[Octave-bug-tracker] [bug #37896] quadcc(f, -Inf, Inf) incorrectly returns 0 (quadgk too)

[Rik]

Update of bug #37896 (project octave):

                  Status:               Need Info => Invalid                
             Open/Closed:                    Open => Closed                 

    _______________________________________________________

Follow-up Comment #3:

Computers are great aids, but they can't replace humans.  Try plotting the
function to get some intuition about it.  Because of the Heaviside function,
the value is 0 for all times less than 0.  And while it has a peak at early
times, it quickly decays off to zero.  When you ask quadcc or quadgk to
integrate a function over an infinite range it will check a couple of points. 
In this case the value at -Inf,0,+Inf which are all zero.  The integration
routine simply can't find anything to integrate and returns 0.

If I use an integration range that includes the non-zero portions, say t = [0,
5000], I get


octave:15> quadcc (f, 0, 5000)
ans =  1.2703e+09
octave:16> ans - quadcc (f, 0, Inf)
ans = -1.0156


And notice that +Inf is acceptable as an endpoint as long as it can find
something to integrate.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?37896>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/