---
title: "Images, gains and delays of a Gaussian plasma lens"
description: "Solve the lens equation for one, two or three images, find their gains and arrival times, and see when interference fringes replace separate copies."
canonical_url: "https://sandboxphysics.com/stories/fast-radio-burst-echoes/gaussian-plasma-lens-images/"
language: "en"
page_type: "story"
alternate_languages:
  zh-CN: "https://sandboxphysics.com/zh/stories/fast-radio-burst-echoes/gaussian-plasma-lens-images/"
last_modified: "2026-11-14"
---

[Fast radio bursts: repeats, echoes and plasma lenses](https://sandboxphysics.com/stories/fast-radio-burst-echoes/)

# Images, gains and delays of a Gaussian plasma lens

Sandbox Physics · 2026-11-14

A Gaussian plasma lens gives one image of a burst, or three. Three appear only when the lens strength exceeds $e^{3/2}/2\approx2.24$ and the source sits in a band of offsets just outside the lens core. Each image has its own gain and delay; together the gains average to one, so the bright spikes at the band edges pay for the deep shadow behind the core. When two copies overlap in time, they interfere and stripe the spectrum with fringes spaced by one over their delay.

Before this chapter, read [how a clump of plasma delays and bends a radio burst](https://sandboxphysics.com/stories/fast-radio-burst-echoes/plasma-delay-and-bending/). It derives the lens equation $u(1+\alpha e^{-u^2})=u'$ and the delay $t(u)$ used here.

## When one image becomes three

The images are the solutions $u$ of $f(u)=u'$ with $f(u)=u\,(1+\alpha e^{-u^2})$. If $f$ only ever increases, each $u'$ has one solution. Three solutions need a stretch where $f$ decreases, that is, where

$$
f'(u)=1+\alpha\,(1-2u^2)\,e^{-u^2}<0 .
$$

The function $(1-2u^2)e^{-u^2}$ has its minimum where its derivative $-2u\,(3-2u^2)\,e^{-u^2}$ vanishes, at $u^2=3/2$, with value $-2e^{-3/2}$. So $f'$ can go negative only if

$$
\alpha>\alpha_{\min}=\frac{e^{3/2}}{2}\approx2.24 .
$$

The points where $f'=0$ are caustics. For the Lab default, $\alpha=6.35$, they sit at $u=0.807$ and $u=1.925$, which map to offsets $u'=3.48$ and $u'=2.23$. A source with $2.23<|u'|<3.48$ has three images; any other offset has one. Cordes and colleagues show the same picture in their Fig. 1 and quote the caustic condition as $\mathrm{DM}_\ell\,d_{\rm sl}/a^2\gtrsim0.65\,\mathrm{pc^2\,AU^{-2}\,cm^{-3}}$ at 1 GHz with $d_{\rm sl}$ in parsecs, which is the same as $\alpha>2.24$. [Lensing of fast radio bursts by plasma structures in host galaxies](https://arxiv.org/abs/1703.06580v1)

## Gain of each image

A narrow strip of source offsets $\mathrm du'$ maps onto a strip $\mathrm du$ of the lens plane. The flux of the image is multiplied by the ratio of their widths:

$$
G=\left|\frac{\mathrm du}{\mathrm du'}\right|=\frac{1}{\left|1+\alpha\,(1-2u^2)\,e^{-u^2}\right|} .
$$

The paper prints this as its Eq. 4 without the absolute value and as Eq. A5 with it; a gain is positive, so we use Eq. A5. Images where the bracket is negative are flipped, and in wave optics they carry an extra phase of $-\pi/2$.

Two limits are worth knowing. Far from the lens, $u\to u'$ and $G\to1$. Directly behind the core, $u'=0$ has the single image $u=0$ with

$$
G(0)=\frac{1}{1+\alpha},
$$

so the default lens dims a source behind its core to $1/7.35=0.136$ of its flux. At a caustic the denominator vanishes and geometric optics predicts infinite gain. Diffraction caps it near $a/r_{\rm F}$, where $r_{\rm F}=\sqrt{\lambda D/2\pi}$ is the Fresnel scale.

