TOPICS
Search

Proof Mining


Proof mining is the logical analysis of mathematical proofs in order to extract explicit quantitative information. Proof interpretations can extract computable bounds, rates of convergence, rates of metastability, or algorithms from proofs that appear to establish only existence or convergence. The method is used especially in analysis, where the extracted bounds may be uniform over parameters that occur in the original theorem. Proof mining is also called proof unwinding.

For example, let (a_n) be a nondecreasing sequence with 0<=a_n<=1. Given epsilon>0 and a function g:N->N, define F(n)=n+g(n) and K=|_1/epsilon_|+1. There is an n<=F^((K))(0) such that

 |a_i-a_j|<epsilon,

for all integers i and j satisfying n<=i<=n+g(n) and n<=j<=n+g(n), where F^((K)) denotes K iterations of F. Otherwise, at each successive starting point F^((j))(0) for 0<=j<K, the sequence would increase by at least epsilon. Its total increase would then exceed 1, contradicting 0<=a_n<=1. This explicit finite bound is a simple instance of a rate of metastability extracted from a qualitative convergence statement.


See also

Constructive Proof, Proof Theory

Explore with Wolfram|Alpha

References

Kohlenbach, U. Applied Proof Theory: Proof Interpretations and Their Use in Mathematics. Berlin, Germany: Springer-Verlag, 2008. https://doi.org/10.1007/978-3-540-77533-1.

Cite this as:

Weisstein, Eric W. "Proof Mining." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/ProofMining.html

Subject classifications