Browsing by Subject "Genetic"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    A Comparison of the Genetic Algorithm and the Mixing Genetic Algorithm
    (2020-07) Gulfam, Muhammad
    Genetic Algorithms (GAs) are optimization techniques inspired by the idea of evolution. They can sometimes take a long time to find the solution to a problem, but it is not always obvious when, or how to configure their various parameters. Recently, a new GA was introduced [8] that has a lot of potential for parallelization. This algorithm, called the Mixing Genetic Algorithm, has shown promising results on the well-known Traveling Salesman Problem. In this work, we have compared the effectiveness of the Mixing GA over a traditional GA on three discrete optimization problems: the OneMax problem and two topologies of the Ising Model (Ising Model on Tree and Ising Model on Ring). The comparison has been done for the success rate at the given time, for the given problem size and size of population. The comparison has been done for, both, serial and parallel implementations. Overall, the success rate for the Mixing GA is better than the traditional GA. We have also compared two population selection methods, namely, tournament selection and generational population selection. The tournament selection outperformed generational population selection for all the problems and problem sizes that we experimented with.
  • Thumbnail Image
    Item
    Multivariate DNA taphonomy: evaluating the effects of environmental context, specimen properties, and laboratory strategies on the preservation and detection of DNA in ancient and challenging specimens.
    (2012-05) Alveshere, Andrea Joanna
    Within their diminutive structures, DNA molecules hold tantalizing potential to address myriad questions about human history, prehistory, and the evolution and dispersal of all forms of life. When accessible and accurate, DNA from ancient and degraded specimens can elucidate many topics of interest to researchers in a variety of fields including archaeology, biological anthropology, forensics, conservation and evolutionary biology, agronomy, and medicine. Despite the great informational potential of genetic studies, the high cost and destructive nature of DNA analyses discourage many researchers from submitting archaeological specimens for testing. A diversity of DNA detection protocols, the limited scope of individual research projects, and a bias toward publishing successful results make it difficult to evaluate the comparative influence of different preservation factors, field methods, and laboratory strategies on the recovery of useful genetic information from ancient and degraded specimens. The work presented in this manuscript is predicated upon the contention that the opportunity to conduct ancient DNA research entails an obligation to make the most of every specimen fragment consumed, every data point collected, and every funding dollar spent. The scope of this project is to develop a system for evaluating whether DNA testing might be appropriate for a given specimen; for determining which steps can be taken to increase the chances of recovering useful data; and for maximizing the contribution of individual research projects, conducted across disparate fields, to the greater body of knowledge on DNA preservation and detection. This endeavor involved: (1) inventorying variables having potential to influence DNA preservation and/or detection; (2) investigating subsets of the candidate variables through case studies of archaeological materials from Kromdraai, Wonderwerk Cave, and Border Cave, South Africa, Silvernale Village, Minnesota, and UV-irradiated forensic-type samples; and (3) the development and validation (via case study data) of the Biomolecular Preservation and Detection Information System (BIOPADIS™), a standardized system for synthesis, management, and analysis of biomolecular taphonomy data. BIOPADIS™ (\bī-’op-ad-is\) comprises a relational database that accommodates all manner of relevant data, a querying capability that makes these data accessible, and a set of statistical approaches appropriate for identifying and evaluating correlations within these multivariate, multi-study data.