Initial Value Problems and Boundary Value Problems

date2026-07-16document_iddoc_16e0df958ec1758f13d6f38b411e423edescription初期値問題と境界値問題を、微分方程式の一般解に条件を課して解の候補を制約する方法として初学者向けに整理する。prerequisites微分方程式とは何か / 導関数の定義と差商type講義content_typelecturestatusactive

mathdifferential-equationsinitial-value-problemboundary-value-problemlecture

1Introduction

This lecture explains how a differential equation and prescribed conditions jointly determine a solution set. Integrating a differential equation commonly produces a family of functions containing arbitrary constants. Conditions constrain that family, but incompatible or dependent conditions may leave no solution or multiple solutions.

For example, integrating

y^{'} = 2 x

gives

y = x^{2} + C .

The constant $C$ is undetermined until a condition is imposed. The condition $y (0) = 1$ yields $C = 1$ and therefore selects the solution $y = x^{2} + 1$ .

2Initial Value Problems

An initial value problem specifies the value of an unknown function and, when required, some of its derivatives at a common reference point. For example,

y^{'} = f (x, y), y (x_{0}) = y_{0}

is a first-order initial value problem.

For a second-order equation, one commonly prescribes both the function and its first derivative:

y^{'}' = F (x, y, y^{'}), y (x_{0}) = y_{0}, y^{'} (x_{0}) = v_{0.}

When the independent variable represents time, these data may specify the current state from which an evolution is determined. The same mathematical form is also used when the independent variable is not time.

3Boundary Value Problems

A boundary value problem specifies conditions at the endpoints of an interval or on the boundary of a domain. For example,

y^{'}' = 0, y (0) = 0, y (1) = 1

is a boundary value problem.

Boundary value problems arise when endpoint or boundary data constrain an entire spatial profile, as in steady-state temperature distributions, the shapes of strings, and other equilibrium configurations.

4Proposition 1: Conditions Constrain Arbitrary Constants

Suppose that a general solution of a differential equation is given by

y = ϕ (x, C),

where $C$ is an arbitrary constant. Imposing $y (x_{0}) = y_{0}$ restricts the admissible values of $C$ to solutions of

ϕ (x_{0}, C) = y_{0.}

The constant $C$ is determined uniquely only if this equation has a unique solution.

4.1Proof

The representation $y = ϕ (x, C)$ gives $y (x_{0}) = ϕ (x_{0}, C)$ . Hence the condition $y (x_{0}) = y_{0}$ requires

ϕ (x_{0}, C) = y_{0.}

Conversely, any value of $C$ satisfying this equation gives

y (x_{0}) = ϕ (x_{0}, C) = y_{0.}

Thus the condition becomes an equation constraining the arbitrary constant. If that equation has no solution, the conditioned problem has no solution within the stated general family; if it has multiple solutions, multiple candidates remain.

This proposition shows that imposing a condition on a general family can be reduced to solving an equation for its arbitrary constants.

5Example 1: A First-Order Initial Value Problem

Consider

y^{'} = - y, y (0) = 1.

Solving the differential equation alone gives

y = C e^{- x} .

The initial condition yields

1 = C e^{0} = C,

y = e^{- x} .

Verification is essential: $y^{'} = - e^{- x} = - y$ and $y (0) = 1$ , so the function satisfies both the equation and the condition.

6Example 2: A Second-Order Boundary Value Problem

Consider

y^{'}' = 0, y (0) = 0, y (1) = 1.

Integrating twice gives $y^{'} = A$ and then

y = A x + B .

The boundary condition $y (0) = 0$ gives $B = 0$ , while $y (1) = 1$ gives $A = 1$ . Therefore,

y = x .

The same basic operation occurs in initial and boundary value problems: determine a general solution and substitute the conditions to constrain its arbitrary constants.

The presence of $n$ arbitrary constants and $n$ conditions does not by itself ensure unique determination. In general, substitution produces a system of equations for the constants, and its solvability and uniqueness must be checked. Linearly independent solutions that span the solution space of a homogeneous linear ODE are called basis solutions. When the homogeneous part of a general linear ODE solution is expressed as a linear combination of basis solutions and the imposed conditions are linear, the resulting system is linear. If its coefficient matrix is nonsingular, the arbitrary constants are uniquely determined.

Let $y_{1}, \dots, y_{n}$ be a basis of solutions of an $n$ th-order homogeneous linear ODE. For standard initial conditions specifying $y, y^{'}, \dots, y^{(n - 1)}$ at the same point $x_{0}$ , the coefficient matrix for the integration constants is

{(y_{j^{(i - 1)}} (x_{0}))}_{{i, j = 1}^{n}} .

Its determinant is the Wronskian $W (y_{1}, \dots, y_{n}) (x_{0})$ . If the Wronskian is nonzero, the matrix is nonsingular and the integration constants are uniquely determined. For boundary conditions imposed at different endpoints or for general linear conditions, one must instead test the separately constructed condition matrix; its determinant is not generally a Wronskian.

data/lecture/math/linear-operator/affine-spaces-and-condition-determination.lecture.n.md data/lecture/math/differential-equations/variation-of-parameters-and-wronskian.lecture.n.md

7Comparison of Initial and Boundary Value Problems

Item	Initial value problem	Boundary value problem
Location of data	Function and derivative values at one reference point	Data at endpoints or on a boundary
Typical interpretation	Determine an evolution from a specified state	Constrain a profile through boundary data
Representative applications	Population growth, cooling, motion	Steady temperature, strings, normal modes
Principal caution	Existence and uniqueness require hypotheses	No solution or multiple solutions may occur

The term normal mode appears in the table but is not developed here. In some problems, only particular oscillations satisfy the boundary conditions; their analysis leads to eigenvalues and eigenfunctions.

8Caution: The Number of Conditions Is Not Sufficient

A second-order differential equation commonly has a general solution containing two arbitrary constants, so two conditions are often required. Nevertheless, two conditions do not guarantee the existence of exactly one solution.

Consider

y^{'}' = 0.

Twice integrating gives

y = A x + B, y^{'} = A .

Now impose the derivative boundary conditions

y^{'} (0) = 0, y^{'} (1) = 1.

The first requires $A = 0$ , whereas the second requires $A = 1$ . These requirements are inconsistent, so the boundary value problem has no solution.

By contrast, if

y^{'} (0) = 0, y^{'} (1) = 0,

then only $A = 0$ is fixed and $B$ remains arbitrary. Hence every constant function

y = B

is a solution, and the problem has infinitely many solutions.

Thus, even when the number of conditions matches the order, the placement of the conditions and the structure of the equation may lead to no solution or multiple solutions. Existence and uniqueness theory analyzes precisely these questions.

data/lecture/math/differential-equations/existence-uniqueness-and-lipschitz-condition.lecture.n.md

9Analysis Procedure

Identify the unknown function and independent variable.
Determine the order of the equation.
Determine whether the conditions are imposed at one reference point or distributed across endpoints or a boundary.
Determine a general solution when one is available.
Substitute the conditions, solve the resulting constraint equations, and identify the admissible constants.
Substitute the result into the original equation and all prescribed conditions to verify it.

10Next Lecture

The next lecture classifies first-order differential equations by form and explains how to identify applicable solution methods.

data/lecture/math/differential-equations/classifying-first-order-odes.lecture.n.md

11Exercises

data/exercise/math/differential-equations/introduction-to-odes-and-direct-integration.exercise.n.md data/exercise/math/differential-equations/existence-uniqueness-and-numerical-methods.exercise.n.md