Is this a bug in NIntegrate?

Question

Fixed in 10.1

Bug is present as of version 10.0.2 checked on windows 7, 64 bit

Is this a bug or I missed something? NIntegrate seems to give a different answer for the same integrand when the option Method -> {Automatic, "SymbolicProcessing" -> 0} is given.

$Version
(* "10.0 for Mac OS X x86 (64-bit) (June 29, 2014)" *)

$Assumptions = w ∈ Reals;
FEx[w_] = (I E^(-25 + 5 I w - w^2/4) Sinh[5 w])/Sqrt[2];
NIntegrate[Conjugate[FEx[w]]*FEx[w], {w, -∞, ∞}]
NIntegrate[Conjugate[FEx[w]]*FEx[w], {w, -∞, ∞}, 
 Method -> {Automatic, "SymbolicProcessing" -> 0}]
(* 0.313329 + 0. I *)
(* 0.626657 + 0. I *)

version 10

enter image description here

version 9

enter image description here

version 8

enter image description here

It seems to have worked in version 7.

version 7

enter image description here

score 3 · Accepted Answer · answered Sep 05 '14 at 23:23

3

Since it was Daniel Lichtblau who added the tag bugs, I would say the answer to your question is "yes".

answered Sep 05 '14 at 23:23

m_goldberg

107,779
16
103
257

Therefore, should be reported to Wolfram Technical Support. – rhermans Sep 06 '14 at 07:25
@rhermans. I think you can take that as done. – m_goldberg Sep 06 '14 at 09:48
Can you explain why? Is it in the sense that you trust somebody already did, you know somebody did, or is it that Wolfram Technical Support actually participates in here? – rhermans Sep 06 '14 at 11:07
4

@rhermans. Google "Daniel Lichtblau". – m_goldberg Sep 06 '14 at 21:20

Is this a bug in NIntegrate?

1 Answers1

Linked