coq-club AT inria.fr
Subject: The Coq mailing list
List archive
- From: Ewen Denney <denney AT irisa.fr>
- To: Benjamin Werner <Benjamin.Werner AT inria.fr>
- Subject: Re: witness of existential
- Date: Wed, 18 Aug 1999 15:50:53 +0200
- Organization: IRISA, Campus de Beaulieu, 35042 Rennes Cedex, France
Thanks a lot. I have one final question on this general topic.
Suppose I want to specify something using a mixture of logical and
computational existentials. For example, my aim is to construct a program
t : T1 -> T2 which realises a relation R : T1 -> T2 -> Prop (ie. the
graph of t should be contained in the relation). However, we do not start
with R but rather a specification of R, Defn : (T1 -> T2 -> Prop) ->
Prop, say.
I want to write something like
exists R . Defn (R) /\ {t: T1 -> T2 | (x:T1) R x (tx)}
but this won't work. For this example I can put the sig outside the
exists but, in general, I have a family of such definitions and it might
prove fiddly to reformulate like this.
I don't want to extract the R, but it must be provided somehow during the
prove.
Ewen
- witness of existential, Ewen Denney
- Re: witness of existential, Jean-Christophe Filliatre
- Re: witness of existential,
Benjamin Werner
- Re: witness of existential, Ewen Denney
- Re: witness of existential, Benjamin Werner
- Re: witness of existential, Ewen Denney
Archive powered by MhonArc 2.6.16.