[floxy]

> Instead, they try to solve everything analytically...

This seems like a common refrain it lots of things I see (not just this one paper). Can anyone give a lay man's explanation why we can't just numerically simulate general relativity? As in, plug a simulation with 100 billion stars in to a super computer and see what comes out.

[ncmncm]

It ought to suffice to simulate the motion of exactly one star, in a circular orbit, for exactly one time-step, just adding up all the effects of each of the 1e11 or so other stars, plus interstellar medium. There are two possible end states: either it follows the circle--no dark matter needed--or it swings wide.