Formulas & Methods

Study Sheet

Formulas & Methods

The Putnam reference sheet: analysis, algebra, and the named integrals in one place

Analysis

Tip
Limits, series, integrals

Limits as derivatives (n(a1/n1)lnan(a^{1/n} - 1) \to \ln a), as Riemann sums (1nf(kn)01f\tfrac1n\sum f(\tfrac kn) \to \int_0^1f), as exponentials ((1+an)nea(1 + \tfrac an)^n \to e^a); Stolz–Cesàro; Taylor with Lagrange remainder; the three series (xn\sum x^n, xnn\sum\tfrac{x^n}{n}, xnn!\sum\tfrac{x^n}{n!}) and their derivatives; ζ(2)=π26\zeta(2) = \tfrac{\pi^2}{6}, η(2)=π212\eta(2) = \tfrac{\pi^2}{12}; telescoping with lag; Cauchy condensation; Abel summation anbn=ANbNAn(bn+1bn)\sum a_nb_n = A_Nb_N - \sum A_n(b_{n+1} - b_n).

Rewrite limits as one of the three shapes; expand in Taylor series to second order; swap sums and integrals only after stating the justification (absolute/uniform convergence, monotone or dominated convergence).

Tip
The named integrals

Feynman: I(a)=afI'(a) = \int\partial_af; Dirichlet 0sinxx=π2\int_0^\infty\tfrac{\sin x}{x} = \tfrac\pi2 and 0sin2xx2=π2\int_0^\infty\tfrac{\sin^2x}{x^2} = \tfrac\pi2; Gaussian ex2=π\int e^{-x^2} = \sqrt\pi and 0ex2cos2bx=π2eb2\int_0^\infty e^{-x^2}\cos 2bx = \tfrac{\sqrt\pi}{2}e^{-b^2}; King abf=abf(a+bx)\int_a^bf = \int_a^bf(a+b-x); Queen/Jack folding; Glasser f(x1x)=f\int f(x - \tfrac1x) = \int f; the 1/x1/x reflection; Frullani 0f(ax)f(bx)x=(f(0)f())lnba\int_0^\infty\tfrac{f(ax) - f(bx)}{x} = (f(0) - f(\infty))\ln\tfrac ba; Gamma Γ(n+1)=n!\Gamma(n+1) = n!, Γ(12)=π\Gamma(\tfrac12) = \sqrt\pi, Γ(s)Γ(1s)=πsinπs\Gamma(s)\Gamma(1-s) = \tfrac{\pi}{\sin\pi s}; Beta B(p,q)=Γ(p)Γ(q)Γ(p+q)B(p,q) = \tfrac{\Gamma(p)\Gamma(q)}{\Gamma(p+q)}; 0dx1+xn=π/nsin(π/n)\int_0^\infty\tfrac{dx}{1+x^n} = \tfrac{\pi/n}{\sin(\pi/n)}; log-sine 0π/2lnsinx=π2ln2\int_0^{\pi/2}\ln\sin x = -\tfrac\pi2\ln2; 01ln(1+x)1+x2=π8ln2\int_0^1\tfrac{\ln(1+x)}{1+x^2} = \tfrac\pi8\ln2; 0ln(1+x2)1+x2=πln2\int_0^\infty\tfrac{\ln(1+x^2)}{1+x^2} = \pi\ln2; Fresnel 0sinx2=12π2\int_0^\infty\sin x^2 = \tfrac12\sqrt{\tfrac\pi2}; Wallis 4n24n21=π2\prod\tfrac{4n^2}{4n^2-1} = \tfrac\pi2; Weierstrass t=tanx2t = \tan\tfrac x2; series expansion and swap.

Order of attack: symmetry (xa+bxx \to a + b - x, x1xx \to \tfrac1x) → parameter (Feynman) → named shape (Frullani, Beta, Gamma, Dirichlet) → substitution (tangent, Weierstrass) → series. Never start with parts.

Tip
Inequalities and convexity

AM–GM, Cauchy–Schwarz (sums and integrals), Jensen, rearrangement, power means; Hermite–Hadamard f(a+b2)1baabff(a)+f(b)2f(\tfrac{a+b}{2}) \le \tfrac{1}{b-a}\int_a^bf \le \tfrac{f(a) + f(b)}{2} for convex ff; tangent-line trick; MVT-derived bounds (ex1+xe^x \ge 1 + x, lnxx1\ln x \le x - 1, sinxx\sin x \le x).

Reminder — The Mean Value Theorem:f(c)=f(b)f(a)bafor some c(a,b)f'(c)=\frac{f(b)-f(a)}{b-a} \quad\text{for some } c \in (a,b)

Identify the equality case first; convert integral inequalities to their sum versions mentally, then apply the same proof.

