Abstract:Fundamental mathematical constants such as e and π are ubiquitous in diverse fields of science, from abstract mathematics and geometry to physics, biology and chemistry<a href="/articles/s41586-021-03229-4#ref-CR1">1</a>,<a href="/articles/s41586-021-03229-4#ref-CR2">2</a>. Nevertheless, for centuries new mathematical formulas relating fundamental constants have been scarce and usually discovered sporadically<a href="#ref-CR3">3</a>,<a href="#ref-CR4">4</a>,<a href="#ref-CR5">5</a>,<a href="/articles/s41586-021-03229-4#ref-CR6">6</a>. Such discoveries are often considered an act of mathematical ingenuity or profound intuition by great mathematicians such as Gauss and Ramanujan<a href="/articles/s41586-021-03229-4#ref-CR7">7</a>. Here we propose a systematic approach that leverages algorithms to discover mathematical formulas for fundamental constants and helps to reveal the underlying structure of the constants. We call this approach 'the Ramanujan Machine'. Our algorithms find dozens of well known formulas as well as previously unknown ones, such as continued fraction representations of π, e, Catalan's constant, and values of the Riemann zeta function. Several conjectures found by our algorithms were (in retrospect) simple to prove, whereas others remain as yet unproved. We present two algorithms that proved useful in finding conjectures: a variant of the meet-in-the-middle algorithm and a gradient descent optimization algorithm tailored to the recurrent structure of continued fractions. Both algorithms are based on matching numerical values; consequently, they conjecture formulas without providing proofs or requiring prior knowledge of the underlying mathematical structure, making this methodology complementary to automated theorem proving<a href="#ref-CR8">8</a>,<a href="#ref-CR9">9</a>,<a href="#ref-CR10">10</a>,<a href="#ref-CR11">11</a>,<a href="#ref-CR12">12</a>,<a href="/articles/s41586-021-03229-4#ref-CR13">13</a>. Our approach is especially attractive when applied to discover formulas for fundamental constants for which no mathematical structure is known, because it reverses the conventional usage of sequential logic in formal proofs. Instead, our work supports a different conceptual framework for research: computer algorithms use numerical data to unveil mathematical structures, thus trying to replace the mathematical intuition of great mathematicians and providing leads to further mathematical research.

A Programmable Approach to Evaluate Ramanujan Sums

A Novel 3780-Point FFT Processor Scheme for the Time Domain Synchronous OFDM System

An Algebraic Generalization of the Ramanujan Sum

On Arithmetic Cosine Transform Algorithm

Evaluations of Ramanujan Continued Fractions

Generating conjectures on fundamental constants with the Ramanujan Machine

Symmetric-Mapping LUT-Based Method and Architecture for Computing <i>X<SUP>Y</SUP></i>-Like Functions

A Memristor Crossbar-Based Computation Scheme with High Precision.

An Area-Efficient In-Memory Implementation Method of Arbitrary Boolean Function Based on SRAM Array

Ramanujan subspace pursuit for signal periodic decomposition

Summation formulae of multiplicative functions over arithmetic progressions and applications

Piecewise Parabolic Approximate Computation Based on an Error-Flattened Segmenter and a Novel Quantizer

A Multiplier Structure Based on A Novel Real-Time Csd Recoding

Proof and generalization of conjectures of Ramanujan Machine

Research on Multiplication Routine Based on Reconfigurable Four-Valued Logic Processor

An 8.8 TFLOPS/W Floating-Point RRAM-Based Compute-in-Memory Macro Using Low Latency Triangle-Style Mantissa Multiplication

Applications of Extended Kummer's Summation Theorem

Universal sums via products of Ramanujan's theta functions

A Parallel Multibit Programing Scheme with High Precision for RRAM-Based Neuromorphic Systems

FIR system identification under congruential summation-triggered communication scheme with binary observations

The Average Size of Ramanujan Sums over Cubic Number Fields