Ma · Mathematics

Expressing Sine and Cosine as Integrals of Bessel Functions

Published By the 9to5Science editorial team

At a glance

  1. 1. Choose half-integer orders

    Use the explicit formulas for the Bessel functions J₁/₂ and J₋₁/₂.

  2. 2. Apply a square-root weight

    Multiplying each function by the square root of its argument leaves a sine or cosine.

  3. 3. Integrate from zero

    Integrate the resulting elementary function to obtain sine or one minus cosine.

  4. 4. Check the domain

    The real formulas below use x ≥ 0; use parity to handle negative x.

Sine and cosine can be written as definite integrals involving Bessel functions. The convenient orders are −12-\tfrac12 and 12\tfrac12, because Bessel functions of these orders reduce to cosine and sine multiplied by a simple power of their argument.

For x≥0x\geq 0, the identities are

sin⁡x=π2∫0xt J−1/2(t) dt,\sin x=\sqrt{\frac{\pi}{2}}\int_0^x \sqrt{t}\,J_{-1/2}(t)\,dt,

and

cos⁡x=1−π2∫0xt J1/2(t) dt.\cos x=1-\sqrt{\frac{\pi}{2}}\int_0^x \sqrt{t}\,J_{1/2}(t)\,dt.

Here JνJ_\nu denotes the Bessel function of the first kind of order ν\nu. The factor t\sqrt{t} matters: these are not formulas for the unweighted integrals ∫0xJν(t) dt\int_0^x J_\nu(t)\,dt.

Why half-integer orders work

For positive tt, the two relevant Bessel functions have elementary forms:

J−1/2(t)=2πtcos⁡t,J1/2(t)=2πtsin⁡t.J_{-1/2}(t)=\sqrt{\frac{2}{\pi t}}\cos t, \qquad J_{1/2}(t)=\sqrt{\frac{2}{\pi t}}\sin t.

Multiplication by t\sqrt{t} cancels the factor 1/t1/\sqrt{t}:

t J−1/2(t)=2πcos⁡t,t J1/2(t)=2πsin⁡t.\sqrt{t}\,J_{-1/2}(t)=\sqrt{\frac{2}{\pi}}\cos t, \qquad \sqrt{t}\,J_{1/2}(t)=\sqrt{\frac{2}{\pi}}\sin t.

The desired identities then follow by ordinary integration. In particular, the integrands after multiplication by t\sqrt t remain finite as tt approaches zero.

Deriving the formulas

For the sine identity, substitute the expression for J−1/2J_{-1/2} into the integral:

∫0xt J−1/2(t) dt=2π∫0xcos⁡t dt=2πsin⁡x.\begin{aligned} \int_0^x \sqrt{t}\,J_{-1/2}(t)\,dt &=\sqrt{\frac{2}{\pi}}\int_0^x \cos t\,dt\\ &=\sqrt{\frac{2}{\pi}}\sin x. \end{aligned}

Multiplying both sides by π/2\sqrt{\pi/2} gives the stated formula.

For cosine, use the order 1/21/2 instead:

∫0xt J1/2(t) dt=2π∫0xsin⁡t dt=2π(1−cos⁡x).\begin{aligned} \int_0^x \sqrt{t}\,J_{1/2}(t)\,dt &=\sqrt{\frac{2}{\pi}}\int_0^x \sin t\,dt\\ &=\sqrt{\frac{2}{\pi}}(1-\cos x). \end{aligned}

Rearranging gives the expression for cos⁡x\cos x.

Worked example: evaluate the sine integral at x=π/2x=\pi/2

Set x=π/2x=\pi/2 in the first identity. Since sin⁡(π/2)=1\sin(\pi/2)=1,

∫0π/2t J−1/2(t) dt=2π.\int_0^{\pi/2}\sqrt{t}\,J_{-1/2}(t)\,dt =\sqrt{\frac{2}{\pi}}.

You can also check directly using the elementary form of the Bessel function:

∫0π/2t J−1/2(t) dt=2π∫0π/2cos⁡t dt=2π.\int_0^{\pi/2}\sqrt{t}\,J_{-1/2}(t)\,dt =\sqrt{\frac{2}{\pi}}\int_0^{\pi/2}\cos t\,dt =\sqrt{\frac{2}{\pi}}.

The same check for the cosine formula gives

∫0π/2t J1/2(t) dt=2π,\int_0^{\pi/2}\sqrt{t}\,J_{1/2}(t)\,dt=\sqrt{\frac{2}{\pi}},

so 1−π/2 2/π=0=cos⁡(π/2)1-\sqrt{\pi/2}\,\sqrt{2/\pi}=0=\cos(\pi/2).

What changes for negative x?

The displayed real integrals use t\sqrt{t}, so they are written for x≥0x\geq0. For x<0x<0, recover the values from parity: sin⁡(−x)=−sin⁡x\sin(-x)=-\sin x and cos⁡(−x)=cos⁡x\cos(-x)=\cos x. Equivalently, apply the integral formulas to ∣x∣|x| and then use those parity rules.

Common mistakes

  • Dropping the weight: The formulas involve t Jν(t)\sqrt{t}\,J_\nu(t), not just Jν(t)J_\nu(t). Removing the weight changes the integral.
  • Swapping the orders: J−1/2J_{-1/2} is proportional to cos⁡t\cos t, while J1/2J_{1/2} is proportional to sin⁡t\sin t.
  • Forgetting the constant in the cosine formula: Integrating sine from 00 to xx gives 1−cos⁡x1-\cos x, so the result must be rearranged as cos⁡x=1−⋯\cos x=1-\cdots.
  • Ignoring the lower limit: The choice of zero makes the sine integral start at zero and fixes the constant in the cosine identity.

These formulas are useful when a problem specifically calls for a Bessel-function representation. They are direct consequences of the elementary half-integer Bessel formulas, rather than special evaluations of general unweighted Bessel integrals.

References

  1. NIST Digital Library of Mathematical Functions, §10.16: Relations to Other Functions

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.