Core Concepts
The author explores the upper bounds and average lengths of variable-length non-overlapping codes, linking them to fixed-length codes.
Abstract
The study delves into non-overlapping codes, focusing on their cardinality and construction methods. It establishes a connection between variable-length and fixed-length codes, providing insights into their sizes and average lengths. The paper concludes that variable-length codes do not offer advantages in terms of cardinality compared to fixed-length ones.
Stats
C(n, q) denotes the maximum size of fixed-length non-overlapping codes over an alphabet of size q.
The best-known upper bound for C(n, q) is given by Levenshtein as C(n, q) ≤ (n - 1)^n * q^(n/n).
The minimal average length L of a q-ary non-overlapping code with cardinality ˜C satisfies ⌈logq ˜C⌉ ≤ L ≤ n.
For a prefix code, the expected length L = P pili satisfies L ≥ Hq(X), where Hq(X) = − P i pi logq pi.
Quotes
"The size of a q-ary variable-length non-overlapping code is upper bounded by C(n, q)."
"Variable-length non-overlapping codes do not offer advantages in terms of cardinality compared to fixed-length ones."