As for every astronomer next to a pool, it is impossible for me not to notice the similarity between the light patterns that refraction creates on the bottom of the pool, and the image of the cosmic web produced by cosmological simulations.
The similarity is not by chance, but there are deep connections between the two!
First, this is how the cosmic web is predicted to be shaped, by all cosmological simulations (regardless of details on physics and numerics): a complex and multi-scale network of filaments, clustering in knots and separated by voids.
This is movie showing the evolution of the cosmic web of gas density, for a tiny slice crossing a random plane of my simulation.
Surprisingly or maybe not, the formation of this network is well predicted by a linear model which is about 50 years old: “the Zeldovich approximation” , by Y. B. Zeldovich (1970).
Conceived in an epoch in which computing was not very accessible, this simple yet incredibly powerful model can well predict the shaping of matter into a cosmic web just using a linear approximation, and nothing more than a simple x = velocity * time formula.
Basically, given an initial distribution of (macro)particle positions in 3D (discretising cosmic mass), and an initial "displacement field" their further evolution can be well approximated by
x(t) = x0 - D*s(t)
where
q=initial position
D=displacement field
s(t) = growth factor (depending on cosmology)
The "displacement" field D is a 3D field, which models the initial tiny displacement of these particles, under the effect of gravity.
Basically, once that the scales of the initial perturbations induced by gravity are set ( which depends on the cosmological model), the rest of the evolution is set too:
particles start detaching or getting closer, depending on the amplitude of the initial displacement - which really looks like a tiny perturbation on a uniform distribution, at its start.
Very interestingly, for a significant fraction of the following evolution, the shaping of the cosmic web under the Zeldovich approximation proceeds without even taking into account the effect of the mutual gravitational interaction of particles - they just fly apart, keeping memory of the initial perturbation, until they get denser in regions where the flows converge.
After this, the approximation is not valid anymore - because particles must "feel each other", somehow (gravity & collisions)
This is a visualisation of this first stage with a simple #Julialang N-body solver of mine.
There is no gravity force in the entire simulation - particle just "cluster" and produce a cosmic web pattern, because of the initial correlation of their displacement and velocity vectors
These caustics of lights from the outside of the pool are seen as the brightest knots from the outside - they are the equivalent of filaments and halos of the cosmic web - regions of spaces where the trajectories of particles collide.
In nature, at this point these matter chunks will attract each other and form bound halos - in the pool they are just randomly fluctuating light condensations.
which very well shows how time and the vertical direction in the pool example are the direction along with the linear "mapping"of the initial conditions take place.
The concept of "mapping" and the analogy with optics makes this piece of cosmology very suitable for a number of cross-field contamination, applications of non-standard numerical techniques, and so on.
Luckily for you, a master in all this is very active on mastodon: @BrunoLevy01 , who also did a fantastic overview on the evil twitter (maybe also here? I lost it, in case):
To conclude, a subtlety here:
the network of light produced by the pool are a good analog of the simulated cosmic web in 2D slices, in the sense that the pattern formed at the bottom of the pool are very similar to slices through 3D distributions of the cosmic web.
(See this example of a tomography through the gas density of a simulation of mine.)
On the other hand this analogy does not well reproduce the cosmic web seen in its full project - one would need for this many different scattering surfaces than the one at the top of the pool.
There will be much more to say - for example that there is an entire and very productive industry of using the Zeldovich formalism combined with the "adhesion" model to produce very realistic, and extremely quick, models of the cosmic web on its largest scales, and for #cosmology applications, bypassing the numerical problem of computing gravity in very clustered regions, etc..
@ranco_vazza I'm fascinated by this simulation, and it's caused me think a bit further afield than the analysis you're using it as an example for. Could this type of mapping could be used to simulate the action of quantum gravity across dimensions?
Add comment