Le 12/09/2019 à 18:55, Stéphane Mottelet a écrit :
I prefer the display after applying
https://codereview.scilab.org/#/c/20981/:
--> x0=%pi/4;h=%eps/2
h =
1.110D-16
--> cos(x0+%i*h)
ans =
0.7071068 - 7.850D-17i
However, we could discuss if the arbitrary switch to "e" mode is
desirable or not, but since Scilab 6.0 we have got used to this
display mixing "v" and "e" mode...
... and IMO it's very suitable. The only rule should be that, */for a
given format's width/*, */the more the number of significant displayed
figures the better./* When the exponent format is used, it takes 4 or 5
digits (D+12, D+123). As a consequence, in v mode, small (absolute)
values should be displayed in normal mode down to 4 non-significant 0
(e.g. 0.000012), and switched to "e" mode beyond.
In v mode, forcing the display of 0.0123456 in "e" mode on the (bad)
reason that some other huge or tiny numbers in the matrix are displayed
in this mode, would print 1.234D-02, and so make invisible 2 significant
figures.
I do not see any reason that would impose the display mode to be
homogeneous over all elements of a matrix, possibly canceling the
display of significant figures in the given format's width.
Regards
Samuel
_______________________________________________
users mailing list
users@lists.scilab.org
http://lists.scilab.org/mailman/listinfo/users