I am a final year ECE student from Greece, studying about Information Theory a question came up.
In the definition/description of LZ77 (Lempel-Ziv's algorithm), following is noted:
"If n > 1, encode the positive integer u ≤ w using a fixed-length code of length log w bits."
My question is: Why using a fixed-length code and not a variable one? (such as a prefix free or any other uniquely decodable one?).
Source of original lecture in English: http://users.ece.northwestern.edu/~rberry/ECE428/Lectures/lec8_notes.pdf
If you know $u\leq w$ using a fixed length code is usually much simpler. Just use the bit representation of $u$. And it makes no difference to the overall efficiency of the LZ77 scheme.