Problem in integration function containing gauss hypergeometric function
Show older comments
I am deriving the coverage probability of heterogeneous network with massive MIMO macro base station. During the process, I am stuck in the in hypergeometric function. The small part of that program is given below.
fun= @(x,y)imag(exp(pi.*1i.*(x).*y.^(-2).*hypergeom(1,[1/2,3/2],-1i.*x./(y.^4))))./x
c= @(y) integral(@(x) fun(x,y), 0.1,1,'ArrayValued',true)
fun1= @(y)((1/2)-((1/pi).*c(y))).*exp(-pi.*y.^2)
c1 = integral(fun1, 0.1,1)
I think iproblem is due to terms(y.^(-2) and and y.^4) in the fun. If any body have any idea please suggest. Any suggestions regarding this will be appreciated.
7 Comments
Jan
on 14 Dec 2017
You forgot to mention, what the problem is.
Shweta Rajoria
on 8 Jan 2018
Walter Roberson
on 8 Jan 2018
Do you mean that it is crashing, or that it is reporting that it did not converge, or that it is taking a long time to compute?
Jan
on 8 Jan 2018
@Shweta Rajoria: You still did not explain, what the problem is, but you mention only, that there is one. What do you observe and what do you want to change?
Shweta Rajoria
on 9 Jan 2018
Walter Roberson
on 9 Jan 2018
Edited: Walter Roberson
on 9 Jan 2018
To within the boundaries of double precision, c1 will be nan. The values involved are way way too big to work with in double precision.
Shweta Rajoria
on 9 Jan 2018
Answers (1)
Walter Roberson
on 8 Jan 2018
0 votes
fun(8/10, 3/10) is about 2.974*10^1912918 .
fun(8/10, 2/10) is about 4.087*10^-24618734963806
fun(8/10, 0.18) is about 2.345*10^207906652598325658
So, fun is very steep, and has a lot of values that overflow floating point representation. You will not be able to integrate that using any normal representation.
Categories
Find more on Numerical Integration and Differentiation in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!