Hadamard well-posedness

Hadamard well-posedness is the criterion, introduced by Jacques Hadamard in 1902, that a mathematical problem must satisfy three conditions to be considered properly set: a solution must exist, the solution must be unique, and the solution must depend continuously on the initial or boundary data. The third condition — continuous dependence — is the most subtle and the most frequently violated in practice. It demands that small changes in the problem's specification produce only small changes in the solution, ensuring that the problem is solvable in the presence of the inevitable measurement and computational errors of physical reality.

Problems that fail the Hadamard criteria are called ill-posed, and they are not merely inconvenient. They signal a mismatch between the mathematical model and the physical situation it purports to describe. The Navier-Stokes equations backward in time are ill-posed: fluid flow is irreversible, and attempting to reconstruct past states from present data amplifies errors exponentially. Inverse problems in imaging, tomography, and geophysics are typically ill-posed, and their solution requires regularization — the deliberate introduction of additional constraints or prior information to restore well-posedness.

Hadamard's criteria encode a deep philosophical assumption: that physical reality is structured in a way that makes mathematical prediction possible. When a problem is ill-posed, the failure is not in the mathematics but in the fit between the mathematical frame and the physical process. The art of applied mathematics is largely the art of reframing ill-posed problems into well-posed ones — of discovering the right variables, the right scales, and the right symmetries that make the problem tractable.

Hadamard well-posedness is often treated as a technical requirement, but it is better understood as an epistemological boundary. It marks the line between problems we can expect to solve and problems that outrun the structure of human knowledge. The universe does not guarantee that its dynamics are well-posed from every perspective. We are lucky that so many are.