---
title: "Henry J. Kelley"
type: "entity"
entity_kind: "person"
status: "hub"
canonical_name: "Henry J. Kelley"
aliases: ["Henry Kelley","H. J. Kelley"]
first_seen: "2026-06-30T00:00:00.000Z"
writer_model: "claude-sonnet-5"
connects_to: ["gradient method in optimal control","adjoint system","backpropagation","[[entity-arthur-bryson|Arthur Bryson]]","steepest descent"]
seek_code_commit: "f8a723c"
---


Aerospace engineer at Grumman who published "Gradient Theory of Optimal Flight Paths" (*ARS Journal* 30, October 1960), a method for optimising rocket and flight trajectories by steepest descent, using a system adjoint to the linearised dynamics to obtain the gradient without solving a two-point boundary-value problem. Matters to this vault as the earliest node of the Kelley/Bryson optimal-control lineage that later accounts of backpropagation point back to. Not the same person as [[entity-truman-lee-kelley]].

## References
- [[claim-kelley-bryson-optimal-control-precursor]] · [[claim-backprop-special-case-kelley-bryson-formula]] · [[claim-kelley-1960-reference-list-cites-no-nyquist-or-bode]]
- 2026-09-25: read alongside Widrow–Hoff 1960, his paper turns out to share only a steepest-descent ancestor (Hadamard, Courant) with LMS, not a mechanism or a citation. ([[claim-kelley-and-widrow-hoff-1960-share-steepest-descent-root-not-backward-sweep-mechanism]])
- 2026-09-26: Werbos's 1988 paper does not name Kelley anywhere in the reference list or in body pp. 350-355 (manual read, no grep receipt). ([[claim-werbos-1988-body-pp-350-355-names-no-isaacs-bellman-pontryagin-kelley-or-bryson]])