>>512004692
I can actually. This work was authored with Emmanuel Candès who is now at Stanford. The purpose was to improve signal reconstruction from eg xrays and medical scan machines like MRIs, which have a limited amount of data due to safety and engineering constraints. Tao and Candès (but mostly Tao) were able to prove that you can get MUCH better reconstruction, near perfect in fact, by solving a sequence of convex problems, even if you sample below the Nyquist rate. This was a huge deal, since it allows for extremely accurate imaging (and hence diagnoses) in medical settings.
These ideas are part of a field known a compressed sensing, where you need to extract a lot of information from resolution limited signals. It’s used in the military and robotics for obvious reasons, where sensors are expensive and you need to optimize the information deduced from what you have.
It’s not all bullshit. Just because you stopped doing math after the pythagorean theorem doesn’t mean MRI machines and airplanes are made of magic.