y=sin⁡(x2)y = \sin(x^2)

Graph of the Chirp y=sin⁡(x2)y = \sin(x^2)

y=sin⁡(x2)y = \sin(x^2) is a sine whose argument is squared. The further xx travels, the faster that argument changes, so the oscillations bunch ever closer together. A wave whose frequency rises as it goes is called a chirp, after the rising call of a bird1.

Domain and symmetry

The domain is all real numbers and the range is the interval [−1,1][-1, 1]. Since (−x)2=x2(-x)^2 = x^2 the function is even, and the graph is symmetric about the yy-axis, bunching up in the same way to the left. It is not periodic.

Instantaneous frequency

The phase is x2x^2, so its rate of change gives the angular frequency at each point.

ddxx2=2x\frac{d}{dx}x^2 = 2x

Since it grows in direct proportion to xx, this wave is called a linear chirp.

How the zeros crowd

We have sin⁡(x2)=0\sin(x^2) = 0 when x2=nπx^2 = n\pi, that is at x=nπx = \sqrt{n\pi}.

nnZero at x=nπx = \sqrt{n\pi}Gap from the previous zero
11≈1.7725\approx 1.7725≈1.7725\approx 1.7725
22≈2.5066\approx 2.5066≈0.7341\approx 0.7341
33≈3.0700\approx 3.0700≈0.5634\approx 0.5634
44≈3.5449\approx 3.5449≈0.4749\approx 0.4749
55≈3.9633\approx 3.9633≈0.4184\approx 0.4184

For large nn the gap between consecutive zeros shrinks roughly like π2n\dfrac{\sqrt{\pi}}{2\sqrt{n}}: a sharp initial contraction followed by a gradual one.

Near the origin

The derivative is y′=2xcos⁡(x2)y' = 2x\cos(x^2), which vanishes at x=0x = 0. Close to the origin the approximation sin⁡u≈u\sin u \approx u applies, so the curve is all but indistinguishable from a parabola.

sin⁡(x2)≈x2\sin(x^2) \approx x^2

Oscillation sets in only once xx has grown somewhat, the first crest appearing where x2=π2x^2 = \dfrac{\pi}{2}, at x≈1.253x \approx 1.253.

Fresnel integrals

This function has no elementary antiderivative; its integral defines the special function known as a Fresnel integral2. Taken out to infinity the value is as follows.

∫0∞sin⁡(x2) dx=π8≈0.6267\int_0^{\infty}\sin(x^2)\,dx = \sqrt{\frac{\pi}{8}} \approx 0.6267

The integrand does not tend to 00, yet the integral is finite, because the oscillations quicken and the positive and negative parts cancel: the convergence is conditional, not absolute. Plotting the two Fresnel integrals as coordinates traces the Cornu spiral, used in computing the diffraction of light.

Applications

Radar systems transmit a long chirp instead of a short powerful pulse and compress it on reception, obtaining fine range resolution without high transmit power3.

  • Pulse compression in radar and sonar
  • Improved resolution in medical ultrasound
  • The signal obtained by sweeping a gradient field in magnetic resonance imaging
  • The gravitational wave from a binary just before merger, itself called a chirp
  1. Chirp, Wikipedia
  2. Fresnel integral, Wikipedia
  3. Pulse compression, Wikipedia