Hi,
tuxguitar is not a software synthesizer, so it can't implements it.
tuxguitar just connect to a MIDI Synthesizer, and the synth is who play the sounds.
See this connection examples:
TuxGuitar -> Java -> Java Sound Synthesizer
TuxGuitar -> ALSA -> Timidity
TuxGuitar -> ALSA -> Fluidsynth
So it's the synthesizer who should to implements the algorithm if you want to heard it.
The "Karplus Strong algorithm" (I personally don't like the name, since it sounds kind of egotistical) can also be generalized to REALLY BIG buffers (several thousand samples)
with a few tricks that have not been published anywhere as far as I know. But you have to know something about complex
numbers and z-transforms in order to get the tuning and decay times to come out right. Read the original CMJ article from 1983 and then think really really hard. It was impractical several decades ago, but is entirely feasible now with fast CPUs. The generalizations are still simple enough not to require multiplications (just addition and right-shift for the /2 in the basic formula). BTW the original patents have now expired in the US, although there may be some still valid in Japan. Good luck, and I am very glad that what I like to call "the guitar algorithm" is still being used after so many years.