> Le 14 août 2018 à 20:37, TB a écrit :
>
> Because float(pi/2) is not exactly pi/2:
>
> sage: n(pi/2 - float(pi/2), 53)
> 0.000
> sage: n(pi/2 - float(pi/2), 54) # also for perc > 54 of course
> 1.11022302462516e-16
and what’s the advantage of numerical_approx(v, digits=10) over
Because float(pi/2) is not exactly pi/2:
sage: n(pi/2 - float(pi/2), 53)
0.000
sage: n(pi/2 - float(pi/2), 54) # also for perc > 54 of course
1.11022302462516e-16
This does mean that:
sage: float(cos(pi/2))
0.0
sage: float(sin(pi))
0.0
Regards,
TB
On 14/08/18 21:02,
I looking for a smart way to fix the following issue that we have when
plotting graphs (see #22050 and #24512 for instance):
cos(pi/2) is not 0 and sin(pi) is not 0 !
sage: G = Graph(4)
sage: _circle_embedding(G, G.vertices())
sage: G._pos
{0: (1.0, 0.0),
1: (6.123233995736766e-17, 1.0), #