Re: [isabelle] Using implications

I agree with your proof. But then, if it is to define how to *approximate* a function f and a function g to be the same by comparing those x's and y's
c, then should the = operator be replaced here? It seems the problem here
is caused by overloading = with a form of approximated =, right? If so, how should one go about introducing a new operator, say for approximation?



On Jan 26, 2011 6:23pm, Brian Huffman <brianh at> wrote:
On Wed, Jan 26, 2011 at 10:14 AM, Eg Gloor egglue at> wrote:

> axiomatization

> S :: "real set" and

> foo :: "real => real" and

> bar :: "real => real" and

> bax :: "real => real" and

> c :: real

> where ax1: "ALL f g. (ALL x y. x > c & y > cfx = gy) --> f = g"

> and ax2: "ALL a b. bax a > c & bax b > c foo (bax a) = bar (bax b)"

Once again, we see that axioms are evil. Your ax1 is already

inconsistent all by itself:

lemma "False"

proof -

let ?f = "%x::real. if x > c then (0::real) else 1"

let ?g = "%x::real. if x > c then (0::real) else 2"

have "?f = ?g"

by (rule ax1 [rule_format], simp)

hence "?fc = ?g c"

by (rule arg_cong)

thus "False"

by simp


-> Brian

This archive was generated by a fusion of Pipermail (Mailman edition) and MHonArc.