2:00 Matthew Andres Moreno: Lessons Learned Administering a Summer Workshop.
The Workshop for Avida-ED Software Development (WAVES) is a ten-week, full-time software development experience that matches early-career participants with mentors working on software for digital evolution and public-facing science outreach.
The workshop came together abruptly in 2020 in response to the sudden need for fully-virtual research experiences.
In addition to outlining the objectives and format of the WAVES workshop, in this talk, I’ll be sharing lessons I learned as the lead workshop coordinator, touching on topics such as
soliciting and reviewing applications,
organizing workshop materials,
devising team-building and social activities,
running virtual group meetings, and
showcasing participant achievements.
As a first-time coordinator now rounding out my second year in the role, there have been a lot of lessons learned!
2:15 Charles Ofria, Emily Dolson, and Matthew Andres Moreno: Empirical: A Scientific Software Library for Research, Education, and Public Engagement.
The objective of the Empirical C++ Library is to facilitate efficient, reliable, and broadly accessible scientific software.
The ubiquitous availability of internet-connected mobile and desktop computing creates incredible potential for low-barrier, hands-on scientific outreach across broad educational and general audiences.
The Emscripten compiler, which enables source code originally written for research computing to run in web browsers, provides an exciting opportunity to realize the potential of research software for public outreach via installation-free access and rich graphical user interfaces.
This talk will provide a hands-on introduction to Empirical’s web tools.
In service of our other objectives — efficiency and reliability — Empirical also includes more general-purpose debugging, metaprogramming, and data management tools, as well as utilities specially relevant to digital evolution such as phylogenetic trackers and the MABE2 framework.
Our presentation will touch on these topics as well.
2:30 Emily Dolson and Abigail Wilson: A randomization-based approach to normalizing phylogenetic metrics?
Phylogenetic tree analysis is a crucial tool for understanding evolutionary history.
When measuring phylogenies of different sizes and shapes, it is important to have viable metrics that can be used for comparison.
However, many phylogeny measurements are very sensitive to tree size.
For example, phylogenetic diversity gets inflated as phylogenies get larger, making it difficult to meaningfully compare over time or between different populations.
This problem is exacerbated in the context of digital evolution, where many assumptions about reconstructed phylogenies from nature are violated and it is unclear whether traditional approaches to normalizing phylogeny metrics are valid.
This presentation will discuss approaches to normalizing phylogeny metrics in computational models.
Specifically, we will be examining a randomization based approach in which thousands of hypothetical trees are generated and metrics for those trees are recorded and stored for normalization, providing a wide spectrum of possible tree shapes and sizes.
The optimal end result of this project is to provide more insightful metrics to a user by producing both a numerical value for their measurement as well as its corresponding percentile mark.
2:45 Santiago Rodriguez Papa, Matthew Andres Moreno, and Charles Ofria: SignalGP-Lite: Event Driven Genetic Programming Library for Large-Scale Artificial Life Applications.
Event-driven genetic programming representations have been shown to outperform traditional imperative representations on interaction-intensive problems.
These representations organize genome content into modules that are triggered in response to environmental signals, simplifying simulation design and implementation.
Existing work developing event-driven genetic programming methodology has largely used the SignalGP library, which caters to traditional program synthesis applications.
The SignalGP-Lite library enables larger-scale artificial life experiments with streamlined agents by reducing control flow overhead and trading run-time flexibility for better performance due to compile-time configuration.
Here, we report benchmarking experiments that show an 8x to 30x speedup.
We also report solution quality equivalent to SignalGP on two benchmark problems originally developed to test the ability of evolved programs to respond to a large number of signals and to modulate signal response based on context.
3:00 Lanea Rohan and Aria Killebrew Bruehl: Planning for the future of MABE2: A summer of documentation and testing.
The second Modular Agent-Based Evolver framework (MABE2) is an open-source research platform that provides accessible tools for conducting evolutionary computation and digital evolution research.
MABE2 reduces the time between constructing a hypothesis and generating results by providing a library of modules that connect to form a variety of experiments.
To promote use among interdisciplinary researchers, modules are connected and adjusted via a simple text interface (i.e., the user does not need to add or edit any code).
However, if the user requires modules beyond the existing library, MABE2 provides a set of practical tools for developing additional modules.
With the understanding that MABE2 is a large piece of software, this summer we created a documentation guide and testing framework as part of the 2021 Workshop for Avida-ED Software Development (WAVES).
In this talk, we will highlight the role of the documentation and testing framework in the MABE2 user experience through a demonstration of constructing and running a custom experiment.
By creating the documentation and testing framework, we hope to make MABE2 more approachable to new users and more useful to the interdisciplinary research community.
3:15 Tait Weicht, Matthew Andres Moreno, and Charles Ofria: Moving scientific computation into the browser.
Providing the means to replicate an experiment is a powerful tool for scientific engagement with both the general public and one’s professional peers.
Modern web browsers are now powerful enough to run complex simulations with dynamic graphical user interfaces inside of a web application, shareable with anyone by just copying a link.
However, a steep learning curve in modern web design and cross-compiling to target web browsers can prevent researchers from taking advantage of this platform.
Empirical, a C++ library, aims to streamline the browser-based deployment of scientific software.
We give an overview of some of the prefabricated Bootstrap-based web components provided by Empirical that can quickly turn a command-line tool into a browser-based one.
We also discuss some approaches to responsive web development to support mobile devices in the context of scientific web applications.