{
  "id": "total-derivative",
  "title": "Total Derivative",
  "lang": "en",
  "source_lang": "en",
  "summary": "The total derivative is the linear map giving a function’s first-order change with respect to all its variables, or its rate of change along a specified dependency.",
  "category": "mathematics",
  "subcategory": "analysis",
  "type": "concept",
  "aliases": [
    "Total differential",
    "Fréchet derivative"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/total-derivative/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/total-derivative/",
      "json": "https://aiwiki.page/zh/total-derivative.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The total derivative of a function of several variables is the linear map that describes its first-order change when all input variables may change together. It generalizes the ordinary derivative of a function of one variable. The term also denotes the derivative of a composite quantity along a specified path or dependency, accounting for both explicit and indirect dependence on the varying parameter. These meanings are connected by the chain rule, but they describe different mathematical objects: a linear map at a point and a rate of change along a path."
    },
    {
      "heading": "Definition as a linear approximation",
      "text": "Let \\(U\\) be an open subset of Euclidean space \\(\\mathbb{R}^n\\), and let\n\n\\[f:U\\to\\mathbb{R}^m.\\]\n\nThe function is totally differentiable at \\(a\\in U\\) if there is a linear map\n\n\\[Df(a):\\mathbb{R}^n\\to\\mathbb{R}^m\\]\n\nsuch that\n\n\\[\\lim_{h\\to0} \\frac{\\|f(a+h)-f(a)-Df(a)h\\|}{\\|h\\|}=0.\\]\n\nHere \\(\\|\\cdot\\|\\) denotes a norm; in finite-dimensional spaces, the choice among equivalent norms does not affect differentiability.\n\nEquivalently,\n\n\\[f(a+h)=f(a)+Df(a)h+r(h), \\qquad \\frac{\\|r(h)\\|}{\\|h\\|}\\to0.\\]\n\nThe remainder is therefore negligible compared with the size of the input displacement. The expression\n\n\\[h\\longmapsto f(a)+Df(a)h\\]\n\nis the first-order affine approximation to \\(f\\) near \\(a\\).\n\nThe derivative \\(Df(a)\\), if it exists, is unique. Other common notations include \\(df_a\\), \\(df(a)\\), and \\(f'(a)\\). In finite-dimensional analysis this definition is also called Fréchet differentiability; its extension to normed spaces is the Fréchet derivative.\n\nThe requirement concerns arbitrary sufficiently small displacements \\(h\\), not merely movement along the coordinate axes. This is the main distinction between total differentiability and the existence of partial derivatives."
    },
    {
      "heading": "Jacobian matrix, gradient, and differential",
      "text": "If \\(f=(f_1,\\ldots,f_m)\\) is totally differentiable, its derivative is represented in standard coordinates by the Jacobian matrix:\n\n\\[Df(a)=J_f(a)= \\begin{pmatrix} \\frac{\\partial f_1}{\\partial x_1}(a)&\\cdots& \\frac{\\partial f_1}{\\partial x_n}(a)\\\\ \\vdots&\\ddots&\\vdots\\\\ \\frac{\\partial f_m}{\\partial x_1}(a)&\\cdots& \\frac{\\partial f_m}{\\partial x_n}(a) \\end{pmatrix}.\\]\n\nThus the first-order output change is\n\n\\[\\Delta f\\approx J_f(a)\\,\\Delta x.\\]\n\nFor a scalar-valued function \\(f:U\\to\\mathbb{R}\\),\n\n\\[Df(a)h = \\sum_{i=1}^{n} \\frac{\\partial f}{\\partial x_i}(a)h_i = \\nabla f(a)\\cdot h,\\]\n\nwhere \\(\\nabla f(a)\\) is the gradient and the final expression uses the Euclidean inner product. The gradient is a vector, while the derivative is a linear functional; the inner product provides the identification between them.\n\nThe same relation is often written as the total differential:\n\n\\[df=\\sum_{i=1}^{n}\\frac{\\partial f}{\\partial x_i}\\,dx_i.