Patrick Cousot. The Essence of Event-B from an Abstract Interpretation Perspective

 


  • Patrick Cousot.
    The Essence of Event-B from an Abstract Interpretation Perspective.
    In Rigorous Engineering of Trustworhy Systems — Essays Dedicated to Jean-Raymond Abrial, Egon Börger and Yamine Aït Ameur (Eds), LNCS 16750, Springer, 2026.
  • Paper (29 pages): PDF (515 KB)
  • Abstract: We explain the basic concepts of Event-B, that is the specification of machines and refinement, in light of abstract interpretation. After formally defining the trace semantics of systems handled by Event-B, we show that machines are specifications of abstractions of this semantics while refinement is an inverse abstract interpretation reducing the degree of nondeterminism of this trace semantics. We conclude by proposing in induction principle for program verification exploiting both property abstraction and program refinement.

  • Keywords: Jean-Raymond Abrial Event-B · Abstraction · Abstract machine · Semantics · Refinement · Verification · Abstract interpretation · Formal methods. 

 

" I have had the pleasure to have my office close to that of Jean-Raymond Abrial
during my thesis in Grenoble and to share a common passion for climbing, mainly
in the Vercors limestone massif and the French and Swiss granite and glacier
Alps. At the time Jean-Raymond was working on data semantics [1] and started
to be increasingly interested in formal program verification and construction.
He had strong ideas on computer science, in particular program design and
verification, that he explained very clearly, and was always ready for discussion
on many subjects with a young ignorant trying to understand abstraction.
When Jean-Raymond left Grenoble, I was given the charge of his class on
data structures, programming, and verification that gave me a strong taste for
teaching, in particular formal methods.
We met later in conferences, meetings, summer schools, and seminars in
particular during his stay at ETH Zurich and sometimes also by chance, like in
Venice, on the Ponte dell’Accademia, where, for a long time on the bridge, we
passionately discussed the genius of Venice architecture and urbanism design.
Unfortunately, Jean-Raymond left Grenoble before I had established the
foundations of abstract interpretation with Radhia and therefore we have never
had scientific collaborations. So I was surprised to be invited by Egon Boerger
and Yamine Aït-Ameur to contribute to this Gedenkschrift with the proposed
theme of “Relating program verification by abstract interpretation with proofs in
Event-B”. Since no one, including myself, seemed to have understood the corre-
spondence during the “Workshop on Refinement and Abstraction” (ETL Osaka apan, 15–17 November 1999), and I had made no progress since then, I thought
taking on the challenge would be interesting.
Although I had read the B-book [2] decades ago, I had to learn Event-B,
almost from scratch [4,7] while knowing that I could not absorb the abundant
literature on Event-B (event-b.org) in the allotted time.
It turns out that I could slightly extend the challenge to an explanation
of the semantics of Event-B in terms of abstract interpretation [16]. Of course
this brings nothing new to Event-B but this formalization helps me to better
understand two related forms of abstractions when respectively applied to models
(as in Event-B) and to properties of these models (as in abstract interpretation).
It might also be useful to conceive Event-B refinements by calculational design
of the concrete from the given abstract using a formal correspondence between
their semantics (such as a Galois connection or equivalent). " [...]

Comments

Popular posts from this blog

Jean-Raymond Abrial (1938, 2025)

Why might we teach "Formal Methods" informally? And to whom? And when? Carroll Morgan

Jean-Raymond Abrial (1938, 2025) English