Algebra

Tip
Linear algebra

Matrix determinant lemma det(I+uvT)=1+vTu\det(I + \mathbf u\mathbf v^{\mathsf T}) = 1 + \mathbf v^{\mathsf T}\mathbf u and det(aI+bJ)=an1(a+nb)\det(aI + bJ) = a^{n-1}(a + nb); Vandermonde i<j(xjxi)\prod_{i<j}(x_j - x_i); circulant eigenvalues by DFT; tridiagonal Toeplitz eigenvalues a+2bcoskπn+1a + 2b\cos\tfrac{k\pi}{n+1}; detp(A)=p(λi)\det p(A) = \prod p(\lambda_i), trAk=λik\mathrm{tr}A^k = \sum\lambda_i^k; Cayley–Hamilton; rank–nullity, rank(AB)min\mathrm{rank}(AB) \le \min; polynomial-entry matrices have rank deg+1\le \deg + 1; GLn(Fq)=(qnqk)|GL_n(\mathbb F_q)| = \prod(q^n - q^k); spectral theorem and Rayleigh quotient; trAk=0\mathrm{tr}A^k = 0 for all knk \le n implies nilpotent.

Row-reduce two rows before any expansion. Look for identity-plus-low-rank. Move to the spectrum for anything involving powers or polynomials of a matrix.

Tip
Polynomials, number theory, abstract algebra

Vieta and Newton; roots of unity filter; auxiliary polynomial with known roots for values at integers; Chebyshev; Eisenstein; abP(a)P(b)a - b \mid P(a) - P(b). Orders and CRT; LTE; Legendre/Kummer; QR supplements and reciprocity; Wilson; multiplicative functions and dnφ(d)=n\sum_{d\mid n}\varphi(d) = n; bounding between squares. Lagrange's theorem; cyclic-group counts (φ(d)\varphi(d) elements of order dd, one subgroup per divisor); orders in SnS_n as lcm of cycle lengths; a polynomial over a field has at most deg\deg roots; finite closed subsets are subgroups.

Reminder — Vieta's formulas:r1+r2=ba,r1r2=car_1+r_2=-\frac{b}{a},\qquad r_1 r_2=\frac{c}{a}

For values of a polynomial at consecutive integers, build the auxiliary polynomial. For group questions, use only closure, Lagrange, and counting — nothing heavier is expected.

Combinatorics, Probability & Geometry

Tip
Combinatorics and probability

Bijections; PIE; derangements; Catalan and the reflection principle; generating functions and xddxx\tfrac{d}{dx}; Erdős–Szekeres; Mantel/Turán, Hall, Euler; extremal and pigeonhole; games (P/N, Nim, strategy stealing). Linearity with indicators; tail-integral E[X]=P(X>t)dtE[X] = \int P(X > t)dt; order statistics E[max]=nn+1E[\max] = \tfrac{n}{n+1}; coupon collector nHnnH_n; gambler's ruin kN\tfrac kN, duration k(Nk)k(N-k); random walk returns (2nn)/4n\binom{2n}{n}/4^n; first-step analysis.

Write indicator sums before distributions. For "expected time," write first-step equations. For random geometry, draw the region and integrate.

Tip
Geometry

Pick's theorem; gcd+1\gcd + 1 lattice points on a segment; Minkowski; Gauss circle count πr2+O(r)\pi r^2 + O(r); complex numbers (rotation, collinearity via real ratios, equilateral criterion a+ωb+ω2c=0a + \omega b + \omega^2c = 0); point-to-line/plane distances; convexity (Helly, Carathéodory); continuous-motion IVT arguments.

Reminder — The Intermediate Value Theorem:f continuous on [a,b], f(a)<k<f(b)  f(c)=k for some c(a,b)f \text{ continuous on } [a,b],\ f(a)<k<f(b) \ \Longrightarrow\ f(c)=k \text{ for some } c\in(a,b)

Count lattice points by rows; prove existence by sweeping a parameter and invoking the IVT; reach for complex numbers when rotations appear.

Concept
The universal problem-solving loop

Read twice. Restate the goal in your own words. List the givens. Pick a representation. Try small cases. Look for symmetry, an invariant, or an extremal object. Compute, then CHECK against a second method or a sanity bound.

At AMC 10/12 and AIME level, the second method is what separates a 44-second guess from a 1212-second certainty: compute a probability two ways (complement and direct), a length via two theorems, a count via a recurrence and a formula. Casework must be organized by a stated criterion so nothing is double-counted; algebra must be checked by substituting back.