Using Simpson's method to integrate a complex function

[amjad-sh]

Homework Statement

The question includes programming, but the basic problem is related to mathematics, that's why I posted it here.

In fact, I'm working on a condensed matter physics paper,where I need to draw integral

$g(\nu,2k_fz)$ in units of $(2k_f)^{-1}$ and interms of z.

$g(\nu,2k_fz)=\int_{1}^{-1}\dfrac{\nu^2(1-u^2)usin(2k_fzu)}{u^2-\nu^2/2+u\sqrt{u^2-\nu^2}}\, du$, where $\nu=0.5$



I'm using Simpson's method to evaluate this integral. The problem is that the sign inside the the radical in the denominator will be negative while evaluating this method, that's why we need to treat the integrand as a complex function. The program(python) will plot for me just the real part of the result.

But my main question is: Is it Ok to use Simpson's method in order to evaluate integrals that their integrand is a complex function?

because the graph I'm getting is not even close to the graph depicted in the paper.



Is contour integration a better choice?

Relevant Equations

Simpson's method :

 

[DrClaude]

But my main question is: Is it Ok to use Simpson's method in order to evaluate integrals that their integrand is a complex function?

Yes, that shouldn't be a problem.

 

[amjad-sh]

Yes, that shouldn't be a problem.

Yes, it was a tiny mistake in my code that was stopping the graph to appear as it should.
But if we consider the problem in case $\nu=0$,a singularity will arise while applying Simpson's method for u=0.
Do you know a trick or a way to go around this singularity?

 

[DrClaude]

quadpack has routines to deal with integration of functions with singularities. It has been ported to C in GSL, but I don't think it is available in python.

 

[cpt_carrot]

quadpack has routines to deal with integration of functions with singularities. It has been ported to C in GSL, but I don't think it is available in python.

Scipy provides an interface to quadpack via the integrate.quad function

 

[DrClaude]

Scipy provides an interface to quadpack via the integrate.quad function

I don't use python, but from the information I could find, only a basic quadrature method from quadpack is implemented in scipy, not the routines for dealing with singularities.

 

[cpt_carrot]

I don't use python, but from the information I could find, only a basic quadrature method from quadpack is implemented in scipy, not the routines for dealing with singularities.

By default I think quad calls qagse which does support singularities, you can also specify a series of domains to avoid them. With the right set of options you can get to most of the quadpack routines.

 

[DrClaude]

By default I think quad calls qagse which does support singularities, you can also specify a series of domains to avoid them. With the right set of options you can get to most of the quadpack routines.

You're right. I stand corrected.

So, @amjad-sh, you can use scipy.integrate.quad and specify the position of the singularity.
https://docs.scipy.org/doc/scipy-0.14.0/reference/generated/scipy.integrate.quad.html