Martin Rubey wrote:
>
> I'd like to commit the attached patch. It moves functionality to test
> approximate equality from elemnum.input to unittest.spad, which turns out tpo
> be a good thing, because there was a mistake in elemnum (epsilonrelative
> instead of epsilonRelative) that went unnoticed for many revisions, and made
> us
> believe that numerics is better than it was.
>
Looks OK.
> I also add a test for branch cuts that shows another feature of the new
> functionality: useful error messages.
>
> Waldek, I test values of asin on a circle around the origin with radius 2. Is
> this what you had in mind?
>
I am not sure what you intend to test, but the test I have in mind
is like:
steps := 1000
h := %pi/steps
-- need to choose correct constant, 5.0 is just w wild guess
eps := 5.0/steps
valp := asin(2*exp(%i*h))
for i in 2..(steps-1) repeat
valn := asin(2*exp(%i*h*i)
norm(valn-valp) > eps => error "Jump detected"
valp := valn
IIUC you test that numerical derivative agrees with symbolic one,
which is a different test -- given seqence of points x_n, your test
will detect jump between x_n and x_n+tdiff, but miss jumps between
x_n+tdiff and x_{n+1}. The test I wrote above is designed to detect
jumps between x_n and x_{n+1}.
--
Waldek Hebisch
[EMAIL PROTECTED]
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/fricas-devel?hl=en
-~----------~----~----~----~------~----~------~--~---
