{
  "id": "initial-value-problem",
  "title": "Initial Value Problem",
  "lang": "en",
  "source_lang": "en",
  "summary": "An initial value problem specifies a differential equation together with the state of its solution at a given starting point, determining how that state evolves.",
  "category": "mathematics",
  "subcategory": "analysis",
  "type": "concept",
  "aliases": [
    "IVP",
    "initial-value problem"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/initial-value-problem/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/initial-value-problem/",
      "json": "https://aiwiki.page/zh/initial-value-problem.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "An initial value problem is a differential equation together with conditions specifying the value of an unknown function, and sometimes its derivatives, at a single initial point. Its purpose is to determine a solution consistent with both the equation and those initial data. Initial value problems are fundamental to the mathematical description of evolution: an equation describes how a system changes, while the initial conditions describe its starting state. The central questions are whether a solution exists, whether it is unique, how long it remains defined, and how sensitively it depends on the data."
    },
    {
      "heading": "Mathematical formulation",
      "text": "For a first-order ordinary differential equation, the standard formulation is\n\n\\[y'(t)=f(t,y(t)),\\qquad y(t_0)=y_0.\\]\n\nHere \\(t\\) is the independent variable, \\(t_0\\) is the initial point, and \\(y_0\\) is the prescribed initial value. The unknown \\(y\\) may be scalar-valued or take values in Euclidean space \\(\\mathbb{R}^n\\), in which case the equation represents a system of \\(n\\) coupled equations.\n\nA classical solution is a differentiable function \\(y:I\\to\\mathbb{R}^n\\), defined on an interval \\(I\\) containing \\(t_0\\), such that\n\n\\[y'(t)=f(t,y(t))\\]\n\nthroughout the interval and \\(y(t_0)=y_0\\). The points \\((t,y(t))\\) must remain in the domain where \\(f\\) is defined.\n\nAlthough \\(t\\) commonly denotes time, it may instead represent distance or another independent variable. “Initial” identifies where the data are prescribed; it does not require that solutions be considered only for \\(t\\geq t_0\\). When the equation permits it, a solution can be continued both forward and backward from the initial point.\n\nHigher-order equations\n\nAn explicit equation of order \\(m\\),\n\n\\[y^{(m)}=F(t,y,y',\\ldots,y^{(m-1)}),\\]\n\nis normally supplemented by \\(m\\) initial conditions:\n\n\\[y(t_0)=a_0,\\quad y'(t_0)=a_1,\\quad \\ldots,\\quad y^{(m-1)}(t_0)=a_{m-1}.\\]\n\nIntroducing the state variables\n\n\\[x_1=y,\\quad x_2=y',\\quad\\ldots,\\quad x_m=y^{(m-1)}\\]\n\nconverts it into a first-order system. Thus the theory of first-order systems also covers explicit higher-order equations.\n\nFor example, the equation of an ideal harmonic oscillator,\n\n\\[x''+\\omega^2x=0,\\]\n\nrequires an initial position \\(x(t_0)=x_0\\) and velocity \\(x'(t_0)=v_0\\). For \\(\\omega>0\\), its solution is\n\n\\[x(t)=x_0\\cos\\bigl(\\omega(t-t_0)\\bigr) +\\frac{v_0}{\\omega}\\sin\\bigl(\\omega(t-t_0)\\bigr).\\]\n\nPrescribing position alone leaves the initial velocity undetermined and therefore does not specify a unique motion."
    },
    {
      "heading": "Initial conditions and boundary conditions",
      "text": "An initial value problem prescribes data at one value of the independent variable. A boundary value problem typically prescribes conditions at different points or on the boundary of a spatial domain.\n\nFor example,\n\n\\[y''+y=0,\\qquad y(0)=0,\\quad y'(0)=1\\]\n\nis an initial value problem with the unique solution \\(y(t)=\\sin t\\). By contrast,\n\n\\[y''+y=0,\\qquad y(0)=0,\\quad y(\\pi)=0\\]\n\nis a boundary value problem with infinitely many solutions \\(y(t)=C\\sin t\\).\n\nThis difference affects both analysis and computation. Initial value problems naturally support step-by-step integration from a starting point. Boundary value problems generally require finding a solution that satisfies conditions imposed at separated locations."
