Superluminal Man Meets The Microquark Kid

Faster than light, smaller than an atom

What do you do when you get an unphysical answer on an exam, and is there any way to mitigate this?

Some combination of carrot and stick is the usual way; of course; when I had control over grading policy it was in the navy and we mainly used the big stick: grossly unphysical answers were conceptual errors and it meant a big loss of points on the question — there was no amount of correct information that would let you have a passing score on that particular problem. To drive this home, it was reinforced with feedback from homework and quiz results — I tried to not let an opportunity pass where I could point out how wrong such an answer was and how much it would cost come test-time. But it was also buttressed by having other divisions using similar grading policy, which is a knob not really available (or at least less accessible) in an academic setting.