My question is whether the following scenario is realistic. Imagine coupling a source of photons (say a laser) into a ring made of an optical fiber using a y-junction. (see picture) If the reflectivity of the fiber is high enough to allow the light to make more than one round trip, the intensity in the fiber should increase over time. Assuming it is possible to open the fiber at some point in time, the intensity of the pulse of light exiting the ring could be orders of magnitude higher than the input beam. This system would not have captured photons per se, but at least made it possible to balance and surpass the losses within the fiber and accumulating photons in the process. What do you think about this?
Yes, you’ve essentially described an actively q-switched fiber laser. The difference is that instead of some light input feeding the resonator, there is a gain medium within the loop. Then an output coupler can be electrically toggled to result in laser pulses. These lasers can produce pulses in the mJ energy range with durations around 10 ns or shorter.