---
title: "Olivier Talagrand"
type: "entity"
entity_kind: "person"
status: "hub"
canonical_name: "Olivier Talagrand"
aliases: []
first_seen: "2026-10-15T00:00:00.000Z"
connects_to: ["4D-Var / variational data assimilation","adjoint method","automatic differentiation vs optimization","ECMWF","Andreas Griewank"]
seek_code_commit: "d40ee39"
---


French meteorologist (LMD/ENS) and co-author, with François-Xavier Le
Dimet, of the 1986 *Tellus* paper that founded 4D-Var variational data
assimilation — the technique still used operationally at ECMWF. Matters to
this vault as the Tier-1 first-person witness behind two claim-notes: his
2018 ECMWF reminiscence states that automatic-differentiation and
optimization researchers "had little contact" with each other
([[claim-talagrand-ad-and-optimization-had-little-contact]]), a third,
independently-arrived-at instance of the vault's ML/AD-unawareness
pattern; and the same talk supplies the corrected history of 4D-Var's own
adjoint method — a documented transmission from a 1979–81 ECMWF seminar
and a folder of optimization algorithms, not an independent rediscovery
([[claim-4d-var-adjoint-method-transmitted-not-independent]]).

Named alongside him in that history, as mentions rather than hubs (no
separate standing in the vault yet): François-Xavier Le Dimet (his
co-author, who supplied the adjoint-equations algorithm from his own
optimal-control training under Jacques-Louis Lions), G. R. Kontarev (gave
the 1979–81 seminar that introduced the technique to ECMWF), and
Academician G. I. Marchuk (whose group Kontarev came from).

## References
- [[claim-talagrand-ad-and-optimization-had-little-contact]]
- [[claim-4d-var-adjoint-method-transmitted-not-independent]]
- Capture: 10-inbox/raw/2026-10-13-hop-4d-var-adjoint-unawareness.md

> [!cali] Cali's daily review 2026-10-15 (Claude Fable 5.1): The page states as fact what its source note flags `[unverified-quote]` ('a documented transmission from a 1979–81 ECMWF seminar … G. R. Kontarev (gave the 1979–81 seminar)'); the passages verify today (slides sha 5080b298… l. 47–54, 76–77), so the fact holds — but the slide says 'Around 1979-80', and two visitors came (Kontarev and Lykossov), Kontarev lectured. 'the technique still used operationally at ECMWF' is in no note, capture or archive (model memory). 'French meteorologist (LMD/ENS)' is true (archive l. 11) but reached no note.