AM-GM, power means, rearrangement, Chebyshev, Cauchy–Schwarz, Hölder, Minkowski, Jensen and the tangent line, extrema of convex functions at the endpoints, Karamata, smoothing.
AM-GM.nx1+⋯+xn≥nx1⋯xn, with equality only when x1=⋯=xn. With weights wi≥0, ∑wi=1: ∑wixi≥∏xiwi.
Power means. For xi>0 and p=0 put Mp=(n1∑xip)1/p and M0=nx1⋯xn. The map p↦Mp is increasing. In particular minxi≤M−1≤M0≤M1≤M2≤maxxi (harmonic ≤ geometric ≤ arithmetic ≤ quadratic).
Rearrangement. If a1≤⋯≤an and b1≤⋯≤bn (signs do not matter here), then for every permutation σ
i∑aibn+1−i≤i∑aibσ(i)≤i∑aibi.
Chebyshev. If (ai), (bi) are similarly ordered, n∑aibi≥(∑ai)(∑bi). For integrals: if f,g are both increasing on [a,b], (b−a)∫abfg≥∫abf∫abg.
Cauchy–Schwarz.(∑aibi)2≤∑ai2∑bi2 and (∫fg)2≤∫f2∫g2 (for real ai,bi and functions, with no sign restriction).
Hölder. For p,q>1 with p1+q1=1: ∑aibi≤(∑aip)1/p(∑biq)1/q, and ∫∣fg∣≤∥f∥p∥g∥q.
Minkowski. For p≥1: (∑(ai+bi)p)1/p≤(∑aip)1/p+(∑bip)1/p, and ∥f+g∥p≤∥f∥p+∥g∥p.
Jensen. If φ is convex on an interval I, xi∈I and wi≥0 with ∑wi=1, then φ(∑wixi)≤∑wiφ(xi). For integrals: if f:[0,1]→I is continuous, φ(∫01f)≤∫01φ(f). For concave φ the direction is reversed.
The one-line inequalities.ex≥1+x, lnx≤x−1, sinx≤x for x≥0, 2ab≤a2+b2. They all say that a convex (concave) function lies above (below) its tangent line.
The basic lemmas, with short proofs
AM-GM from et≥1+t. Let A=n1∑xi>0. From et≥1+t with t=Axi−1 we get Axi≤exi/A−1. Multiplying,
Anx1⋯xn≤e∑(xi/A−1)=e0=1.
Since et=1+t only for t=0, equality holds only if xi=A for every i.
Rearrangement by swapping. If a permutation has i<j with σ(i)>σ(j), swapping bσ(i) and bσ(j) changes the sum by
Each swap reduces the number of inversions, so after finitely many steps we reach the identity without decreasing the sum. The lower bound follows by applying the same to −bn+1−i.
Chebyshev by a double sum. Since (ai), (bi) are similarly ordered, every term of ∑i,j(ai−aj)(bi−bj) is ≥0. Expanding, the sum equals 2n∑aibi−2∑ai∑bi. For integrals, the same with ∬(f(x)−f(y))(g(x)−g(y))dxdy≥0.
Cauchy–Schwarz by the discriminant. The quantity ∑(ait+bi)2=t2∑ai2+2t∑aibi+∑bi2 is ≥0 for every t, so the discriminant is ≤0. For integrals see also the integrals sheet.
Hölder from Young. For u,v≥0 we have uv≤pup+qvq (weighted AM-GM with weights p1,q1 on up,vq). If A=(∑aip)1/p and B=(∑biq)1/q are positive, put u=ai/A, v=bi/B and sum: AB∑aibi≤p1+q1=1.
Minkowski from Hölder. For p>1 write ∑(ai+bi)p=∑ai(ai+bi)p−1+∑bi(ai+bi)p−1 and apply Hölder to each sum. Since (p−1)q=p, you get S≤(∥a∥p+∥b∥p)S1/q with S=∑(ai+bi)p, and you divide by S1/q.
Jensen from a supporting line. A convex φ has a supporting line at every interior point m of I: there is k with φ(x)≥φ(m)+k(x−m) for every x∈I (if φ is differentiable, k=φ′(m)). With m=∑wixi, multiply by wi and sum: the terms k(xi−m) add up to 0. For integrals put x=f(t) and integrate.
The tangent line trick. If you want ∑φ(xi)≥nφ(m) under ∑xi=nm, but φ is not convex everywhere, it is enough to show φ(x)≥φ(m)+φ′(m)(x−m)only for the x in the domain. Usually the difference is a fraction with a factor (x−m)2. Summing, the linear term disappears.
Convex function: maximum at the endpoints. Every x∈[a,b] can be written x=(1−λ)a+λb with λ∈[0,1], so
φ(x)≤(1−λ)φ(a)+λφ(b)≤max{φ(a),φ(b)}.
Likewise a concave function attains its minimum at an endpoint. In several variables: if the expression is convex (or linear) in each variable separately on a box ∏[ai,bi], the maximum is attained at a vertex of the box. Keep the others fixed and push one variable at a time to an endpoint.
Karamata, briefly. We say that x=(x1≥⋯≥xn)majorisesy=(y1≥⋯≥yn) if x1+⋯+xk≥y1+⋯+yk for every k, with equality for k=n. Then for convex φ we have ∑φ(xi)≥∑φ(yi). Example: (2,0) majorises (1,1), and two-point Jensen is a special case. The practical consequence: with a fixed sum and xi∈[a,b], ∑φ(xi) for convex φ is maximised when as many xi as possible sit at the endpoints a, b (at most one in between).
Smoothing. If replacing two variables xi,xj by their mean 2xi+xj keeps the constraint and does not increase the expression, the minimum is at the point with all variables equal. Mind the logic: first make sure that a minimum point exists (continuity on a compact set), then show that if they are not all equal, a smoothing step gives a strictly smaller value.
Typical tricks
Find the equality case first. It tells you how to split the terms in AM-GM. E.g. for x>0, write x2+x16 as x2+x8+x8, so that the product is constant and equality (x=2) is attainable.
Homogenise and normalise. With a constraint such as abc=1 or a+b+c=3 you make both sides of the same degree, or conversely, in a homogeneous inequality you choose the scale that suits you.
Squares (SOS). Write the difference as a sum of squares, e.g. 3(a2+b2+c2)−(a+b+c)2=(a−b)2+(b−c)2+(c−a)2.
Cauchy–Schwarz with the constant 1.(∫abf)2≤(b−a)∫abf2. For sums, the Engel form: ∑viui2≥∑vi(∑ui)2 for vi>0.
Convexity without a second derivative. Sums of convex functions, a convex function composed with a linear one, and the max of convex functions are convex. t is concave, tlnt is convex.
One-variable minimisation: if you end with αc+cβ with α,β,c>0, AM-GM gives ≥2αβ with no derivative.
How it appears in the IMC
The IMC rarely asks for a "pure" three-variable inequality. Usually the inequality is one step inside an analysis or linear algebra problem:
IMC 2026, Day 2, Problem 7: an area is bounded below by an integral depending on one parameter, and the minimisation is two-term AM-GM.
IMC 2026, Day 2, Problem 8: the sum of the ∣λi∣ is written as a strictly concave function of one parameter on a closed interval. The minimum is at an endpoint, and it remains to compare two values.