Answers are grounded in this paper and cite their sources.
This note develops new corrected endpoints that address quadratic exactness and overshoot minimization in smoothing algorithms.
The zero-lag endpoint of the exponentially weighted two-rate family, XEPMA, has zero first kernel moment at every smoothness order but is quadratic-exact (zero second moment as well) only at order s = 1. This note develops the two named endpoints that correct the second moment away from that order. A single moment knob, the term c(s) Δ^2 EMA^[s+2], adjusts the kernel's second moment while leaving its gain and lag untouched, so it composes with the endpoint at any order without re-deriving either. Setting c(s) to cancel the second moment gives QuadraticXEPMA, quadratic-exact at every real s >= 1; setting the same knob to the exact step-overshoot minimiser gives DampedXEPMA, whose step response sits below the endpoint's own overshoot at every order and period. We give the closed forms, the continuous-limit optima, and an honest account of the cost: quadratic exactness roughly doubles the endpoint's noise gain, so the full correction is a completeness result rather than a deployment default, while the damped endpoint is Pareto-useful on overshoot. Every displayed quantity is verified to machine precision by scripts released with the reference implementation.
No takes yet. Share an insight, caveat, or question.
Marcus John Hendry Don (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: