Power-law comparison for vanishing derivative regularization

ID: power-law-comparison-for-vanishing-derivative-regularization

For with endpoint values zero and one, the zero-penalty minimizer has infinite derivative penalty. Use instead. The mismatch is at most , while the derivative penalty is . Balancing and by gives an upper energy bound of order . Nonnegativity then proves convergence of minimum energies to zero. This is a recovery construction; no claim of a matching lower asymptotic constant is needed.

New to topics? Read the docs here!