feedstuffs, no procedure was able to consistently discriminate the ME values of individual feeds within feedstuffs determined in vivo, suggesting that the quest for an accurate and precise ME predictive approach among and within feeds, may remain to be identified. (C) 2004 Elsevier B.V. All rights reserved.

Inter-laboratory variation of in vitro cumulative gas production profiles of feeds using manual and automated methods

A study was conducted to estimate variation among laboratories and between manual and automated techniques of measuring pressure on the resulting gas production profiles (GPP). Eight feeds (molassed sugarbeet feed, grass silage, maize silage, soyabean hulls, maize gluten feed, whole crop wheat silage, wheat, glucose) were milled to pass a I mm screen and sent to three laboratories (ADAS Nutritional Sciences Research Unit, UK; Institute of Grassland and Environmental Research (IGER), UK; Wageningen University, The Netherlands). Each laboratory measured GPP over 144 h using standardised procedures with manual pressure transducers (MPT) and automated pressure systems (APS). The APS at ADAS used a pressure transducer and bottles in a shaking water bath, while the APS at Wageningen and IGER used a pressure sensor and bottles held in a stationary rack. Apparent dry matter degradability (ADDM) was estimated at the end of the incubation. GPP were fitted to a modified Michaelis-Menten model assuming a single phase of gas production, and GPP were described in terms of the asymptotic volume of gas produced (A), the time to half A (B), the time of maximum gas production rate (t(RM) (gas)) and maximum gas production rate (R-M (gas)). There were effects (P<0.001) of substrate on all parameters. However, MPT produced more (P<0.001) gas, but with longer (P<0.001) B and t(RM gas) (P<0.05) and lower (P<0.001) R-M gas compared to APS. There was no difference between apparatus in ADDM estimates. Interactions occurred between substrate and apparatus, substrate and laboratory, and laboratory and apparatus. However, when mean values for MPT were regressed from the individual laboratories, relationships were good (i.e., adjusted R-2 = 0.827 or higher). Good relationships were also observed with APS, although they were weaker than for MPT (i.e., adjusted R-2 = 0.723 or higher). The relationships between mean MPT and mean APS data were also good (i.e., adjusted R 2 = 0. 844 or higher). Data suggest that, although laboratory and method of measuring pressure are sources of variation in GPP estimation, it should be possible using appropriate mathematical models to standardise data among laboratories so that data from one laboratory could be extrapolated to others. This would allow development of a database of GPP data from many diverse feeds. (c) 2005 Published by Elsevier B.V.