{
  "aliases": null,
  "backlinks": null,
  "body": "\n## Statement\n\nLet $X$ be a real [[normed space|03J5RAKB30]] with [[continuous dual space|4A82S5FVRT]] $X^*$.\nLet $f \\colon X \\to [-\\infty, +\\infty]$ be a function with values in the [[extended real line|0GSZ7B0CGR]], with [[Fenchel conjugate|T7V2VPY61S]] $f^* \\colon X^* \\to [-\\infty, +\\infty]$.\nDefine the biconjugate $f^{**} \\colon X \\to [-\\infty, +\\infty]$ by $f^{**}(x) = \\sup_{y \\in X^*} \\big(y(x) - f^*(y)\\big)$.\nThen $f^{**} = f$ if and only if one of the following holds:\n\n1. $f$ is proper, [[convex|07Q7CNM5WF]] and lower [[semicontinuous|RN4G8SGP6P]], where proper means that $f$ never takes the value $-\\infty$ and is not identically $+\\infty$;\n2. $f$ is identically $+\\infty$;\n3. $f$ is identically $-\\infty$.\n\n## Remarks\n\nFor every $f$, the function $f^{**}$ is a supremum of continuous affine functions, so it is convex and lower semicontinuous, and the Fenchel–Young inequality gives $f^{**} \\le f$.\nIf $f$ takes the value $-\\infty$ somewhere, then $f^*$ is identically $+\\infty$ and $f^{**}$ is identically $-\\infty$, so $f^{**} = f$ forces case 3.\nOtherwise $f^{**} = f$ forces $f$ to be convex and lower semicontinuous, which is case 1 or case 2.\nFor the \"if\" direction, take $x_0$ and a real $r \u003c f(x_0)$.\nThe point $(x_0, r)$ lies outside the closed convex [[epigraph|8BJ5M76FSX]] of $f$, and the Hahn–Banach separation theorem gives a continuous affine minorant of $f$ with value above $r$ at $x_0$.\nEvery such minorant is below $f^{**}$, so $f^{**}(x_0) \\ge f(x_0)$.\n\nExample: $f(x) = \\abs{x}$ on $\\R$ has $f^* = 0$ on $[-1, 1]$ and $+\\infty$ elsewhere, and $f^{**}(x) = \\sup_{\\abs{y} \\le 1} xy = \\abs{x}$.\nLower semicontinuity is necessary: for $f = 0$ on $(0, 1)$ and $f = +\\infty$ elsewhere on $\\R$, the function is convex and proper, but $f^{**}$ is $0$ on the closed interval $[0, 1]$.\nConvexity is necessary: for $f(x) = \\min\\big((x - 1)^2, (x + 1)^2\\big)$, the biconjugate is $0$ on $[-1, 1]$ and agrees with $f$ outside.\n\n## Explanations\n\n### L1 human\n\nTaking the conjugate twice returns the function you started with, as long as the function is convex and has no gaps in its graph from below.\nIn general, the double conjugate is the largest convex, lower semicontinuous function that stays below $f$.\nSo a convex closed function is completely described by the family of lines that lie below it.\n\n### L2 human\n\nThe theorem is the basis of duality in convex optimization.\nA primal problem and its dual are related through conjugates, and the theorem says that dualizing twice gives back the primal problem.\nIt also shows that $f \\mapsto f^*$ is a bijection between proper convex lower semicontinuous functions on $X$ and on the dual side, when $X$ is reflexive.\nIt extends the involution property of the classical Legendre transformation to nonsmooth functions.\nThe same statement holds in every Hausdorff locally convex space.\n",
  "commit": "ff8a40749dc16545e3d2fc637f6b0196e81d0bfa",
  "front": {
    "about": [
      "T7V2VPY61S"
    ],
    "kind": "statement",
    "msc": [
      "52A41",
      "46N10",
      "26B25"
    ],
    "name": "Fenchel–Moreau theorem",
    "role": "theorem",
    "same_as": {
      "wikidata": "Q5443005",
      "wikipedia": "Fenchel–Moreau theorem"
    },
    "sources": [
      {
        "url": "https://en.wikipedia.org/wiki/Fenchel–Moreau_theorem"
      }
    ],
    "tag": "XZFYB2PBPD",
    "uses": [
      "T7V2VPY61S",
      "07Q7CNM5WF",
      "RN4G8SGP6P",
      "03J5RAKB30",
      "4A82S5FVRT",
      "0GSZ7B0CGR",
      "8BJ5M76FSX"
    ]
  },
  "kind": "statement",
  "link": "https://tamnd.github.io/math2/node/XZFYB2PBPD/",
  "links": [
    {
      "field": "about",
      "target": "T7V2VPY61S"
    },
    {
      "field": "uses",
      "target": "T7V2VPY61S"
    },
    {
      "field": "uses",
      "target": "07Q7CNM5WF"
    },
    {
      "field": "uses",
      "target": "RN4G8SGP6P"
    },
    {
      "field": "uses",
      "target": "03J5RAKB30"
    },
    {
      "field": "uses",
      "target": "4A82S5FVRT"
    },
    {
      "field": "uses",
      "target": "0GSZ7B0CGR"
    },
    {
      "field": "uses",
      "target": "8BJ5M76FSX"
    }
  ],
  "markdown": "https://tamnd.github.io/math2/node/XZFYB2PBPD.md",
  "msc": [
    "52A41",
    "46N10",
    "26B25"
  ],
  "name": "Fenchel–Moreau theorem",
  "prov": null,
  "role": "theorem",
  "source": "https://github.com/tamnd/math2/blob/main/nodes/XZF/XZFYB2PBPD.md",
  "status": null,
  "tag": "XZFYB2PBPD",
  "uri": "m2://XZFYB2PBPD",
  "url": "https://tamnd.github.io/math2/wiki/fenchel-moreau-theorem/",
  "writing": {
    "tag": "XZFYB2PBPD",
    "kind": "statement",
    "level": "W0",
    "next": [
      {
        "what": "an inventory row",
        "note": "the row of the inventory that the node accounts for",
        "have": 0,
        "want": 1
      },
      {
        "what": "a source with a locator",
        "have": 0,
        "want": 1
      }
    ]
  }
}
