Continuing improvements in computing hardware are poised to transform capabilities for in silico modeling of cross-scale phenomena underlying major open questions in evolutionary biology and artificial life, such as transitions in individuality, eco-evolutionary dynamics, and rare evolutionary events. Emerging ML/AI-oriented hardware accelerators, like the 850,000 processor CerebrasWafer Scale Engine (WSE), hold particular promise. However, many practical challenges remain in conducting informative evolution experiments that efficiently utilize these platforms’ large processor counts. Here, we focus on the problem of extracting phylogenetic information from agent-based evolution on the WSE platform. This goal drove significant refinements to decentralized in silico phylogenetic tracking, reported here. These improvements yield order-of-magnitude performance improvements. We also present an asynchronous island-based genetic algorithm (GA) framework forWSE hardware. Emulated and on-hardware GA benchmarks with a simple tracking-enabled agent model clock upwards of 1 million generations a minute for population sizes reaching 16 million agents. This pace enables quadrillions of agent replication events a day. We validate phylogenetic reconstructions from these trials and demonstrate their suitability for inference of underlying evolutionary conditions. In particular, we demonstrate extraction, from wafer-scale simulation, of clear phylometric signals that differentiate runs with adaptive dynamics enabled versus disabled. Together, these benchmark and validation trials reflect strong potential for highly scalable agent-based evolution simulation that is both efficient and observable. Developed capabilities will bring entirely new classes of previously intractable research questions within reach, benefiting further explorations within the evolutionary biology and artificial life communities across a variety of emerging high-performance computing platforms.

The addition of parasites to a host population can drive an escalation in the host population's phenotypic complexity – even in the absence of a direct fitness advantage for this increase. Parasites restrict certain regions of the genotype space, decreasing the fitness and the probability of survival of particular host phenotypes. While many artificial life frameworks model a direct correlation between genotype and fitness, the structure of genotype-phenotype maps can have important effects on evolutionary dynamics. Using a simple coarse-grained model for phenotypic transitions during evolution, we show that the escalation in phenotypic complexity under neutral co-evolution is dependent on the structure of the genotype-phenotype map. We discuss these results using the metaphor of evolutionary spandrels and highlight how these structural considerations might allow us to capture biological phenomena more accurately.

Alife has made fundamental contributions to our understanding of how evolutionary processes work. I will highlight a few of these instances, as well as ongoing work by my group and others embracing digital organisms within more traditional biological boundaries. However, there is a history of artificial life studies that are often overlooked by related disciplines. I don't mean this as a critique of either field. Instead, I would argue it's more of an opportunity. Artificial life has always been pushing the boundaries of truly interdisciplinary science, and as traditional fields expand their own horizons, old discoveries from the artificial life community are waiting to be newly embraced. This has been the promise of interdisciplinary fields, and Alife is well positioned to deliver.

Cooperation is a defining attribute of life as we know it, from the delicate interactions of intracellular components to social behavior in groups. However, defection and exploitation are at least as ubiquitous. Evolutionary game theory is a successful tool for investigating how cooperation may be maintained despite large advantages for defection. The Prisoners Dilemma is one such game where spatial structure can maintain cooperation, but only if the benefit-to-cost ratio (b/c) is greater than some threshold, which appears to be the average number of neighbors (k). However, this inequality was tested only for regular spatial and irregular non-spatial networks. In this paper, we use networks in Cartesian space that are based on radii of interactions. We investigate whether the b/c > k threshold holds for these irregular spatial networks, and we use a much broader range of k than previously studied. We find that this rule, and other related inequalities, hold well for the larger radii even when there is noise in the expected neighborhood size. As the expected neighborhood size increases, so does the variation in the empirical edge distribution. However, the variation in the threshold for cooperation decreases. This paper is a first step in a broad investigation of how uncertainty affects the outcome of game theoretic simulations.

Fitness landscapes are visual metaphors that appeal to our intuition for real-world landscapes to help us understand how populations evolve. The object inspiring the metaphor is better described as a networks composed of all possible genotypes, but they are frequently simplified to a surface where the fitness of each genotype is represented by elevation. Selection drives evolving populations to ascend the landscape until they are dominated by genotypes from which no further beneficial mutations are likely, known as a peak. However, by allowing for environmental change, former peaks can vanish, forcing populations to resume adapting. To explore how changing environments affect adaptation, we used the digital evolution platform, Avida, wherein we could manipulate the organisms’ environment as they are subject to natural evolutionary forces. We found that transient exposure to alternate environments frequently resulted in more fit genotypes. Negative-frequency-dependent environments, in particular, yielded strong fitness benefits after returning to the original environment. Furthermore, we explored how such environmental change could yield adaptive benefits via valley crossing and how such knowledge could be exploited in systems where improving the rate of adaption is beneficial.