Stopping criteria guaranteeing user-specified local error in optimization-based ODE parameter estimation

Develop stopping criteria for optimization-based parameter estimation methods for ordinary differential equation (ODE) models that guarantee the recovered parameter values are within a user-specified local error tolerance, even in cases where convergence of the optimization procedure can be proven.

Background

The paper contrasts optimization-based and algebra-based approaches for parameter estimation in ODE models. It points out a key limitation of optimization-based methods: even when convergence can be established, there is no established way to certify that the returned parameters meet a user-specified local error tolerance.

This gap concerns practical stopping criteria that can guarantee accuracy upon termination, an issue highlighted as not yet known in the current literature referenced by the authors.