\\]\n\nPrecisely, at \\(a\\), each \\(dx_i\\) is the coordinate linear functional \\(h\\mapsto h_i\\), and \\(df_a\\) is their linear combination. Applied to a displacement \\(h\\), this is an exact identity between linear maps. By contrast,\n\n\\[f(a+h)-f(a)\\approx df_a(h)\\]\n\nis generally only an approximation to a finite change.\n\nFor example, if\n\n\\[f(x,y)=x^2y+\\sin y,\\]\n\nthen\n\n\\[df=2xy\\,dx+(x^2+\\cos y)\\,dy.\\]\n\nConsequently, for small increments \\(\\Delta x,\\Delta y\\),\n\n\\[\\Delta f \\approx 2xy\\,\\Delta x+(x^2+\\cos y)\\,\\Delta y.\\]"
    },
    {
      "heading": "Relation to partial and directional derivatives",
      "text": "A partial derivative measures variation in one coordinate while the remaining coordinates are held fixed. A directional derivative measures variation along a specified direction \\(v\\):\n\n\\[D_vf(a)= \\lim_{t\\to0}\\frac{f(a+tv)-f(a)}{t},\\]\n\nwhen the limit exists.\n\nIf \\(f\\) is totally differentiable at \\(a\\), all these directional derivatives exist and\n\n\\[D_vf(a)=Df(a)v.\\]\n\nIn particular, the directional derivative depends linearly on \\(v\\). Total differentiability also implies that \\(f\\) is continuous at \\(a\\).\n\nThe converses do not hold in general. Even continuity together with the existence of every directional derivative is insufficient.\n\nFor example, define\n\n\\[f(x,y)= \\begin{cases} \\dfrac{x^3}{x^2+y^2},&(x,y)\\ne(0,0),\\\\[6pt] 0,&(x,y)=(0,0). \\end{cases}\\]\n\nSince \\(|f(x,y)|\\le |x|\\), this function is continuous at the origin. For every nonzero direction \\(v=(a,b)\\),\n\n\\[D_vf(0,0)=\\frac{a^3}{a^2+b^2}.\\]\n\nHowever, this expression is not linear in \\(v\\). The coordinate partial derivatives are\n\n\\[f_x(0,0)=1,\\qquad f_y(0,0)=0,\\]\n\nso a total derivative, if it existed, would have to send \\((a,b)\\) to \\(a\\). In the direction \\((1,1)\\), the actual directional derivative is \\(1/2\\), not \\(1\\). Therefore the function is not totally differentiable at the origin.\n\nA standard sufficient condition is that all first partial derivatives exist in a neighborhood of \\(a\\) and are continuous at \\(a\\). Under this condition, \\(f\\) is totally differentiable at \\(a\\). Continuity of the partial derivatives is sufficient, but not necessary."
    },
    {
      "heading": "Total derivative along a path",
      "text": "Suppose the inputs depend on a parameter \\(t\\):\n\n\\[x(t)=\\bigl(x_1(t),\\ldots,x_n(t)\\bigr).\\]\n\nIf \\(x\\) is differentiable at \\(t\\) and \\(f\\) is totally differentiable at \\(x(t)\\), the chain rule gives\n\n\\[\\frac{d}{dt}f(x(t)) = Df(x(t))\\,x'(t).\\]\n\nFor scalar-valued \\(f\\),\n\n\\[\\frac{df}{dt} = \\sum_{i=1}^{n} \\frac{\\partial f}{\\partial x_i} \\frac{dx_i}{dt}.\\]\n\nThe partial derivatives are evaluated at the current point \\(x(t)\\).\n\nIf a quantity also depends explicitly on \\(t\\), write it as \\(F(t,x(t))\\). Then\n\n\\[\\frac{d}{dt}F(t,x(t)) = \\frac{\\partial F}{\\partial t} + \\sum_{i=1}^{n} \\frac{\\partial F}{\\partial x_i}\\frac{dx_i}{dt}.\\]\n\nThe first term describes explicit dependence on \\(t\\); the remaining terms describe dependence transmitted through the changing inputs.\n\nFor example, let\n\n\\[F(t,x,y)=tx^2+y,\\qquad x(t)=t^2,\\qquad y(t)=\\sin t.\\]\n\nThen\n\n\\[\\frac{dF}{dt} = x^2+2tx(2t)+\\cos t = 5t^4+\\cos t.\\]\n\nHolding \\(x\\) and \\(y\\) fixed would instead give only\n\n\\[\\frac{\\partial F}{\\partial t}=x^2.\\]\n\nThus a total derivative along a path cannot be determined without specifying which quantities vary and how they depend on the parameter."
