Printing floating points
Bruce Carneal
bcarneal at gmail.com
Thu Jan 28 15:13:57 UTC 2021
On Thursday, 28 January 2021 at 08:30:40 UTC, Ola Fosheim Grøstad
wrote:
> On Thursday, 28 January 2021 at 02:25:57 UTC, Bruce Carneal
> wrote:
>> Regardless, you clearly do not understand what I'm asking in
>> question 1) and I do not think it likely that additional
>> interaction on this topic will be useful to either of us.
>
> Asking for a mapping does not really make much sense without
> specifying the domain/codomain, but my answer actually holds
> even then.
For the problem under discussion the domain is 0 .. 2**NBits
(binary FP) and the codomain is the problem specified string
output of the function. This has not changed and I had thought
it was clear to all concerned throughout.
The correctness check for the mapping in this case is trivial, by
which I mean: obvious, readily understood, not needing further
explanation, ... a *good* thing in this context.
>
> Anyway, if you _actually_ are interested in this topic then you
> could start by looking at automatic test case generation using
> SMT solvers.
I am actually interested in proving correctness by establishing
full coverage of the domain with something less expensive than
direct enumeration, less expensive than a for loop across all
possibilities (an SMT-like solution). This is what I thought
that you said you had accomplished with a trivial coverage, your
"no problem" remark. IOW, I thought you claimed to have solved
the problem.
I found such a claim remarkable, I did not see it, so I sought
clarification. IIUC now, I was mistaken then. You have no
particular insight to offer toward the solution of the above
problem and, by your lights, never made such a claim.
Apparently I took your well meaning remarks much too seriously.
More information about the Digitalmars-d
mailing list