{
  "id": "youngs-inequality",
  "title": "Young's Inequality",
  "lang": "en",
  "source_lang": "en",
  "summary": "Young's inequality bounds a product by powers with conjugate exponents, with generalizations involving convex functions and convolution norms.",
  "category": "mathematics",
  "subcategory": "analysis",
  "type": "concept",
  "aliases": [
    "Young's inequality for products",
    "Young's product inequality"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/youngs-inequality/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/youngs-inequality/",
      "json": "https://aiwiki.page/zh/youngs-inequality.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Young's inequality is a fundamental inequality in mathematical analysis that bounds the product of two nonnegative numbers by a weighted sum of their powers. If \\(p,q>1\\) satisfy \\(1/p+1/q=1\\), then\n\n\\[ab\\leq \\frac{a^p}{p}+\\frac{b^q}{q} \\qquad (a,b\\geq 0).\\]\n\nNamed after the mathematician William Henry Young, it is widely used to estimate products, prove integral inequalities, and control terms in differential equations. The name also refers to a more general inequality involving inverse functions and to a related norm estimate for convolution."
    },
    {
      "heading": "Product form and equality",
      "text": "The exponents \\(p\\) and \\(q\\) are called conjugate exponents. Each determines the other:\n\n\\[q=\\frac{p}{p-1}, \\qquad p=\\frac{q}{q-1}.\\]\n\nYoung's inequality states that\n\n\\[ab\\leq \\frac{a^p}{p}+\\frac{b^q}{q},\\]\n\nwith equality if and only if\n\n\\[a^p=b^q.\\]\n\nEquivalent equality conditions are \\(b=a^{p-1}\\) and \\(a=b^{q-1}\\).\n\nThe simplest case is \\(p=q=2\\):\n\n\\[ab\\leq \\frac{a^2+b^2}{2}.\\]\n\nThis is equivalent to \\((a-b)^2\\geq0\\), with equality precisely when \\(a=b\\).\n\nFor example, with \\(p=3\\) and \\(q=3/2\\),\n\n\\[ab\\leq \\frac{a^3}{3}+\\frac{2b^{3/2}}{3}.\\]\n\nUnlike an estimate using squares alone, this version separates a product into terms with different growth rates.\n\nFor real or complex \\(a,b\\), the same statement applies to their absolute values:\n\n\\[|ab|\\leq \\frac{|a|^p}{p}+\\frac{|b|^q}{q}.\\]"
    },
    {
      "heading": "Proofs and connection with convexity",
      "text": "Proof by minimization\n\nFix \\(b\\geq0\\) and consider\n\n\\[F(a)=\\frac{a^p}{p}-ab+\\frac{b^q}{q}, \\qquad a\\geq0.\\]\n\nIts derivative is\n\n\\[F'(a)=a^{p-1}-b.\\]\n\nThe minimum occurs at\n\n\\[a=b^{1/(p-1)}=b^{q-1}.\\]\n\nAt this point, \\(a^p=b^q\\) and \\(ab=b^q\\), so\n\n\\[F(a)=b^q\\left(\\frac1p+\\frac1q-1\\right)=0.\\]\n\nThus \\(F(a)\\geq0\\), proving both the inequality and its equality condition.\n\nWeighted arithmetic–geometric mean\n\nThe weighted arithmetic–geometric mean inequality states that, for \\(x,y\\geq0\\) and \\(0<\\theta<1\\),\n\n\\[x^\\theta y^{1-\\theta} \\leq \\theta x+(1-\\theta)y.\\]\n\nTaking\n\n\\[x=a^p,\\qquad y=b^q,\\qquad \\theta=\\frac1p\\]\n\ngives Young's inequality, since \\(1-\\theta=1/q\\).\n\nConversely, substituting \\(a=x^{1/p}\\) and \\(b=y^{1/q}\\) recovers the weighted arithmetic–geometric mean inequality. The two statements are therefore equivalent under this change of variables.\n\nThis places Young's inequality within the theory of convex functions. The weighted arithmetic–geometric mean inequality follows from the concavity of the logarithm, or equivalently from Jensen's inequality."
