How Fast Do Algorithms Improve? [Point of View]

Algorithms determine which calculations computers use to solve problems and are one of the central pillars computer science. As algorithms improve, they enable scientists tackle larger explore new domains scientific techniques [1] , rid="ref2" xmlns:xlink="http://www.w3.org/1999/xlink">[2] . Bold claims have been made about pace algorithmic progress. For example, President’s Council Advisors on Science Technology (PCAST), a body senior that advise U.S. President, wrote in 2010 “in many areas, performance gains due improvements vastly exceeded even dramatic increased processor speed” rid="ref3" xmlns:xlink="http://www.w3.org/1999/xlink">[3] However, this conclusion was supported based data from progress linear solvers rid="ref4" xmlns:xlink="http://www.w3.org/1999/xlink">[4] is just single example. With no guarantee representative general, it unclear how broadly conclusions, such as PCAST’s, should be interpreted. Is faster most algorithms? Just some? How much average?