Other meanings of Initial value problem
Differential equations
An initial value problem is a differential equation together with conditions that specify the unknown function, or selected derivatives of it, at one starting point. The conditions turn a family of possible solutions into a problem of finding the particular trajectory that passes through prescribed initial data.
An initial value problem specifies a differential equation and the value of the unknown function and, when necessary, its derivatives at one initial point. A first-order problem commonly has the form y' = f(t,y), with y(t0) = y0; a second-order problem may prescribe both y(t0) and y'(t0).1
The initial data select one member from the broader collection of solutions of the differential equation. In mechanics, for example, position and velocity at a specified time determine the initial state of a moving body, while in population models the initial value may represent the population at the beginning of an observation period. The independent variable need not be time: spatial, financial, and other evolution problems can also be posed as initial value problems.
The central theoretical questions are whether a solution exists near the initial point and whether the stated data determine only one solution. For a first-order equation y' = f(t,y), continuity of f near the initial point supports local existence, while additional regularity in the dependent variable—often a local Lipschitz condition—supports uniqueness in the Picard–Lindelöf theorem.
These are local conclusions: a solution may exist and be unique only until it reaches a singularity, becomes unbounded, or leaves the region where the hypotheses hold. Linear systems with continuous coefficients have stronger global results on any interval where the coefficients remain defined. If uniqueness fails, several solution curves can pass through the same initial state; if existence fails, no differentiable curve satisfies both the equation and the data.
Initial value problems are solved exactly when integration, substitution, series, transforms, or other analytic methods produce a usable formula; otherwise, numerical methods approximate the solution step by step from the initial state. Euler's method advances using the current slope, while Runge–Kutta methods sample several slopes within each step and generally achieve higher accuracy for comparable step sizes.2
Numerical integration introduces truncation and round-off error, so step-size selection, stability, and error control matter. Stiff problems can contain rapidly decaying components alongside slowly varying ones; explicit methods may then require impractically small steps, motivating implicit methods and specialized solvers.3 The computed answer is therefore an approximation whose reliability depends on both the mathematical model and the numerical method.
Initial value problems also arise in settings where the initial data are functions, vectors, or constrained states rather than a single number. A system of first-order equations can represent a higher-order equation, and its initial condition is a vector in phase space; trajectories that begin nearby may nevertheless separate rapidly in sensitive or chaotic systems.4
Boundary value problems differ because their conditions are imposed at more than one point, and converting one type into the other can change both solvability and numerical behavior. Some equations have a well-defined solution forward in the independent variable but not backward, particularly when the dynamics amplify measurement or rounding errors. Initial data may also be uncertain, leading to ensemble or probabilistic calculations rather than one deterministic trajectory. These issues connect IVPs with dynamical systems, control theory, inverse problems, and scientific simulation.
Initial value problems may be formulated for scalar equations, systems of equations, or higher-order equations rewritten as first-order systems.
Help improve the encyclopedia. Reports go straight to the site manager.