Pythagorean Theorem
$$a^2 + b^2 = c^2 \implies \|x\|_2 = \sqrt{\sum x_i^2}$$
Establishes the fundamental Euclidean metric and $L_2$ norm. Corner stone of inner products, distance geometry, and cosine similarity across high-dimensional AI embedding spaces.
Logarithms (John Napier, 1614)
$$\log(x \cdot y) = \log(x) + \log(y)$$
Transforms joint probability products into numerically tractable sums. Prevents floating-point underflow on GPUs when maximizing log-likelihood objectives.
The Derivative (Newton & Leibniz, 1668)
$$\frac{df}{dt} = \lim_{h \to 0} \frac{f(t+h) - f(t)}{h}$$
Quantifies instantaneous rate of change. Foundation of gradient descent ($\theta_{t+1} = \theta_t - \eta \nabla \mathcal{L}$) and the chain rule powering backpropagation.
Wave Equation (d'Alembert, 1746)
$$\frac{\partial^2 u}{\partial t^2} = c^2 \nabla^2 u$$
Describes propagation of disturbances at speed $c$. Crucial for acoustic signal modeling, speech synthesis, seismology, and wave-neural operator architectures.
Fourier Transform (J. Fourier, 1822)
$$\hat{f}(\xi) = \int_{-\infty}^\infty f(x) e^{-2\pi i x \xi} dx$$
Decomposes continuous functions into orthogonal frequency harmonics. Enables fast convolutions in $\mathcal{O}(N \log N)$ (FFT) and grounds Fourier Neural Operators (FNO).
Navier-Stokes Equations (1845)
$$\rho \left( \frac{\partial u}{\partial t} + u \cdot \nabla u \right) = -\nabla p + \mu \nabla^2 u + f$$
Nonlinear momentum conservation laws for viscous fluids. Benchmark problem par excellence for Physics-Informed Neural Networks and Clay Millennium Prize problem.
Second Law of Thermodynamics (Boltzmann, 1877)
$$S = k_B \log W, \quad dS \ge 0$$
Connects macroscopic entropy to accessible microstates $W$. Foundation for Boltzmann machines, simulated annealing, and modern score-based diffusion models.
Schrödinger Equation (1926)
$$i\hbar \frac{\partial}{\partial t} \Psi(r, t) = \hat{H} \Psi(r, t)$$
Governs deterministic evolution of quantum probability amplitudes in Hilbert space. Steers quantum computing algorithms and Quantum PINNs (Q-PINNs).
Chaos Theory / Logistic Map (May, 1976)
$$x_{t+1} = k x_t(1 - x_t)$$
Proves simple deterministic rules can produce aperiodic chaotic trajectories. Essential for understanding hyperparameter sensitivity and bifurcations in deep non-convex optimization.