## Where the flux goes

A lens made only of a phase delay cannot create or destroy energy. Average the summed gain over all source offsets:

$$
\int\sum_j G_j\,\mathrm du'=\int\mathrm du .
$$

Each lens-plane point $u$ is the image of exactly one offset, so integrating over every image of every offset counts each $u$ once. Over a wide range the average of $\sum_j G_j$ is therefore one. The shadow behind the core is balanced by the extra flux near the caustics. Our independent wave-optics script checks this without ray tracing: for $\alpha=5$ the mean of the full diffraction intensity over offsets from $-8$ to $8$ is 1.0000000 to seven digits.

## Delays and dispersion of each copy

With $u'-u=\alpha u\,e^{-u^2}$ from the lens equation and $t_{g0}\alpha=t_{\rm DM0}$, the delay at an image is

$$
t(u)=t_{\rm DM0}\,e^{-u^2}\left(1+\alpha u^2 e^{-u^2}\right),
$$

Eq. 15 of the 2017 paper. Each copy also crosses its own electron column, $\mathrm{DM}_\ell\,e^{-u^2}$, so the copies have different apparent dispersion measures.

**Worked example.** Take the Lab default ($\alpha=6.35$, $t_{\rm DM0}=6.91\,\mathrm{ms}$) with the source at $u'=2.30$. The three images are at $u=2.181$, $1.700$ and $0.347$. For the last one, $e^{-u^2}=0.887$ and $\alpha u^2e^{-2u^2}=0.601$, so $t=6.91\times(0.887+0.601)=10.3\,\mathrm{ms}$. Its bracket is $1+6.35\times0.759\times0.887=5.28$, so $G=0.190$, and it crosses $0.6\times0.887=0.532\,\mathrm{pc\,cm^{-3}}$. The other two arrive at 0.075 and 0.776 ms with gains 1.87 and 1.46. The copies spread over 10.2 ms and their apparent dispersion measures differ by 0.527.

## Copies change with frequency

Both $\alpha$ and $t_{\rm DM0}$ scale as $\nu^{-2}$. For the same geometry at 800 MHz, $\alpha=3.57$ and the three copies arrive at 0.028, 3.05 and 4.59 ms. At 400 MHz, $\alpha=14.3$; the caustic band has moved outward and the source now sits in the shadow, with one copy of gain 0.07 at 20.2 ms. A lens therefore brightens some frequencies and dims others, and its echoes come later at low frequency. That is why lensed bursts can look narrow-band and can drift downward in frequency.

## When copies overlap: fringes

If two copies arrive closer together than the burst width, their electric fields add before a detector squares them. For copies with gains $G_1$ and $G_2$ and delay difference $\Delta t$,

$$
|E|^2\propto G_1+G_2+2\sqrt{G_1G_2}\,\cos(2\pi\nu\,\Delta t+\Delta\varphi),
$$

so the spectrum is striped with fringes spaced by $1/\Delta t$. In the worked example the two bright copies are $0.70\,\mathrm{ms}$ apart, giving fringes 1.43 kHz apart. In the Lab's microsecond preset they are 2.35 µs apart and the fringes are 426 kHz apart. Cordes and colleagues discuss this frequency structure in their Sec. 2.6. Hessels and colleagues point out that overlapping images give interference structure on small time and frequency scales, qualitatively like the structure they see in bursts of FRB 121102. [FRB 121102 bursts show complex time-frequency structure, Sec. 4.3](https://arxiv.org/abs/1811.10748v1)

Geometric optics treats each copy as a ray. It holds when the Fresnel scale is much smaller than the lens. For the default lens $r_{\rm F}/a=3.5\times10^{-4}$. Our wave-optics script evaluates the full diffraction integral and compares it with the ray sum. At $\alpha=5$, $u'=2.6$, they differ by 3 percent at $r_{\rm F}/a=0.05$ and by 0.007 percent at $0.01$. A Fourier transform of the diffracted field across a band recovers copy delays of 0.279 and 3.937 µs, against 0.287 and 3.937 µs from the delay formula, within the 0.125 µs resolution.

