If it could be recursive perpetual motion would also work:
https://www.youtube.com/watch?v=EiZU3BvqvP4

While the equation can work in both way, I would say it can't. Unless you can somehow break the laws of physics and ignore entropy you would be able to create energy.
You can for instance make it work in reverse by making matter out of light but it would take a lot of energy.

Energy will always be conserved because it transforms but it can't be used to create more energy.

My previous answer might have been a bit hurried and probably misses the point.

If I remember the first topic I have ever posted on fractalforums was related to this as well. I'll see if I can find it in the broken layout.

http://www.fractalforums.com/new-theories-and-research/fractals-entropy-and-probabilities/

but the formulas of physics, especially E=mc² aren't recursive.

E=mc² is not really a law of physics from a modern point of view, but a conversion of units from Joules to kg; mass and energy are not "transformed" into each other, they are one and the same thing.

Current laws of physics are recursive and chaotic and almost every physical system is chaotic and hence has fractal attractors.

Abstractly we can describe the universe by a (large) set of numbers, call them \( z(t) \) (we don't know exactly what they are yet, but let's say all the stuff that goes into the standard model + general relativity). Laws of physics have the form
\( \frac{dz}{dt} = f(z) \), starting from \( z(0) = "the\ big\ bang" \). We don't know exactly what f(z) is yet, not do we know what \( z(0) \) is.
If you consider discrete times \( t_0=0, t_1 = h, t_2=2h, ... \ etc \) with h a small time step (say the Planck time), and call \( z(t_k) = z_k \) we get
\( \frac{dz}{dt} \approx \frac{z_{k+1}-z_k}{h} \) and the "laws of the universe" take the form
\( z_{k+1} = z_k +hf(z_k) \).
If you make a toy universe where z is just a field of complex numbers in a plane with complex coordinate c and \( f(z) = (z^2-z +c)/h \) we get the Mandelbrot equation.
The "equation of the universe" has a similar form, but a bit more complicated.

For example cosmologists have tried to figure out what the real \( z(0) \) is by trying various things and see if they could produce the observed distribution of matter. This is like making a Julia set.

Could you give a specific example? (the simpler the better)

Simplest example is probably the double pendulum.