This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: Particle accelerators are expensive. Not just to build or to run, but also to predict the outcome of measurements obtained from them.
Predicting the outcomes of particle collisions is essential to understanding the measured results seen in colliders. These predictions use quantum field theory (QFT), the language that describes particle behavior, and they have a major problem, according to Perimeter research faculty member Michael Borinsky. "We take a theoretical model of physics as input, translate it into the QFT language, and it's very hard to produce predictions," he says.
"What you put in encodes a law of nature, how particles interact locally, and what you want is some prediction of what follows on a larger scale. It's a pipeline from these local, very simple rules to things that can be measured." But predictions aren't just intellectually difficult. They're computationally difficult, more so with every digit calculated, which translates directly into expensive computer hours.
And as scientists investigate phenomena at increasingly tiny length scales, the predictions need to be more precise. Every digit counts, but each new digit requires significantly more compute. And significantly more money.
"This pipeline, this algorithm, has a flaw. It's very difficult and it's very slow," Borinsky explains. "It's just very hard to make these predictions, and they get more expensive if we want more accurate results.
Getting more digits becomes much harder at each step." Some of Borinsky's research on tropical QFTs was recently published in Ann. At its core, the QFT prediction problem is mathematical: predictions increase their computational demands factorially instead of exponentially, which outpaces exponential growth. That also means the cost to run these predictions increases factorially.
Extract — continue reading at the source.