## Lab tasks

1. In [the plasma-lens echo Lab](https://sandboxphysics.com/labs/frb-plasma-lens-echoes/), start from the default and drag the source slowly toward the lens. Note the offset at which two copies brighten together and then vanish.
2. Choose the shadow preset and compare the gain readout with $1/(1+\alpha)$.
3. Change the observing frequency from 800 to 400 MHz without moving anything. Count the copies at each end and explain the change with the caustic band.

## Exercises

1. Show that the caustic condition $f'(u)=0$ has solutions only for $\alpha>e^{3/2}/2$.
2. For $\alpha=6.35$, check that $u=1.925$ is a caustic and find the offset it maps to.
3. A lens has $\alpha=3$. What fraction of a source's flux reaches us when it sits directly behind the core, and what is its delay in units of $t_{\rm DM0}$?
4. Two copies with gains 1.87 and 1.46 overlap completely. Find the brightest and faintest summed intensity across the fringes.

Answers

1. $f'=1+\alpha g(u)$ with $g=(1-2u^2)e^{-u^2}$. The minimum of $g$ is $-2e^{-3/2}$ at $u^2=3/2$, so $f'<0$ somewhere only if $1-2\alpha e^{-3/2}<0$, that is $\alpha>e^{3/2}/2=2.24$.
2. At $u=1.925$, $u^2=3.706$ and $e^{-u^2}=0.0246$. Then $1+6.35\times(1-7.41)\times0.0246=1-1.00\approx0$, a caustic. Its offset is $1.925\times(1+6.35\times0.0246)=2.23$.
3. $G=1/(1+3)=0.25$, a quarter of the flux. With $u=0$ the delay is $t_{\rm DM0}\,e^{0}(1+0)=t_{\rm DM0}$.
4. The intensity swings between $(\sqrt{1.87}+\sqrt{1.46})^2=6.62$ and $(\sqrt{1.87}-\sqrt{1.46})^2=0.03$ times the unlensed value. Their incoherent sum is 3.33, the mean of the two.

Next: [telling source repetition from path effects with timing statistics](https://sandboxphysics.com/stories/fast-radio-burst-echoes/repetition-statistics-source-or-path/).

## Try it in the Lab

[FRB echoes through a plasma lens](https://sandboxphysics.com/labs/frb-plasma-lens-echoes/): Move the source toward the caustic and watch two copies brighten and merge.

## Sources

1. [Lensing of fast radio bursts by plasma structures in host galaxies](https://arxiv.org/abs/1703.06580v1)
2. [FRB 121102 bursts show complex time-frequency structure, Sec. 4.3](https://arxiv.org/abs/1811.10748v1)

## Keep reading

Fast radio bursts: repeats, echoes and plasma lenses · Reading path

### From one burst to source-or-path tests

1. [Article **Is a repeating radio flash the source repeating, or its path?**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/repeat-or-echo/)
2. [Article **How a clump of plasma delays and bends a radio burst**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/plasma-delay-and-bending/)
3. [You are here **Images, gains and delays of a Gaussian plasma lens**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/gaussian-plasma-lens-images/)
4. [Article **Telling source repetition from path effects with timing statistics**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/repetition-statistics-source-or-path/)
5. [Article **Search one CHIME burst for a lens echo and test R3's repeats**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/fit-a-chime-burst-and-r3-repeats/)
6. [Article **How to read the CHIME paper on the 16-day FRB cycle**](https://sandboxphysics.com/stories/fast-radio-burst-echoes/reading-the-chime-r3-periodicity-paper/)
