On Tuesday, October 7, 2014 1:33:19 PM UTC-7, Jeroen Demeyer wrote:
>
>
> > The ramifications for coercion discovery are also quite big:
> No, you're understanding it wrong. The parent of the result of any
> computation would be the same as now. The result of adding 2 RealNumbers
> with differin
This is of historical interest:
http://wstein.org/sage_strategy.jpg
This was from about 10 years ago -- late 2004 -- it's my planning
sheet from when I was first getting serious about "Sage" (then called
Manin).
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.
On Tuesday, October 7, 2014 9:32:56 PM UTC+1, Jeroen Demeyer wrote:
>
> I would argue that the result would be *more* predictable and certainly
> more accurate than now.
>
How do you explain somebody when precision is lost due to the extra
rounding for coercion?
Is composition of "compatible"
On 2014-10-07 17:45, Nils Bruin wrote:
Another conclusion one can draw from that is that floats got modelled
wrong: that precision is a property of the element, not the parent.
That cannot really work, since we really want RealField(100)(1/3) to be
different from RealField(53)(1/3)
That's a v
On 2014-10-07 19:03, Volker Braun wrote:
IMHO way too complicated, and too few people understand coercion as it
is already. I'd rather have one unnecessary rounding than ring
operations where I can't predict the outcome without reading the source
first.
I would argue that the result would be *mor
On Tuesday, October 7, 2014 7:14:57 PM UTC+1, Nils Bruin wrote:
>
> I'd say in p-adics elements simply carry around their own precision. The
> different parents simply supply a cap on how big that precision can be.
>
Thats just another way of saying that they implement precision both in the
pare
Hey Jeroen,
My understanding of why we coerce to the lower precision is to have at
least the required amount of significant figures, and I'd guess doing
things this way is more numberically stable. In the same vein, I think
we're better off just letting the coercion to go through normally rat
On Tuesday, October 7, 2014 10:03:54 AM UTC-7, Volker Braun wrote:
>
> p-adics are even more weird, they coerce in the wrong direction. And, to
> patch up that mistake, have a precision in the parent and another precision
> in the element:
>
I'd say in p-adics elements simply carry around their
IMHO way too complicated, and too few people understand coercion as it is
already. I'd rather have one unnecessary rounding than ring operations
where I can't predict the outcome without reading the source first.
The only takeaway is: If you care about the last decimal places then start
by def
Just an update because I was not clear in my other message:
> The real issue is the FinitePoset code:
The actual bug I reported was created by #12536
(http://trac.sagemath.org/ticket/12536).
Indeed, it is incorrect to relabel the (automatically computed) linear
extension of a Poset A (when A is
On Tuesday, October 7, 2014 7:03:54 AM UTC-7, Jeroen Demeyer wrote:
>
> Hello,
>
> Consider the following Sage session:
>
> sage: a = RR(-1)
> sage: b = 1.0001
> sage: a + b
> 0.000
> sage: a._add_(b)
> 1.03397576569128e-24
>
> Here, the coercion model makes
Hello,
Consider the following Sage session:
sage: a = RR(-1)
sage: b = 1.0001
sage: a + b
0.000
sage: a._add_(b)
1.03397576569128e-24
Here, the coercion model makes the output less precise than what is
possible: a + b first reduces the precision of b, rounding t
12 matches
Mail list logo
