3
$\begingroup$

I have two integrals that I suspect can be expressed as elliptic integrals:

$$\int_0^{2\pi} d\phi' \frac{1}{( 1 + \alpha^2 + \beta^2 - 2 \beta \cos(\phi') )^{3/2}} $$ $$\int_0^{2\pi} d\phi' \frac{\cos(\phi')}{( 1 + \alpha^2 + \beta^2 - 2 \beta \cos(\phi') )^{3/2}}$$

Is there a way to have Mathematica attempt to transform these into elliptic integrals?

$\endgroup$

3 Answers 3

3
$\begingroup$

When doing the indefinite integrals and taking the limits, you get the results very fast.

int1[p_, a_, b_] = Integrate[1/(1 + a^2 + b^2 - 2 b Cos[p])^(3/2), p, 
           Assumptions -> 
               a \[Element] Reals && b \[Element] Reals && 0 <= p <= 2 Pi]

lim1t = Limit[int1[p, a, b], p -> 2 Pi, Direction -> 1, 
           Assumptions -> a \[Element] Reals && b \[Element] Reals]

(*    (4 EllipticE[-((4 b)/(a^2 + (-1 + b)^2))])/(Sqrt[
        a^2 + (-1 + b)^2] (a^2 + (1 + b)^2))    *)

lim1b = Limit[int1[p, a, b], p -> 0, Direction -> -1, 
           Assumptions -> a \[Element] Reals && b \[Element] Reals]

(*    0    *)

int2[p_, a_, b_] = Integrate[Cos[p]/(1 + a^2 + b^2 - 2 b Cos[p])^(3/2), p, 
                     Assumptions -> a \[Element] Reals && b \[Element] Reals]

lim2t = Limit[int2[p, a, b], p -> 2 Pi, Direction -> 1, 
            Assumptions -> a \[Element] Reals && b \[Element] Reals]

(*    (2 (1 + a^2 + b^2) EllipticE[-((4 b)/(a^2 + (-1 + b)^2))] - 
       2 (a^2 + (1 + b)^2) EllipticK[-((4 b)/(a^2 + (-1 + b)^2))])/(Sqrt[
       a^2 + (-1 + b)^2] b (a^2 + (1 + b)^2))    *)

lim2b = Limit[int2[p, a, b], p -> 0, Direction -> -1, 
            Assumptions -> a \[Element] Reals && b \[Element] Reals]

(*    0    *)

So the definite integrals are lim1t and lim2t.

$\endgroup$
2
$\begingroup$

In version 12.0

Integrate[1/(1 + \[Alpha]^2 + \[Beta]^2 - 2 \[Beta] Cos[\[Phi]])^(3/
 2), {\[Phi], 0, 2*Pi}, GenerateConditions -> False]

$$\frac{4 E\left(-\frac{4 \beta }{\alpha ^2+(\beta -1)^2}\right)}{\sqrt{\alpha ^2+(\beta -1)^2} \left(\alpha ^2+(\beta +1)^2\right)} $$

Integrate[Cos[\[Phi]]/(1 + \[Alpha]^2 + \[Beta]^2 - 2 \[Beta] Cos[\[Phi]])^(3/
 2), {\[Phi], 0, 2*Pi}, GenerateConditions -> False]

$$\frac{2 \left(\alpha ^2+\beta ^2+1\right) E\left(-\frac{4 \beta }{\alpha ^2+(\beta -1)^2}\right)-2 \left(\alpha ^2+(\beta +1)^2\right) K\left(-\frac{4 \beta }{\alpha ^2+(\beta -1)^2}\right)}{\beta \sqrt{\alpha ^2+(\beta -1)^2} \left(\alpha ^2+(\beta +1)^2\right)} $$

$\endgroup$
0
$\begingroup$

I was able to do this after a manual transformation (related to the original integrals by a factor of $2 (\alpha^2 + (1-\beta)^2)^{3/2}$) :

$Assumptions = 
  beta >= 0 && alpha \[Element] Reals && (1 - beta)^2 + alpha^2 != 0;

i1 = Integrate[ ( 
    1 + 4 beta Sin[u]^2/((1 - beta)^2 + alpha^2))^(-3/2), {u, 0, Pi}]

i2 = Integrate[ 
  Cos[2 u] ( 1 + 4 beta Sin[u]^2/((1 - beta)^2 + alpha^2))^(-3/2), {u,
    0, Pi}]

and then waiting a very long time (around an hour). The results comes back as not very illuminating sums of EllipticE and EllipticK functions.

Perhaps Mathematica would have given me a similar result for Integrate[]'s of the original integrands? However, I was too impatient, and Aborted the evaluation, thinking that it was hung.

$\endgroup$

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.