Simulating the Cosmos

PUBLISHERS WEEKLY 8 MAI 2023

University of Edinburgh cosmologist Davé's dense debut comes up short in attempting to detail how astronomers use computer simulations to study the cosmos. Delving into "what it takes to build a universe on a computer," Davé explains that accounting for every star in the observable universe, itself only a fraction of outer space, would exceed the world's cumulative computing power, and that a comprehensive simulation would take "over 100,000 times the age of the Universe" to compute. The solution is to make compromises: astronomers reduce the complexity of simulations (and thus the processing power required to run them) by restricting their focus to discrete portions of space and lumping clusters of stars into a single data point. There are some intriguing tidbits (early simulations in the mid-1980s revealed that cosmic matter is arranged in "filaments, sheets and nodes," not randomly), but jargon-filled discussions of the formulas used to model cosmic bodies are difficult to follow, as when the author outlines a technique for simplifying the calculation of gravity: "The hard bit is computing the Fourier transform, but fortunately computer scientists have invented an algorithm known as a fast Fourier transform (FFT), which scales with the number of grid cells as Nlog2N." Burdened by technical language, this is tough going. Photos.