Through the eighteenth century, calculus produced extraordinary results while resting on conceptual foundations that no one could quite explain. Infinitesimals — quantities smaller than any positive number but not zero — were used freely. Series were summed without anyone establishing they converged. Functions were assumed to be as well-behaved as their formulas suggested. Bishop George Berkeley mocked the practice in 1734 as the ghosts of departed quantities. The repair came in a thirty-year burst at the end of the nineteenth century — Cauchy in the 1820s, then Bolzano, Weierstrass, Riemann, Dedekind, Cantor — and produced real analysis: the ε-δ discipline that separated intuitive calculation from rigorous mathematics.
Real analysis is calculus done with every limit, derivative, and integral defined precisely and every theorem proved from first principles. The apparatus orbits one central property of the reals: completeness. A Cauchy sequence is one whose terms eventually get arbitrarily close to each other; in the rationals such a sequence may converge to nothing rational (partial sums for √2 are Cauchy but the limit is irrational), while in the reals every Cauchy sequence converges. That single fact distinguishes ℝ from ℚ and makes calculus on the reals work. Out of completeness fall the structural theorems the eighteenth century used freely without proof — Bolzano-Weierstrass on bounded sequences, Heine-Borel identifying compactness with closed-and-bounded subsets, the intermediate-value and extreme-value theorems, and the conditions under which the Fundamental Theorem of Calculus holds.
The second great theme is what happens when you extend the basic ideas. Series converge absolutely, conditionally, or not at all, and conditional series can have different sums depending on summation order. Uniform convergence preserves continuity in the limit; pointwise convergence does not. The Riemann integral works for most functions in elementary calculus but fails on those with too many discontinuities, and the more powerful Lebesgue integral — built on measure theory — repairs this. Pushing further leads into function spaces (Lᵖ, Banach, Hilbert), functional analysis, partial differential equations, rigorous quantum mechanics, and signal processing. The discipline of rigorous proof real analysis embodies is what separated mathematics from intuitive calculation; Berkeley's ghosts of departed quantities needed a hundred and fifty years to become a science.
Functional analysis — real and complex analysis on infinite-dimensional spaces — is essential to the rigorous formulation of quantum mechanics (wavefunctions live in a Hilbert space), signal processing (the Fourier transform on Lᵖ), partial differential equations (Sobolev spaces, weak solutions), and machine learning theory (kernel methods operate in reproducing kernel Hilbert spaces). Numerical analysis uses real-analytic estimates of error, convergence rates, and stability. Stochastic analysis — real analysis adapted to random processes, with Brownian motion and Itô calculus — underwrites mathematical finance (Black-Scholes), control theory, and modern probability. The discipline of rigorous proof real analysis embodies is what distinguishes mathematics from intuitive calculation.