A numerical technique for two-point boundary value problems that treats unknown boundary data as parameters, integrates an initial-value problem from one boundary, and adjusts the parameters (via root-finding or optimization) so the solution satisfies the other boundary condition.

Principle

Transform the boundary value problem into a parameter-selection problem by defining a shooting function that maps guessed initial data to the residual at the target boundary, then solve for zeros of that function.

Demonstration

For y'' = f(x,y,y'), with y(a)=alpha and y(b)=beta, choose a trial y'(a)=p, integrate the IVP from a to b to obtain y(b;p), and then apply a root-finding method to solve S(p)=y(b;p)-beta=0 for p.

Misapplication

Applying single-shot shooting to stiff or highly sensitive BVPs without stabilization; using straightforward shooting on problems with many oscillatory modes often leads to numerical instability or failure to converge.

Consequence

When successful, shooting reduces the BVP to standard IVP integration plus scalar or low-dimensional root finding, allowing reuse of high-quality ODE solvers but possibly missing multiple solutions or suffering from poor conditioning.

Reversal

Direct discretization methods such as finite differences or collocation convert the BVP into a global algebraic system and avoid treating boundary conditions as adjustable initial parameters.

Boundary

Appropriate for boundary value problems for ordinary differential equations, especially low-dimensional ODEs with well-behaved sensitivity; not generally suitable for large-scale PDE boundary problems or BVPs with extreme stiffness unless augmented (e.g., multiple shooting).

Semantic Tension

Tension exists with multiple shooting and direct discretization: shooting emphasizes IVP integration and parameter root-finding, whereas collocation/discretization treat the entire domain simultaneously and may be more stable.

Synthesis

Shooting Method = treat unknown initial data as parameters, integrate as an IVP, and apply a root-finding loop to satisfy end boundary conditions; effective when sensitivity is manageable and IVP solvers are robust.