    },
    {
      "heading": "Existence and uniqueness",
      "text": "Initial data do not automatically select exactly one solution. Existence and uniqueness depend on the regularity of the equation and on the point where the data are prescribed.\n\nLocal existence\n\nPeano’s existence theorem states, in a standard finite-dimensional formulation, that if \\(f(t,y)\\) is continuous in an open neighborhood of \\((t_0,y_0)\\), then at least one local solution exists through that point.\n\nContinuity alone does not guarantee uniqueness. Consider\n\n\\[y'=\\sqrt{|y|},\\qquad y(0)=0.\\]\n\nThe identically zero function is a solution. For every \\(a\\geq0\\), another solution is\n\n\\[y_a(t)= \\begin{cases} 0,&t\\leq a,\\\\[2mm] \\dfrac{(t-a)^2}{4},&t>a. \\end{cases}\\]\n\nThese solutions remain at zero for different lengths of time before becoming positive. The right-hand side is continuous but is not locally Lipschitz in \\(y\\) at zero.\n\nLocal uniqueness\n\nThe Picard–Lindelöf theorem provides a standard sufficient condition for local existence and uniqueness. If \\(f\\) is continuous near \\((t_0,y_0)\\) and satisfies a local Lipschitz condition in the state variable, uniformly for nearby \\(t\\), then the initial value problem has a unique local solution.\n\nThe relevant inequality is\n\n\\[\\|f(t,u)-f(t,v)\\|\\leq L\\|u-v\\|\\]\n\nfor states and times in a suitable neighborhood. A sufficient condition is that \\(f\\) have continuous first partial derivatives with respect to the state variables near the initial point.\n\nThese hypotheses are sufficient, not necessary: uniqueness can hold even when local Lipschitz continuity fails.\n\nIntegral formulation and Picard iteration\n\nFor continuous \\(f\\), the differential problem is equivalent to the integral equation\n\n\\[y(t)=y_0+\\int_{t_0}^{t}f(s,y(s))\\,ds.\\]\n\nThis follows from the fundamental theorem of calculus.\n\nStarting with \\(y^{(0)}(t)=y_0\\), Picard iteration defines\n\n\\[y^{(k+1)}(t) =y_0+\\int_{t_0}^{t}f(s,y^{(k)}(s))\\,ds.\\]\n\nUnder the hypotheses of the Picard–Lindelöf theorem, these approximations converge on a sufficiently short interval. One proof treats the integral operator as a contraction mapping and applies the Banach fixed-point theorem."
    },
    {
      "heading": "Local and global solutions",
      "text": "A local solution is defined near the initial point. A global solution exists throughout the entire time range under consideration, such as all \\(t\\geq t_0\\).\n\nEven smooth equations can have solutions that cease to exist after a finite time. For\n\n\\[y'=y^2,\\qquad y(0)=1,\\]\n\nthe solution is\n\n\\[y(t)=\\frac{1}{1-t}.\\]\n\nIts maximal interval containing zero is \\((-\\infty,1)\\). The solution becomes unbounded as \\(t\\to1^-\\), so it cannot be extended through \\(t=1\\) as a finite-valued classical solution. This is called finite-time blow-up.\n\nUnder standard local existence and uniqueness assumptions, a solution has a unique maximal interval of existence. If an endpoint of that interval is finite, the solution cannot remain within a compact subset of the equation’s open time–state domain as it approaches that endpoint. It must become unbounded, approach the boundary of the domain, or otherwise leave every such compact subset.\n\nSuitable growth conditions rule out finite-time blow-up. For example, if \\(f\\) is defined for all states, is continuous and locally Lipschitz in the state variable, and satisfies\n\n\\[\\|f(t,y)\\|\\leq a(t)+b(t)\\|y\\|,\\]\n\nwhere \\(a\\) and \\(b\\) are nonnegative and integrable on finite time intervals, then bounds obtained from the integral equation prevent finite-time escape to infinity."
    },
    {
      "heading": "Dependence on initial data",
      "text": "A problem is called well-posed when a solution exists, is unique, and depends continuously on the prescribed data in the chosen mathematical setting.\n\nIf \\(f\\) has Lipschitz constant \\(L\\) in a region containing two solutions \\(y\\) and \\(z\\), then, for forward times on their common interval,\n\n\\[\\|y(t)-z(t)\\| \\leq e^{L(t-t_0)}\\|y(t_0)-z(t_0)\\|.\\]\n\nThis estimate, obtained using Grönwall’s inequality, quantifies how initial uncertainty can propagate.\n\nContinuous dependence does not mean that errors remain small indefinitely. In some dynamical systems, nearby trajectories separate rapidly, limiting long-term prediction. Such sensitivity, studied in chaos theory, is compatible with existence, uniqueness, and continuous dependence over every fixed finite interval.\n\nWhen \\(f\\) is sufficiently differentiable, sensitivity to the initial state can itself be described by an initial value problem. Writing\n\n\\[S(t)=\\frac{\\partial y(t;y_0)}{\\partial y_0},\\]\n\nthe sensitivity matrix satisfies\n\n\\[S'(t)=D_yf(t,y(t))S(t),\\qquad S(t_0)=I.\\]\n\nHere \\(D_yf\\) is the Jacobian matrix of \\(f\\) with respect to the state and \\(I\\) is the identity matrix."
