An official website of the United States government
The concept of hidden genes was recently introduced in genetic algorithms (GAs) to handle systems architecture optimization problems, where the number of design variables is variable. Selecting the hidden genes in a chromosome determines the architecture of the solution. This paper presents two categories of mechanisms for selecting (assigning) the hidden genes in the chromosomes of GAs. These mechanisms dictate how the chromosome evolves in the presence of hidden genes. In the proposed mechanisms, a tag is assigned for each gene; this tag determines whether the gene is hidden or not. In the first category of mechanisms, the tags evolve using stochastic operations. Eight different variations in this category are proposed and compared through numerical testing. The second category introduces logical operations for tags evolution. Both categories are tested on the problem of interplanetary trajectory optimization for a space mission to Jupiter, as well as on mathematical optimization problems. Several numerical experiments were designed and conducted to optimize the selection of the hidden genes algorithm parameters. The numerical results presented in this paper demonstrate that the proposed concept of tags and the assignment mechanisms enable the hidden genes genetic algorithms (HGGA) to find better solutions.
more »
« less
Abstracting and formalising the design co-evolution model
Abstract Co-evolution accounts have generally been used to describe how problems and solutions both change during the design process. More generally, problems and solutions can be considered as analytic categories, where change is seen to occur within categories or across categories. There are more categories of interest than just problems and solutions, for example, the participants in a design process (such as members of a design team or different design teams) and categories defined by design ontologies (such as function-behaviour-structure or concept-knowledge). In this paper, we consider the co-evolution of different analytic categories (not just problems and solutions), by focussing on how changes to a category originate either from inside or outside that category. We then illustrate this approach by applying it to data from a single design session using three different systems of categorisation (problems and solutions, different designers and function, behaviour and structure). This allows us to represent the reciprocal influence of change within and between these different categories, while using a common notation and common approach to graphing quantitative data. Our approach demonstrates how research traditions that are currently distinct from each other (such as co-evolution, collaboration and function-behaviour-structure) can be connected by a single analytic approach.
more »
« less
- Award ID(s):
- 1762415
- PAR ID:
- 10356546
- Date Published:
- Journal Name:
- Design Science
- Volume:
- 8
- ISSN:
- 2053-4701
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Lockton, Dan; Lenzi, Sara (Ed.)Designers advance in the design processes by creating and expanding the design space where the solution they develop unfolds. This process requires the co- evolution of the problem and the solution spaces through design state changes. In this paper, we provide a methodology to capture how designers create, structure and expand their design space across time. Design verbalizations from a team of three professional engineers are coded into design elements from the Function-Behavior- Structure ontology to identify the characteristics of design state changes. Three types of changes can occur: a change within the problem space, a change within the solution space or a change between the problem and the solution spaces or inversely. The paper explores how to represent such changes by generating a network of design concepts. By tracking the evolution of the design space over time, we represent how the design space expands as the design activity progresses.more » « less
-
Iterative first-order methods such as gradient descent and its variants are widely used for solving optimization and machine learning problems. There has been recent interest in analytic or numerically efficient methods for computing worst-case performance bounds for such algorithms, for example over the class of strongly convex loss functions. A popular approach is to assume the algorithm has a fixed size (fixed dimension, or memory) and that its structure is parameterized by one or two hyperparameters, for example a learning rate and a momentum parameter. Then, a Lyapunov function is sought to certify robust stability and subsequent optimization can be performed to find optimal hyperparameter tunings. In the present work, we instead fix the constraints that characterize the loss function and apply techniques from robust control synthesis to directly search over algorithms. This approach yields stronger results than those previously available, since the bounds produced hold over algorithms with an arbitrary, but finite, amount of memory rather than just holding for algorithms with a prescribed structure.more » « less
-
null (Ed.)An adjunction is a pair of functors related by a pair of natural transformations, and relating a pair of categories. It displays how a structure, or a concept, projects from each category to the other, and back. Adjunctions are the common denominator of Galois connections, representation theories, spectra, and generalized quantifiers. We call an adjunction nuclear when its categories determine each other. We show that every adjunction can be resolved into a nuclear adjunction. The resolution is idempotent in a strict sense. The resulting nucleus displays the concept that was implicit in the original adjunction, just as the singular value decomposition of an adjoint pair of linear operators displays their canonical bases. In his seminal early work, Ross Street described an adjunction between monads and comonads in 2-categories. Lifting the nucleus construction, we show that the resulting Street monad on monads is strictly idempotent, and extracts the nucleus of a monad. A dual treatment achieves the same for comonads. This uncovers remarkably concrete applications behind a notable fragment of pure 2-category theory. The other way around, driven by the tasks and methods of machine learning and data analysis, the nucleus construction also seems to uncover remarkably pure and general mathematical content lurking behind the daily practices of network computation and data analysis.more » « less
-
Many variables in biological research—from body size to life-history timing to environmental characteristics—are measured continuously (e.g. body mass in kilograms) but analysed as categories (e.g. large versus small), which can lower statistical power and change interpretation. We conducted a mini-review of 72 recent publications in six popular ecology, evolution and behaviour journals to quantify the prevalence of categorization. We then summarized commonly categorized metrics and simulated a dataset to demonstrate the drawbacks of categorization using common variables and realistic examples. We show that categorizing continuous variables is common (31% of publications reviewed). We also underscore that predictor variables can and should be collected and analysed continuously. Finally, we provide recommendations on how to keep variables continuous throughout the entire scientific process. Together, these pieces comprise an actionable guide to increasing statistical power and facilitating large synthesis studies by simply leaving continuous variables alone. Overcoming the pitfalls of categorizing continuous variables will allow ecologists, ethologists and evolutionary biologists to continue making trustworthy conclusions about natural processes, along with predictions about their responses to climate change and other environmental contexts.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1017/dsj.2022.10
