Continuous derivative jets on a closed interval #
On a nondegenerate closed time interval, derivatives at the endpoints are one-sided.
An interior derivative with a continuous extension to the closed interval is the derivative
within that interval, including at its endpoints. Applying this observation successively
identifies every field of a continuous derivative jet with iteratedDerivWithin.
The jet criterion contDiffOn_Icc_iff_exists_continuous_jet characterizes finite-order
smoothness on a closed interval using only continuous fields and ordinary interior
derivatives. It allows time derivatives of functions on closed slabs to be specified by
continuous extensions without choosing an extension of the function outside the slab.
The codomain may itself be a space of continuous multilinear maps.
The endpoint argument uses Mathlib's
hasFDerivWithinAt_closure_of_tendsto_fderiv; higher orders use
iteratedDerivWithin_succ and the native criterion for ContDiffOn in terms of iterated
derivatives. No completeness assumption on the codomain is needed.
ContDiffOn.isBounded_image_deriv bounds the derivative of a C¹ function on a closed interval:
it agrees with the continuous within-interval derivative away from the two endpoints.
A continuous extension of an interior derivative is the derivative within the closed interval, also at its two endpoints.
Every field in a finite continuous interior derivative jet is the corresponding iterated derivative within the closed interval. In particular its endpoint values are canonical.
A finite continuous derivative jet in the interior yields smoothness up to both endpoints. Only the jet values on the closed interval matter.
Finite-order smoothness on a nondegenerate closed interval is equivalent to the existence of continuous derivative fields satisfying the ordinary derivative identities in the interior. There is no condition on the fields outside the interval or above the specified order.
The derivative of a function that is C¹ on a closed interval has bounded image on that
interval. At the two endpoints deriv f may differ from the one-sided derivative, but these
contribute only two values.