    },
    {
      "heading": "Parameterized forms",
      "text": "A major practical advantage of Young's inequality is that its two terms can be rescaled. For every \\(\\delta>0\\),\n\n\\[ab\\leq \\frac{\\delta}{p}a^p+ \\frac{\\delta^{-q/p}}{q}b^q.\\]\n\nThis follows by applying the basic inequality to\n\n\\[\\delta^{1/p}a \\quad\\text{and}\\quad \\delta^{-1/p}b.\\]\n\nAn equivalent form, often called Young's inequality with epsilon, is\n\n\\[ab\\leq \\varepsilon a^p+C_{p,\\varepsilon}b^q,\\]\n\nwhere\n\n\\[C_{p,\\varepsilon} =\\frac1q(p\\varepsilon)^{-q/p}.\\]\n\nHere \\(\\varepsilon>0\\) can be chosen freely, but making it smaller increases the coefficient of \\(b^q\\).\n\nFor \\(p=q=2\\), this becomes\n\n\\[ab\\leq \\varepsilon a^2+\\frac{b^2}{4\\varepsilon}.\\]\n\nFor example, if nonnegative quantities \\(X,Y,A\\) satisfy\n\n\\[X^2\\leq A+XY,\\]\n\nthen choosing \\(\\varepsilon=1/2\\) gives\n\n\\[XY\\leq \\frac12X^2+\\frac12Y^2,\\]\n\nand hence\n\n\\[X^2\\leq 2A+Y^2.\\]\n\nThis procedure is known as absorption: a sufficiently small multiple of a term on the right-hand side is moved to the left-hand side. It is common in estimates for partial differential equations and in the calculus of variations."
    },
    {
      "heading": "Integral form and Hölder's inequality",
      "text": "Applying the scalar inequality pointwise to measurable functions \\(f,g\\) and integrating yields\n\n\\[\\int_X |fg|\\,d\\mu \\leq \\frac1p\\int_X |f|^p\\,d\\mu+ \\frac1q\\int_X |g|^q\\,d\\mu,\\]\n\nwhenever the right-hand side is finite. Here \\((X,\\mu)\\) is a measure space, and the integrals may be understood as Lebesgue integrals.\n\nA normalization argument gives Hölder's inequality:\n\n\\[\\int_X |fg|\\,d\\mu \\leq \\|f\\|_p\\|g\\|_q.\\]\n\nIndeed, if both norms are nonzero, apply Young's inequality to\n\n\\[\\frac{|f(x)|}{\\|f\\|_p} \\quad\\text{and}\\quad \\frac{|g(x)|}{\\|g\\|_q}.\\]\n\nAfter integration, the right-hand side is \\(1/p+1/q=1\\).\n\nThus Young's inequality provides a pointwise foundation for a central estimate in L^p spaces. Hölder's inequality in turn controls the integral pairing between conjugate spaces and underlies important results in functional analysis.\n\nFor \\(p=q=2\\), Hölder's inequality is the integral version of the Cauchy–Schwarz inequality."
    },
    {
      "heading": "Generalized Young inequality",
      "text": "The power-function inequality is a special case of a broader result involving an increasing function and its inverse.\n\nLet\n\n\\[f:[0,\\infty)\\longrightarrow[0,\\infty)\\]\n\nbe continuous, strictly increasing, satisfy \\(f(0)=0\\), and tend to infinity as its argument tends to infinity. Its inverse function is then defined throughout \\([0,\\infty)\\). For \\(a,b\\geq0\\),\n\n\\[ab\\leq \\int_0^a f(t)\\,dt+ \\int_0^b f^{-1}(s)\\,ds,\\]\n\nwith equality if and only if \\(b=f(a)\\).\n\nGeometrically, the two integrals describe areas associated with the graph of \\(f\\) and its inverse. Their sum equals the rectangle area \\(ab\\) when \\((a,b)\\) lies on the graph; otherwise, the sum includes an additional nonnegative area.\n\nTaking\n\n\\[f(t)=t^{p-1}\\]\n\ngives\n\n\\[f^{-1}(s)=s^{q-1},\\]\n\nand the two integrals become \\(a^p/p\\) and \\(b^q/q\\). This recovers the familiar product inequality."
