Recent years have experienced growth in experimental mathematics, utilizing modern computational tools for mathematical research.
This practice is widespread, involving many researchers across various fields, including number theory, algebra, analysis, geometry, and topology.
Key tasks include generating examples, creating plots, manipulating algebra, testing conjectures, and exploring proofs.
Computation aids mathematicians by providing time-efficient conjecture testing, largely due to the development of advanced symbolic mathematics software.
Significant tools:
Maple and Mathematica: Leading symbolic mathematics software.
Magma: Computational algebra package from the University of Sydney.
Other tools for integer sequences and constants.
Gordon Moore's prediction in 1965 about hardware complexity has shown consistent accuracy, leading to unprecedented hardware advancements.
Experimental Mathematics Journal: Established 12 years ago, reflects this shift in mathematical research perspective.
Past perception that “real mathematicians don’t compute” has evolved significantly, with extensive computational papers contributing to mainstream mathematics.
Definition: Searches for linear relationships satisfied by a set of numerical values represented by vectors.
Integer relation algorithm finds integer coefficients such that a linear combination of the vector equals zero.
Requires high precision and can lead to significant computational requirements; for instance, a 50,000-digit calculation was necessary for a recent result.
The PSLQ algorithm: The best-known method for detecting integer relations, recognized for its efficiency.
A notable application of PSLQ: The BBP formula for π:
Allows for the calculation of binary or hexadecimal digits of π without computing preceding digits, conserving memory.
Example of similar calculations includes the binary expansion for log 2.
The BBP approach has led to various other formulas allowing for arbitrary-position digit calculations for different constants.
Despite efforts, no BBP-type formula for π exists in bases other than powers of two.
Recent findings indicate the computational exploration has significant implications for the normality of mathematical constants, including π and log 2.
Definitions:
b-normal: A real number is b-normal if every possible string of digits in base b appears with a limiting frequency.
Many known constants are suspected to be normal, but proven examples are few.
Multi-zeta constants generalize the Riemann zeta function and are crucial for modern mathematical evaluations.
Calculations of complex sums lead to interesting numerical identities, heuristically verified through computer experiments instead of traditional proof methods.
Advanced numerical methods offer high-precision results, often exceeding typical software limits.
Innovations such as tanh-sinh quadrature significantly improve integration accuracy over traditional methods, revealing new mathematical patterns and facilitating research implementations.
Ramanujan discovered noteworthy relationships involving continued fractions, experimentally leading to conclusions about logarithmic constants in mathematics.
The intersection of computational experimentation and traditional mathematical research is increasingly prominent.
Future advances in software and hardware promise further impact on research methodologies and findings.
Ongoing exploration raises questions about rigorous proof versus computational evidence, reaffirming the necessity of both in the mathematical community.