    },
    {
      "heading": "Composition and higher derivatives",
      "text": "For differentiable maps\n\n\\[f:\\mathbb{R}^n\\to\\mathbb{R}^m, \\qquad g:\\mathbb{R}^m\\to\\mathbb{R}^p,\\]\n\nthe derivative of their composition is\n\n\\[D(g\\circ f)(a)=Dg(f(a))\\circ Df(a).\\]\n\nIn coordinates, composition of these linear maps becomes matrix multiplication:\n\n\\[J_{g\\circ f}(a)=J_g(f(a))J_f(a).\\]\n\nThis is the general chain rule underlying the scalar path formulas.\n\nWhen the derivative map is itself differentiable, the second total derivative \\(D^2f(a)\\) is a bilinear map. For a twice continuously differentiable scalar-valued function, it is represented by the Hessian matrix, and the second-order Taylor expansion is\n\n\\[f(a+h) = f(a)+Df(a)h +\\frac12h^{\\mathsf T}H_f(a)h +o(\\|h\\|^2).\\]\n\nFor a twice differentiable path \\(x(t)\\) and a sufficiently smooth scalar-valued \\(f\\),\n\n\\[\\frac{d^2}{dt^2}f(x(t)) = x'(t)^{\\mathsf T}H_f(x(t))x'(t) + \\nabla f(x(t))\\cdot x''(t).\\]\n\nThe second derivative along a path therefore depends both on the curvature of the function and on the acceleration of the path."
    },
    {
      "heading": "Applications",
      "text": "Sensitivity and computation\n\nIn sensitivity analysis, the total derivative describes how simultaneous input changes affect an output. Its Jacobian collects all first-order input–output sensitivities. This is useful in numerical models, parameter estimation, and optimization.\n\nAutomatic differentiation applies the chain rule systematically to computational operations. Forward-mode differentiation can compute a Jacobian–vector product \\(Df(a)v\\), while reverse-mode differentiation can compute the gradient of a scalar output without explicitly constructing every intermediate Jacobian.\n\nThese are local, first-order descriptions. For finite perturbations, neglected nonlinear terms can be important.\n\nDynamics and transport\n\nIf a scalar field \\(u(t,x)\\) is evaluated along a trajectory satisfying\n\n\\[x'(t)=v(t,x(t)),\\]\n\nthen\n\n\\[\\frac{d}{dt}u(t,x(t)) = \\frac{\\partial u}{\\partial t} + v\\cdot\\nabla u.\\]\n\nIn continuum mechanics, this rate along a moving particle is called the material derivative. It distinguishes change observed while following the particle from change observed at a fixed spatial location.\n\nIn the calculus of variations, expressions such as\n\n\\[\\frac{d}{dt}\\left(\\frac{\\partial L}{\\partial\\dot q_i}\\right)\\]\n\nin the Euler–Lagrange equations are total derivatives along a trajectory. They account for all dependencies of the expression on time, position, and velocity."
    },
    {
      "heading": "Extensions beyond Euclidean spaces",
      "text": "For normed vector spaces \\(X\\) and \\(Y\\), Fréchet differentiability uses the same remainder condition, with the derivative required to be a bounded linear map \\(X\\to Y\\). In infinite-dimensional spaces, boundedness must be stated explicitly; in finite dimensions, every linear map is bounded.\n\nOn a smooth manifold, the derivative of a smooth map \\(f:M\\to N\\) at \\(p\\) is a linear map between tangent spaces:\n\n\\[df_p:T_pM\\to T_{f(p)}N.\\]\n\nCoordinate charts represent this map by a Jacobian, but the map itself is independent of the chosen coordinates. For a real-valued smooth function, \\(df\\) is a differential one-form, assigning a linear functional to each tangent space."
    }
  ],
  "links": [
    "mathematical-function",
    "derivative",
    "open-set",
    "euclidean-space",
    "linear-map",
    "norm-mathematics",
    "jacobian-matrix",
    "gradient",
    "partial-derivative",
    "directional-derivative",
    "continuous-function",
    "chain-rule",
    "function-composition",
    "hessian-matrix",
    "taylor-series",
    "sensitivity-analysis",
    "mathematical-optimization",
    "automatic-differentiation",
    "calculus-of-variations",
    "euler-lagrange-equations",
    "normed-vector-space",
    "manifold",
    "tangent-space",
    "differential-form"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}