Data Skeptic Modelling Evolution
May 9, 2024
Evolutionary biologist and software engineer Ben Haller discusses modeling population genetics over time, delving into natural selection sources like resources, mate preferences, and competition. Explore the capabilities of SLiM for diverse evolutionary scenarios, its impact on research, and transitioning from software engineering to evolutionary biology with a focus on modeling multiple species and ecological interactions.
AI Snips
Chapters
Transcript
Episode notes
Override Defaults For Custom Behaviors
- Override SLiM defaults by writing callbacks for behaviors like mate choice when default fitness-proportional selection isn't appropriate.
- Example: script mate selection by evaluating nearest neighbors' size or resources and add stochasticity for realism.
Large Manual And Workshops For Onboarding
- SLiM has a steep learning curve: combined SLiM and Eidos manuals exceed 1,000 pages with 150+ example recipes.
- Haller runs workshops to help users pass the initial learning hurdle before diverging into varied modeling paths.
Profile First Then Rework Algorithms
- Optimize SLiM models by profiling to find hotspots and then change algorithms rather than micro-optimizing code.
- Aim to reduce algorithmic complexity (e.g., from n squared to n log n) for biggest performance gains in large simulations.