Understanding how galaxies form and evolve requires increasingly sophisticated simulations, but with greater physical realism comes a substantially higher cost of entry. Fortunately, high-quality science doesn't always have to mean paying this price. Sometimes, being smarter about how we spend our computational resources (i.e. thinking like a cheapskate) can get us just as far, or even further!
In this talk, I'll explore two complementary approaches to getting more astrophysical "bang for your buck". The first is cosmological zoom simulations, where concentrating computational resources on only the regions of the Universe we care about dramatically reduces the cost of achieving our scientific goals. I'll introduce the concept of zoom simulations, showcase some of the science they have enabled, and discuss the technical challenges I've encountered in developing and optimising these simulations.
Producing a simulation is only half the battle, though. To understand what simulations predict for real observations, we need to leave the world of theoretical outputs and "forward model" into the observational world. A process rife with uncertainty and computational cost of its own. This is where Synthesizer comes in: an open-source framework for efficiently generating synthetic observables, with performance and flexibility at its foundations. I'll introduce the philosophy behind its development, demonstrate some of its capabilities, and highlight the science it has enabled so far.
Ultimately, being a computational cheapskate isn't just about doing the same science for less. It's about making more ambitious science possible with the resources we already have, pushing the boundaries of what we thought we could achieve.
Description