Help in Bessel functions overflow in Matlab

Hello Everyone,

   I have a trouble about bessel function overflow in matlab. Does anyone have problem with bessel function overflow?

for example:

n=1078;

m1a= 431.2918;

x=(-1).*besselh(n,2,m1a);

than x return 4.9654e+286 -5.7074e+302i.

if n change to 1079;

than x return -inf.

Is there any solution to prevent the command "besselh" from overflowing? Because I want more data with higher n in besselh or besselj.

I really appreciate for your help,

Thank you,

Martin Chang

[e-mail]

Answer 1

[Walter Roberson]

You will need to switch to a higher precision toolbox such as the Symbolic Mathematics toolbox.

[martinyao]

Dear Roberson, Thanks for your answer.I have tried to use command "vpa" in Symbolic Mathematics toolbox. But the result is still inf.

For example,

n=1079;

m1a= 431.2918;

vpa x 500

x=(-1).*besselh(n,2,m1a);

x still return -inf.

Did I have a mistake in using the Symbolic Mathematics toolbox.? Or is it not your means?

Thank you , I really appreciate for your help.

Martin Chang [e-mail]

[Josh Meyer]

Yes, the answer is still -Inf because you are doing a numeric computation. Symbolic Math Toolbox doesn't have a besselh function, so you just need to manually construct it using besselj and bessely for the real and imaginary components.

Try something like this instead:

syms Z v
Z = sym('431.2918');
v = sym('1079');
y = -1.*(besselj(v,Z) - 1i*bessely(v,Z));
vpa(y)

ans =

- 1.1772358960337809504005397055303e-307 - 2.7337988321687724066318412501641e303i