Galaxy-galaxy strong lensing
A galaxy inside a cluster can be a strong lens in its own right, splitting a background galaxy into several images within a few arcseconds of itself. Change the galaxy and the cluster around it, drag the background source, and see how the area enclosed by the galaxy's tangential caustic grows and shrinks.
Drag the background galaxy (a circular Gaussian with a standard deviation of 0.03″) in either panel, or focus one and use the arrow keys. Solid lines are the tangential critical curve (amber) and caustic (white); dashed lines are the radial ones, or for the singular profile, which has no radial critical curve, its cut (the image of the galaxy's centre, a circle in the source plane; crossing it adds or removes one image). The strongly shaded area inside the tangential caustic is the GGSL cross-section, and the faint shading is the rest of the area with three or more images. Where a dotted box is drawn in the source plane, the source is kept inside it so that all of its images stay in the image-plane view.
The singular profile has no core, so there are no core-halved or core-doubled curves to compare with.
What this shows
Following Meneghetti et al. (2020), the GGSL cross-section here is the area of the source plane enclosed by the galaxy's tangential caustic. That caustic opens only when something breaks the galaxy's circular symmetry. For a round galaxy with no external shear, every point of its tangential critical curve (the Einstein ring) maps to the same point behind the galaxy, so the caustic is a single point and the cross-section is exactly 0 arcsec²; the Isolated environment shows this. Inside a cluster, the cluster's shear stretches that point into a diamond with four cusps. The ellipticity of real galaxies opens the caustic too, but this round model leaves it out, so here all of the cross-section comes from the cluster's shear.
The cluster's convergence and shear also bring the galaxy closer to the critical density for lensing. As κc + γc approaches 1 the galaxy approaches the cluster's own critical curve, and the cross-section grows steeply: for the default galaxy with γc = 0.15 it is 0.031 arcsec² at κc = 0, 0.213 arcsec² at κc = 0.6 and 0.93 arcsec² at κc = 0.8. Compactness matters less. At the same mass (the same velocity dispersion and truncation radius), shrinking the core from 1.5 kpc to 0.1 kpc raises the cross-section only from 0.167 arcsec² to 0.231 arcsec², while raising the velocity dispersion from 120 km/s to 280 km/s takes it from 0.045 arcsec² to 0.87 arcsec²; the core matters most for the least massive galaxies, as the chart shows. The profile selector makes the same comparison: at 180 km/s in the default field (κc = 0.6, γc = 0.15) the cross-section is 0.365 arcsec² for the singular isothermal sphere, 0.326 arcsec² for the cored one (0.5 kpc core) and 0.213 arcsec² for the truncated one (0.5 kpc core, 19 kpc truncation radius), so removing the core adds about 12% while truncating the halo removes about a third. Because the cross-section depends on how much mass a galaxy holds near its centre, counts of galaxy-galaxy strong lensing events in clusters are a probe of how dark matter is distributed in cluster galaxies.
The area with three or more images is a different number. It also counts the sources inside the radial caustic, which have three images even when they lie outside the tangential caustic (sources inside both have five). With the defaults, the tangential caustic encloses 0.213 arcsec² while three or more images occur over 0.431 arcsec². The radial caustic is far more sensitive to the core than the tangential one: the area with three or more images is 0.888 arcsec² with a 0.1 kpc core and 0.177 arcsec² with a 1.5 kpc core.
Model and assumptions
By default the member galaxy is a dual pseudo-isothermal sphere (dPIE): a projected density that is flat inside the core radius, falls as 1/r like an isothermal sphere in between, and steepens to 1/r³ beyond the truncation radius. Its velocity dispersion σ is the equivalent singular isothermal value, defined by the lensing strength 4π(σ/c)² Dls/Ds. Lenstool's σdPIE (Elíasdóttir et al. 2007) is related to it by σ² = (3/2) σdPIE² (1 − rcore²/rcut²), so 180 km/s here is about 147 km/s there for the default radii. The total mass is πσ²rcut/G, which does not depend on the core radius. The galaxy sits at the redshift of Abell S1063, z = 0.348.
The galaxy profile selector offers three cases of this profile. The singular isothermal sphere fixes the core radius at 0 and the truncation radius at infinity: its density falls as 1/r all the way in, its deflection is the same, 4π(σ/c)² Dls/Ds, at every radius, and it has no radial critical curve (the galaxy's centre maps instead to a circle, the cut). The cored isothermal sphere keeps the core but fixes the truncation radius at infinity. The truncated profile, the dPIE, lets both vary. Without truncation the mass is infinite, so the mass readout is shown only for the dPIE. The Isolated environment fixes κc and γc at 0.
Over the few arcseconds around the galaxy, the cluster is treated as a constant convergence κc and a constant shear γc, both given for the chosen source redshift. The shear stretches images along the horizontal axis, as if the cluster centre lay above or below the frame. κc + γc is kept at or below 0.95, so the galaxy lies outside the cluster's own critical curve; the shear slider's range shrinks as the convergence grows. Close to that limit the galaxy's critical curves grow to several arcseconds and its outer images lie tens of arcseconds away, where a constant κc and γc is no longer a fair description of a cluster. The Einstein radius is where the mean convergence inside a circle, cluster included, reaches 1.
The lensing Jacobian has two eigenvalues, 1 − κ − |γ| (tangential) and 1 − κ + |γ| (radial), where κ and γ are the total convergence and shear of galaxy and cluster. Each is contoured separately on a fine polar mesh of the image plane to give the tangential and radial critical curves, which are mapped to the source plane to give the caustics. The cross-section is the area of the points that the tangential caustic winds around, summed along 2048 scan lines. The area with three or more images is measured by mapping the mesh triangles to the source plane and counting, at each point of a grid around the caustics, how many land on it. The chart uses a coarser mesh than the readouts, and its marker is computed on that same mesh, so it can differ from the readout by about 0.1%. The background galaxy is a circular Gaussian with a standard deviation of 0.03″. Distances use a flat ΛCDM cosmology with H0 = 70 km/s/Mpc and Ωm = 0.3. This is an illustration of the physics, not a model of a particular galaxy.
References
Elíasdóttir et al. 2007 (arXiv:0710.5636), on the dPIE profile in cluster lens models. Meneghetti et al. 2020, Science 369, 1347 (arXiv:2009.04471), on the excess of galaxy-galaxy strong lensing in observed clusters over simulations. See also Dutra, Natarajan and Gilman 2025, on steeper member profiles and this discrepancy.