2
$\begingroup$

I need some help in evaluating the following limit with assumptions:

Limit[Result, L -> Infinity, Assumptions -> A ∈ Integers && A > 0 && 
    B ∈ Integers && B > 0 && C ∈ Integers && C > 0 && D > 0 && F > 0]

where

Result := C D (B F)^(A + L) (-C D + B F)^-A 
     ((C D + B F)^(-1 - L) - ((-C D + B F)^(-1 - L)
     Gamma[1 + A + L] Hypergeometric2F1[1 + L, 1 + A + L, 
     2 + L, -((C D + B F)/(-C D + B F))])/(Gamma[A] Gamma[2 + L]))

It should come out to 0, and I am sure of this. I just need to verify.

The reason is that I plugged in values for all of the parameters (except L), and then summed over Result from L=0 to L=100000 - the sum converges to 1.

Mathematica is unable to evaluate this limit as I defined it. Is there a proper way?

Thanks in advance for any help.

$\endgroup$
6
  • 3
    $\begingroup$ Did you type this correctly? When I simplify this I get C D (B F)^(A+L) (-C D+B F)^-A. BTW, D is a predefined keyword for derivative. You may use some other variable. $\endgroup$ Commented Jan 18, 2017 at 3:48
  • $\begingroup$ I think I did type correctly... what did you simplify to get that !? $\endgroup$
    – user10189
    Commented Jan 18, 2017 at 3:50
  • $\begingroup$ I just used Simplify[Result]. $\endgroup$ Commented Jan 18, 2017 at 3:51
  • $\begingroup$ Strange. I did not get that. I have version 10.4. $\endgroup$
    – user10189
    Commented Jan 18, 2017 at 3:52
  • 1
    $\begingroup$ I noticed that in the code you have posted, there is a new line in the Result. This is the reason why it gives different results. $\endgroup$ Commented Jan 18, 2017 at 4:04

1 Answer 1

2
$\begingroup$

It is convenient to replace C by a B F/D, where a > 0. Also, specify that L is positive.

Simplify[Result /. C -> a B F/D, B > 0 && F > 0 && a > 0 && L > 0]
(* (1 - a)^-A a B F (B F)^L (((1 + a) B F)^(-1 - L) - ((-(-1 + a) B F)^(-1 - L)
   Gamma[1 + A + L] Hypergeometric2F1[1 + L, 1 + A + L, 2 + L, 
   (1 + a)/(-1 + a)])/(Gamma[A] Gamma[2 + L])) *)

It is not uncommon for Limit to fail when dealing with hypergeometric functions. So, use Series instead.

Series[%, {L, Infinity, 1}, Assumptions -> B > 0 && F > 0 && L > 0] // Normal
(* ((1 - a)^-A a (B F)^L ((1 + a) B F)^-L)/(1 + a) *)
FullSimplify[%, B > 0 && F > 0 && L > 0 && a > 0]
(* (1 - a)^-A a (1 + a)^(-1 - L) *)
Limit[%, L -> Infinity, Assumptions -> a > 0]
(* 0 *)

as desired.

$\endgroup$
1
  • $\begingroup$ This is simply brilliant. My next step was to compute the series anyway! $\endgroup$
    – user10189
    Commented Jan 18, 2017 at 6:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.