A situation where one or more of the hallmarks of well-posedness—existence, uniqueness, or continuous dependence on data—fail for a mathematical problem, often producing nonphysical, nonunique, or unstable solutions that amplify data errors.

Principle

Ill-posedness arises when the mapping from data to solution is not well-behaved (e.g., is not continuous or not injective), frequently occurring in inverse problems, analytic continuation, or degenerate differential operators.

Demonstration

The classical backward heat equation is ill-posed: small measurement noise in final-time temperature leads to exponentially growing errors when attempting to reconstruct earlier states, showing non-continuous dependence on data.

Misapplication

Attempting to solve an ill-posed inverse problem by direct inversion without regularization can produce spurious, wildly varying solutions that fit noisy data but have no predictive value.

Consequence

Ill-posedness necessitates regularization, model refinement, or additional constraints (priors) to obtain stable, meaningful approximate solutions and quantifiable uncertainty statements for inference or control.

Reversal

The reversed condition is well-posedness where solutions exist uniquely and depend continuously on data, permitting direct interpretation and stable numerics without ad hoc stabilization.

Boundary

Applies to the mathematical formulation and chosen topology; ill-posedness in one norm may be resolved by changing function spaces, adding physical constraints, or reformulating the problem, but some problems remain intrinsically ill-posed.

Semantic Tension

Competes with pragmatic solution concepts: an ill-posed model can still be useful if coupled with principled regularization or probabilistic priors; tension is between theoretical failure of Hadamard criteria and practical recoverability.

Synthesis

Ill-posedness describes the breakdown of existence, uniqueness, or stability in a problem's data-to-solution map, signaling the need for additional information, regularization, or reformulation to produce usable results.