How could a galaxy lose all its dark matter ?

This paper attempts to simulate the formation of that galaxy lacking in dark matter. It's somewhat taken for granted that while you can fake the appearance of additional dark matter through gravitational encounters - streaming motions along the line of sight are hard to distinguish from rotation, making it look as though a galaxy is rotating very quickly when actually it's just being disrupted. The opposite case, where a galaxy is made to appear deficient in dark matter but actually still has a lot of it, is probably not possible. So the authors try and get a tidal encounter to really remove as much dark matter from their target galaxy as possible.

As we all know, if you do this suddenly and completely, a galaxy will explode. But that requires magic. If you remove it gradually and leave a bit in the centre, there's no reason a galaxy can't survive quite happily. Oh, it'll be more vulnerable to more encounters in the future, but if it's left well alone it'll be fine.

Under the implicit assumption that this particularly weird object legitimately requires a weird progenitor, they therefore start with a target galaxy of unusually, but not exceptionally, low concentration. This means its dark matter halo is unusually extended and therefore easier to remove as it orbits around a more massive galactic thief. No doubt someone will calculate exactly how improbable this is and argue that the chances of detecting a galaxy that has a 1% chance of existing are a million to one, such are the dodgy statistics that seem to be in vogue in astronomy.

Anyway, their target galaxy has stars and dark matter, whereas the burglar galaxy (my term) is a purely analytic, fixed potential. This is a reasonable way to begin, although eventually I'd hope they add in a particle model for the second galaxy as well as including gas and star formation. Gas, in particular, could change things dramatically because it's collisional, but it's reasonable to speculate the target galaxy might have already depleted its gas supply. Using particles for the burglar's halo could also be important, since dynamical friction can increase the chance of a merger.

The authors find that indeed large amounts of dark matter can be stripped by this simple tidal process of close encounters, and the galaxy still survives. This is an important point, but what the paper does not yet adequately demonstrate (it's only submitted, not accepted) is how well the results compare to observations. NGC1052-DF2 is interesting because (criticism notwithstanding) it seems to have little or no dark matter not just in its outer halo, but everywhere. The authors say they reproduce the object's mass halo, but don't give a figure to demonstrate this or quantifiably compare the velocity dispersion of the stars in the simulations with the observations. Without this, the result that the most unbound dark matter can be removed is neither surprising nor novel. It's a decent beginning but the main claim hasn't been made very convincingly yet.
https://arxiv.org/abs/1804.06421