{
  "aliases": null,
  "backlinks": [
    {
      "field": "defines",
      "target": "ECDJ8DD9RV"
    },
    {
      "field": "uses",
      "target": "WAE4ZGJXJ7"
    }
  ],
  "body": "\n## Summary\n\nThe Radon–Nikodym derivative $d\\nu / d\\mu$ is the density of one measure $\\nu$ with respect to another measure $\\mu$.\nIt exists when both measures are σ-finite and $\\nu$ is absolutely continuous with respect to $\\mu$, by the [[Radon–Nikodym theorem|1ZSZS5G1Q4]].\nIt turns integrals against $\\nu$ into integrals against $\\mu$.\nProbability densities, likelihood ratios and conditional expectations are all Radon–Nikodym derivatives.\n\n## Explanations\n\n### L1 human\n\nThink of $\\mu$ as a way to measure size and $\\nu$ as a way to measure mass.\nThe derivative $d\\nu / d\\mu$ at a point is the mass per unit size near that point.\nFor the standard normal distribution $\\nu$ and Lebesgue measure $\\mu$ on the line, the derivative is the bell curve $e^{-x^2/2}/\\sqrt{2\\pi}$.\nThe name comes from the rule $d\\nu = \\frac{d\\nu}{d\\mu} \\, d\\mu$, which looks like the chain rule of calculus.\n",
  "commit": "547b6eddb8f2feb8d579f2d633d22a27b7e132fc",
  "front": {
    "kind": "concept",
    "msc": [
      "28A15"
    ],
    "name": "Radon–Nikodym derivative",
    "sources": [
      {
        "url": "https://en.wikipedia.org/wiki/Radon–Nikodym_theorem"
      }
    ],
    "tag": "1NTKG64F7N"
  },
  "kind": "concept",
  "link": "https://math2.tamnd.com/node/1NTKG64F7N/",
  "links": null,
  "markdown": "https://math2.tamnd.com/node/1NTKG64F7N.md",
  "msc": [
    "28A15"
  ],
  "name": "Radon–Nikodym derivative",
  "prov": null,
  "role": "",
  "source": "https://github.com/tamnd/math2/blob/main/nodes/1NT/1NTKG64F7N.md",
  "status": null,
  "tag": "1NTKG64F7N",
  "uri": "m2://1NTKG64F7N",
  "url": "https://math2.tamnd.com/wiki/radon-nikodym-derivative/",
  "writing": {
    "tag": "1NTKG64F7N",
    "kind": "concept",
    "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
      }
    ]
  }
}
