The is the for functions on (0,∞), with rescaling t↦γt in the role of translation. This chapter states the definition and the two properties that the rest of Part II uses.
Standard material. The paper uses the transform throughout and defines it on page 17.
Definition
For a function f on (0,∞) and a s,
M(f)(s)=∫0∞f(t)tstdt
The substitution t=eu turns this into ∫f(eu)esudu over the whole real line. For s=iy that is the Fourier transform of f in the variable u=lnt. If f is zero outside a bounded interval [A,B] with A>0, the converges for every complex s, and M(f) is an .
Property 1: rescaling becomes multiplication
The transform of f(γt) is γ−s⋅M(f)(s)
Proof: substitute u=γt. The measure dt/t does not change, and ts=γ−sus. Consequently on (0,∞), which is a weighted sum of rescalings, becomes multiplication of transforms: M(A(φ)f)(s)=M(φ)(s)⋅M(f)(s) for the A(φ) of chapter 10.
Property 2: the real part of s weighs large t against small t
Write s=σ+iy. Then ∣ts∣=tσ. For σ>0 this factor is large where t>1 and small where t<1. So if f≥0 is zero outside [A,B],
Aσ⋅M(f)(0)≤M(f)(σ)≤Bσ⋅M(f)(0)for σ≥0
A function concentrated at t>1 has a transform that grows as σ increases, and one concentrated at t<1 has a transform that shrinks.
Lab 11.1 the transform of a bump along the vertical line Re s = σ
the bump f(t)dashed line: t = 1
|M(f)(iy)||M(f)(σ + iy)|
Both curves are computed by numerical integration. Moving the bump to the right of t=1 makes the orange curve larger than the blue one for σ>0 and smaller for σ<0; moving it to the left of 1 reverses this.
Things to try
Center the bump at t=1. The two curves nearly coincide for every σ: the factor tσ is close to 1 across the bump.
Narrow the bump. Both curves spread out to larger y: a narrow function has a wide transform.
Put the bump at t≈3 and compare the readout with 3σ.
Exercises
Let f(t)=1 for 1≤t≤T and 0 otherwise. Compute M(f)(s).Solution
∫1Tts−1dt=sTs−1(s=0),lnT(s=0)
For which s does rescaling by γ=2 leave the transform unchanged?Solution
The factor is 2−s, which equals 1 when sln2=2πim, that is, s=ln22πim for an integer m. These are the of the circle in chapter 10.
Show that the transform of t⋅f′(t) is −s⋅M(f)(s), for smooth f that is zero outside a bounded interval [A,B] with A>0.Solution
∫0∞tf′(t)tstdt=∫0∞f′(t)tsdt=−s∫0∞f(t)ts−1dt=−s⋅M(f)(s) The middle step is integration by parts; the boundary terms vanish because f is zero outside [A,B].
Check your understanding
Why is the measure dt/t used in place of dt?
It makes every integral converge
It is unchanged by rescaling t↦γt, which is what makes rescaling turn into multiplication by γ−s
It is just a convention with no consequences
With u=γt, du/u=dt/t. With plain dt an extra factor γ would appear. In the coordinate lnt it is ordinary length, and rescaling is translation.
A function f≥0 is zero outside [0.1,0.5]. As σ increases from 0, M(f)(σ):
shrinks, because tσ<1 where f lives
grows
stays the same
By property 2 the value lies between 0.1σ and 0.5σ times M(f)(0). Functions living at t>1 behave the opposite way, which matters later because the proof uses test functions supported at t>1.
You make the bump narrower. Its transform along a vertical line:
becomes narrower too
is unchanged
spreads out to larger heights y
A function concentrated in a short range of lnt needs high frequencies to describe it. Later this means a narrow test function "sees" zeros at greater heights.