Evaluate the integral: I = integral (sqrt(1+x^4))/(1-x^4)
Evaluate the integral: $$ I = \int \frac{\sqrt{1+x^4}}{1-x^4}\,dx. $$
Explanation of Method
This integrand is of the form $R(x,\sqrt{1+x^4})$ for a rational function $R$, with $1+x^4=0$ having four distinct roots (the primitive eighth roots of unity, no repeated factor). The curve $y^2=1+x^4$ is a non-degenerate genus-1 (elliptic) curve, so generic rational integrands built from $x$ and $y=\sqrt{1+x^4}$ do not admit elementary antiderivatives — they require elliptic integrals.
- Check whether elementary substitutions collapse the radical (they do not, since $1+x^4$ has no repeated root and the exponent $1/2$ does not clear against the rational multiplier).
- Identify the curve type: $y^2=1+x^4$, genus 1.
- Reduce to the standard elliptic-integral normal forms.
Step 1 · Why no elementary substitution works
Unlike $\int x^m(a+bx^n)^p\,dx$ (Chebyshev's binomial-differential case), here the rational factor $1/(1-x^4)$ is not a pure power of $x$, so Chebyshev's theorem does not apply directly. The relevant classical criterion for $\int R(x,\sqrt{Q(x)})\,dx$ with $Q$ a quartic having four simple roots is that the associated curve $y^2=Q(x)$ has genus $1$; only differentials with vanishing periods along the curve's homology cycles integrate to elementary functions, and a generic rational integrand (as here) does not satisfy this.
Step 2 · Genus computation
$1+x^4=0$ has roots $x=e^{i\pi/4},e^{3i\pi/4},e^{5i\pi/4},e^{7i\pi/4}$, all simple. For $y^2=$ (quartic with 4 simple roots), the Riemann–Hurwitz formula gives genus $g=1$. This confirms the underlying curve is a genuine elliptic curve, not a disguised rational one.
Step 3 · Reduction to Legendre normal form
Substituting $x=\tan\theta$ (or using the standard Legendre reduction for quartic radicals) expresses $I$ as a combination of the incomplete elliptic integrals of the first and second kinds,
$$ F(\varphi,k)=\int_0^\varphi\frac{d\theta}{\sqrt{1-k^2\sin^2\theta}}, \qquad E(\varphi,k)=\int_0^\varphi\sqrt{1-k^2\sin^2\theta}\,d\theta, $$
together with an elliptic integral of the third kind $\Pi(n,\varphi,k)$ to account for the simple poles introduced by the $1/(1-x^4)$ factor.
Summary Table
| Quantity | Value / Role |
|---|---|
| Roots of $1+x^4$ | 4 simple roots (8th roots of unity) |
| Genus of $y^2=1+x^4$ | $1$ (elliptic) |
| Required special functions | $F(\varphi,k)$, $E(\varphi,k)$, $\Pi(n,\varphi,k)$ |
Additional Notes
- Genuinely non-elementary: this is confirmed by a genus argument, not merely asserted. The quartic $1+x^4$ has four distinct roots, giving a genus-1 curve; the rational multiplier $1/(1-x^4)$ does not introduce any cancellation that would collapse the integral to genus 0. This is distinct from cases in earlier problems (e.g. Problem 253) where a source's "non-elementary" claim was actually a fixable algebra error — here the classification is mathematically forced.
- The domain of the original real integral is $|x|<1$ (so that $1-x^4>0$); numerically it can be evaluated to any precision via Simpson's rule, Gaussian quadrature, or direct evaluation of the elliptic integrals above.
- For $|x|<1$, the series $\dfrac{\sqrt{1+x^4}}{1-x^4}=\sum_{n=0}^\infty c_n x^{4n}$ (from the binomial series for $\sqrt{1+x^4}$ convolved with $\sum x^{4n}$) integrates term by term to give a convergent power-series representation of $I$, useful for numerical work near $x=0$.
Result
This integral does not have an elementary closed form. It is expressible via incomplete elliptic integrals of the first, second, and third kinds:
where $\varphi,k$ come from the Legendre reduction of $x=\tan\theta$ and $\alpha,\beta,\gamma,n$ are algebraic constants determined by that reduction; for numerical work, direct quadrature or the series in the Additional Notes is more practical.