Hmm, I think you got it in reverse. We are talking about of properties of real physical things all the time here. We just can't expect to instantiate any random piece of math in the real world - like take a specific vector subspace and try to use it as a katana, that would be silly.

We take a physical system and apply some mathematical properties to it. In the case of bullet I think what is interesting is to model it's energy, E = 1/2 mv*v. Now, if we take the kinetic energy of a bullet, and apply it to any other object, and then instantiate this object in the real world the lethality of the object depends on the nature of the object. A ball of hay of a diameter of 1 m in 1 atm 10 meters away with the kinetic energy of a bullet does not sound lethal. An object more like a bullet, like an iron nail, probably can be. Gummy bear? Who knows.

The point is, the maths here are about physical things, although it can take a lot of experimental work to manufacture a specific physical system that exhibits behaviour which is modeled with specific piece of math.

Quasi-particles are not just math. They have measurable properties that are like properties of real particles (just like "holes" in P-type semiconductors can be measured to behave like quasi-positrons). Or, replying to your proposition:

"With the holodeck safeties off, even holographic bullets can kill." - Picard, First Contact.

:)

That's true for values of "you" reading this in 2015, at least. :)