In general, solitons are single-crest waves which travel at constant speed and don't loose their shape (due to their non-dispersivity), and there are many examples of them in the real world.
Now in QFT a soliton can be defined as a single crest which travels through a potential that (when we consider two dimensions) has the form of a sinusoid in the x-direction and has a value in the y direction that is the same as the corresponding point in the x-direction, more or less like the surface of a frozen sea with sinusoid waves.
Now one part of the soliton lies between two crests of the sinusoid, while the other part lies between the next two crests of the sinusoid, and it's stretched and moves in the y-direction. The y-direction extends to infinity on both sides. For those interested in the math take a look at Sine Gordon equation.
A soliton in QFT can be represented by this photograph:
Both sides of the rod about which the little rods rotate extend to infinity.
Now does a soliton in QFT really exist, or is it a mathematical construction? Or more concrete, does the described potential really exist? And IF they exist, how do they make themselves detectable?
"First of all, most of the work on solitons is formal theory - analyses of soliton solutions in QFTs that are interesting theoretically but that are known not to describe the Universe around us." Where it must, of course, be noted that he wrote "most". – Deschele Schilder May 26 '17 at 10:13