Good morning, i computed the following integral using Integrate and i get a different result from the one obtained with NIntegrate. Here is the integrand
expr = (b (b - 2 d + Sqrt[4 + b^2 - 4 b d]) (c + I x) x (3 + b^2 c^2 + 2 I b^2 c x - (-6 + b^2) x^2 + 3 x^4) (-1 + a x (-2 I c + a x)))/((2 + b^2 + 2 d (d - Sqrt[4 + b^2 - 4 b d]) + b (-4 d + Sqrt[4 + b^2 - 4 b d])) \[Pi]^3 (-I c + a x) (1 - b c -I b x + x^2)^2 (1 + b c + I b x + x^2)^2 (1 + a^2 x^2));`
I use Integrate
result = Integrate[expr, {x, -Infinity, Infinity}, Assumptions -> 1/2 (b + Sqrt[4 + b^2 - 4 b d]) > -1 && 1/2 (b + Sqrt[4 + b^2 - 4 b d]) < 1 && c > -1 && c < 1 && d > 1 &&
a >= 0]
and then i compare the results
result/.{a->2,b->0.65,c->0.25,d->1.7} 0.341887 +0. I NIntegrate[expr/.{a->2,b->0.65,c->0.25,d->1.7},{x,-Infinity,Infinity}] -0.00340441-8.67362*10^-19 I
As you can see i get different results even if the assumptions are satisfied. Why?
FullSimplify
, remove the constant pre-factor and use theRubi
package in order to integrate analytically. $\endgroup$