    },
    {
      "heading": "Linear initial value problems",
      "text": "A linear system has the form\n\n\\[y'(t)=A(t)y(t)+b(t),\\qquad y(t_0)=y_0.\\]\n\nIf the matrix \\(A(t)\\) and vector \\(b(t)\\) are continuous on an interval containing \\(t_0\\), there is a unique solution throughout that interval.\n\nFor constant \\(A\\), the solution can be written using the matrix exponential:\n\n\\[y(t)=e^{A(t-t_0)}y_0+ \\int_{t_0}^{t}e^{A(t-s)}b(s)\\,ds.\\]\n\nThe first term describes evolution from the initial state without forcing; the integral describes the contribution of the forcing term.\n\nFor the scalar equation\n\n\\[y'=\\lambda y,\\qquad y(t_0)=y_0,\\]\n\nthe solution is\n\n\\[y(t)=y_0e^{\\lambda(t-t_0)}.\\]\n\nThis equation is both a basic growth-and-decay model and an important test problem for numerical algorithms."
    },
    {
      "heading": "Numerical solution",
      "text": "Most nonlinear initial value problems do not have useful closed-form solutions. Numerical analysis therefore studies algorithms that approximate the solution at discrete times.\n\nThe explicit Euler method uses\n\n\\[y_{n+1}=y_n+h f(t_n,y_n),\\qquad t_{n+1}=t_n+h.\\]\n\nIt advances along the slope evaluated at the current point. Under suitable smoothness assumptions, its accumulated error over a fixed finite interval is \\(O(h)\\).\n\nMore accurate approaches include Runge–Kutta methods, which evaluate slopes at several points within a step, and multistep methods, which use information from previous steps. Adaptive algorithms estimate local error and adjust the step size to meet specified tolerances.\n\nAccuracy and numerical stability are distinct concerns. Applied to \\(y'=\\lambda y\\), explicit Euler gives\n\n\\[y_{n+1}=(1+h\\lambda)y_n.\\]\n\nFor a negative real \\(\\lambda\\), the exact solution decays, but the numerical solution decays only when\n\n\\[|1+h\\lambda|<1.\\]\n\nThus a numerically unstable step size can produce growth even when the true solution decreases.\n\nA problem is commonly described as stiff when stability constraints force certain explicit methods to take steps much smaller than the solution’s visible variation would otherwise require. Implicit methods are often used in this setting, although they generally require solving algebraic equations at each step."
    },
    {
      "heading": "Partial differential equations",
      "text": "For a partial differential equation, the initial data are usually functions of spatial position rather than a finite list of numbers. Such problems are often called Cauchy problems.\n\nFor the heat equation on all of space,\n\n\\[\\frac{\\partial u}{\\partial t}=\\kappa\\Delta u, \\qquad u(x,0)=u_0(x),\\]\n\nthe initial temperature field \\(u_0\\) determines subsequent diffusion within an appropriate solution class.\n\nFor the wave equation,\n\n\\[\\frac{\\partial^2u}{\\partial t^2}=c^2\\Delta u,\\]\n\ntwo initial fields are normally prescribed:\n\n\\[u(x,0)=u_0(x),\\qquad \\frac{\\partial u}{\\partial t}(x,0)=v_0(x).\\]\n\nThey represent initial displacement and velocity.\n\nOn a bounded spatial domain, initial data generally must be supplemented by boundary conditions, producing an initial-boundary value problem. For classical solutions, compatibility conditions may be necessary where the initial time meets the spatial boundary.\n\nThe number and placement of prescribed data depend on the equation’s structure. Not every surface is suitable for Cauchy data, and existence, uniqueness, and continuous dependence depend on the regularity class used. When classical solutions cease to exist, a weak solution formulation may be required; additional admissibility conditions can then be needed to select a unique evolution."
    },
    {
      "heading": "Applications",
      "text": "Initial value problems occur wherever a model specifies rates of change together with a starting state. In classical mechanics, initial positions and velocities determine motion under prescribed forces. Electrical circuit models use initial capacitor voltages and inductor currents. Reaction kinetics uses initial concentrations, while population models use initial population sizes.\n\nSpatially discretized evolution equations also produce large systems of ordinary differential equations. The resulting initial value problems connect time-integration algorithms with spatial methods such as the finite difference method and the finite element method. Their reliability depends jointly on the underlying model, the initial data, the spatial approximation, and the numerical integration scheme."
    }
  ],
  "links": [
    "differential-equation",
    "ordinary-differential-equation",
    "euclidean-space",
    "boundary-value-problem",
    "continuous-function",
    "lipschitz-continuity",
    "integral",
    "fundamental-theorem-of-calculus",
    "contraction-mapping",
    "banach-fixed-point-theorem",
    "dynamical-system",
    "chaos-theory",
    "jacobian-matrix",
    "matrix",
    "matrix-exponential",
    "numerical-analysis",
    "eulers-method",
    "runge-kutta-methods",
    "numerical-stability",
    "partial-differential-equation",
    "heat-equation",
    "wave-equation",
    "weak-solution",
    "classical-mechanics",
    "finite-difference-method",
    "finite-element-method"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}