Regularized Newton Method with Global $O(1/k^2)$ Convergence

Re:Regularized Newton Method with Global $O(1/k^2)$ Convergence by robbedpeter on Tuesday December 07, 2021 @11:04AM (#29474036)

It's funny how many complex ideas are "just" simpler ideas mapped to the right sequence of steps. The jargon and naming conventions can obscure the idea from outsiders, but allows for brief and accurate communication. If they'd explained the idea Barney-style, they'd have needed twice as many pages.

Thanks for the breakdown, it makes the paper more accessible.

[ Parent | on HN ]

Re:Regularized Newton Method with Global $O(1/k^2)$ Convergence by tubby12345 on Tuesday December 07, 2021 @02:48PM (#29477334)

Math is about formalizing "simple ideas" so that they can be reasoned about exactly.

[ Parent | on HN ]

Re:Regularized Newton Method with Global $O(1/k^2)$ Convergence by whatshisface on Tuesday December 07, 2021 @12:14PM (#29475251)

Every idea in math is like that, because the rules of math are to do everything in little steps.

[ Parent | on HN ]

Re:Regularized Newton Method with Global $O(1/k^2)$ Convergence by sorokod on Tuesday December 07, 2021 @11:51AM (#29474883)

Well a simpler idea is in way like the initial guess for Newton-Raphson. Sometimes you get convergence and sometimes you don't

[ Parent | on HN ]