4
$\begingroup$

I have an expression defined as

Ν[Δ_, l_] = FunctionExpand[((2^l (Δ + l - 1) Gamma[Δ + l - 1]^2 Gamma[Δ])/
              (Gamma[Δ - 1] Gamma[(Δ + l)/2]^4 Gamma[1/8 - ((Δ - l)/2)]^2 *
               Gamma[1/8 - ((2 - Δ - l)/2)]^2)), 
             Assumptions -> {Δ ∈ Integers, l ∈ Integers, Δ >= 0, l >= 0}];

The problem is that it diverges naively when I try to compute Ν[1,0] with the divergence coming from the fact that Mathematica evaluates Ν[k,0] to be

Ν[k, 0]
(* (2^(-4 + 2 k) (-1 + k)^2 Gamma[-(1/2) + k/2]^2)/(π Gamma[
    1/8 - k/2]^2 Gamma[-(7/8) + k/2]^2 Gamma[k/2]^2) *)

Which means that the numerator becomes a product 0 ComplexInfinity. But when I use FullSimplify on the above, I get

Ν[k, 0] // FullSimplify
(* (4^(-1 + k) Gamma[(1 + k)/2]^2)/(π Gamma[
      1/8 - k/2]^2 Gamma[-(7/8) + k/2]^2 Gamma[k/2]^2) *)

Is there a way to make Mathematica do this simplification by default? I am using this function in a sum, so I have to sum over $l = 0$ terms too, which is causing annoying errors.

$\endgroup$

1 Answer 1

4
$\begingroup$

How about taking the limit instead:

f[kin_, Lin_] := Module[{num, den, L, k, t},
  num = Limit[(2^(-4 + 3*L + 2*k)*(-1 + k)*(-1 + L + k)*
      Gamma[-1/2 + L/2 + k/2]^2), {L -> Lin, k -> kin}];
  den = Limit[(Pi*Gamma[1/8 + L/2 - k/2]^2*Gamma[-7/8 + L/2 + k/2]^2*
      Gamma[L/2 + k/2]^2), {L -> Lin, k -> kin}];
  t = num/den;
  FunctionExpand[t, 
   Assumptions -> {k ∈ Integers, L ∈ Integers, 
     k >= 0, L >= 0}
   ]
  ]

And now

f[1, 0]
(* 0 *)

And

 f[k, 0] // FullSimplify

Mathematica graphics

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.