    },
    {
      "heading": "Convex conjugates and the Fenchel–Young inequality",
      "text": "A broader formulation uses the convex conjugate. For a proper extended-real-valued function \\(\\Phi\\) on \\(\\mathbb R\\), its conjugate is\n\n\\[\\Phi^*(y)=\\sup_x\\{xy-\\Phi(x)\\}.\\]\n\nThe definition immediately implies the Fenchel–Young inequality\n\n\\[xy\\leq \\Phi(x)+\\Phi^*(y).\\]\n\nFor\n\n\\[\\Phi(x)=\\frac{|x|^p}{p},\\]\n\nthe conjugate is\n\n\\[\\Phi^*(y)=\\frac{|y|^q}{q}.\\]\n\nConsequently, Fenchel–Young includes the power form of Young's inequality.\n\nFor a proper lower-semicontinuous convex function, equality holds precisely when \\(y\\) is a subgradient of \\(\\Phi\\) at \\(x\\). At a point where \\(\\Phi\\) is differentiable, this condition becomes\n\n\\[y=\\Phi'(x).\\]\n\nIn a finite-dimensional vector space, the scalar product \\(xy\\) is replaced by the pairing \\(\\langle x,y\\rangle\\):\n\n\\[\\langle x,y\\rangle\\leq \\Phi(x)+\\Phi^*(y).\\]\n\nThis formulation is important in convex optimization and duality. It also allows product estimates based on growth functions other than powers."
    },
    {
      "heading": "Young's convolution inequality",
      "text": "A distinct but related result bearing Young's name concerns convolution. On \\(\\mathbb R^n\\), convolution is defined by\n\n\\[(f*g)(x)=\\int_{\\mathbb R^n}f(x-y)g(y)\\,dy.\\]\n\nIf \\(1\\leq p,q,r\\leq\\infty\\) satisfy\n\n\\[1+\\frac1r=\\frac1p+\\frac1q,\\]\n\nwith \\(1/\\infty=0\\), then\n\n\\[\\|f*g\\|_r\\leq \\|f\\|_p\\|g\\|_q.\\]\n\nImportant cases include\n\n\\[\\|f*g\\|_p\\leq \\|f\\|_1\\|g\\|_p\\]\n\nand, for conjugate exponents,\n\n\\[\\|f*g\\|_\\infty\\leq \\|f\\|_p\\|g\\|_q.\\]\n\nThe convolution inequality controls how convolution changes integrability. It is used in harmonic analysis, in smoothing estimates, and in the study of equations such as the heat equation. It should be distinguished from the scalar product inequality: its conclusion concerns a function norm, and its exponent relation involves three exponents rather than two."
    },
    {
      "heading": "Scope and limitations",
      "text": "The standard power form requires \\(p,q>1\\). The condition \\(1/p+1/q=1\\) is essential: the exponents are chosen so that the two power functions are convex conjugates. Arbitrary exponents do not produce the same global estimate with coefficients \\(1/p\\) and \\(1/q\\).\n\nAlthough Hölder's and convolution inequalities include endpoint cases involving \\(1\\) and \\(\\infty\\), the scalar expression \\(b^q/q\\) cannot be interpreted at \\(q=\\infty\\) by ordinary substitution. A convex-conjugate formulation handles such endpoints through extended-real-valued functions.\n\nYoung's inequality is especially useful for separating mixed terms into independently controlled quantities. That separation can lose information, however: equality occurs only when the two inputs satisfy the stated power relation. In parameterized estimates, the freedom to reduce one coefficient is always balanced by growth in the other coefficient."
    }
  ],
  "links": [
    "mathematical-analysis",
    "absolute-value",
    "derivative",
    "arithmetic-geometric-mean-inequality",
    "convex-function",
    "jensens-inequality",
    "partial-differential-equation",
    "calculus-of-variations",
    "lebesgue-integral",
    "holders-inequality",
    "lp-space",
    "functional-analysis",
    "cauchy-schwarz-inequality",
    "inverse-function",
    "convex-optimization",
    "lagrangian-duality",
    "convolution",
    "harmonic-analysis",
    "heat-equation"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}