The Encoding of Natural Numbers as Nested Parentheses Strings with Associated Probability Distributions

We provide an efficient encoding of the natural numbers {0,1,2,3,…} as strings of nested parentheses {(),(()),(()()),((())),…}, or considered inversely, an efficient enumeration of such strings. The technique is based on the recursive definition of the Catalan numbers. The probability distributions arising from this encoding are explored. Applications of this encoding to prefix-free data encoding and recursive function theory are briefly considered.

💡 Research Summary

The paper introduces a novel bijective encoding that maps each natural number to a uniquely structured string of nested parentheses, such as (), (()), (()()), ((())), … . The construction is rooted in the recursive definition of the Catalan numbers, which count the number of correctly balanced parenthesis strings with a given number of pairs. By exploiting the cumulative sums of Catalan numbers, the authors decompose any integer k into a pair (m, ℓ) where m is the smallest index satisfying ∑{i=0}^{m} C_i > k and ℓ = k − ∑{i=0}^{m‑1} C_i. The integer ℓ is then recursively encoded in the same fashion, yielding a final string of the form ( E₁ E₂ … E_m ) with each E_i being a nested sub‑string generated from the recursive step. This algorithm runs in O(log k) time and uses logarithmic space, because each recursion reduces the depth m. The inverse mapping—parsing a parenthesis string back to its natural number—is equally efficient, requiring a single left‑to‑right scan while maintaining a stack depth counter.

Beyond the combinatorial construction, the authors study the probability distribution naturally induced by the encoding. They assign a uniform probability 1/C_m to each string of depth m, and then weight the depth itself by a geometric factor α^m (0 < α < 1), resulting in a normalized prior P(k) = (1 − α) α^m / C_m. This distribution has the desirable property that the total mass sums to one over all natural numbers, and it concentrates most of its probability on short strings when α is small. The special case α = ½ produces an expected code length that is asymptotically comparable to the Shannon entropy of a uniformly distributed integer, demonstrating near‑optimal compression performance.

Two principal applications are explored. First, the encoding yields a prefix‑free code without the need for explicit delimiters: no encoded string is a prefix of another, which makes it directly suitable for streaming or concatenated data transmission. Compared with Huffman or arithmetic coding, the parenthesis code is conceptually simpler and can be generated on‑the‑fly with constant‑time per symbol, while still achieving comparable average lengths for appropriately chosen α. Second, the nested structure mirrors the call stack of recursive functions, providing a concrete representation for recursion trees in the context of computability theory. The authors argue that this representation facilitates analysis of recursive algorithms, enables direct translation between λ‑calculus expressions and parenthesis strings, and connects to the theory of infinite automata (∞‑automata) where states correspond to nesting levels.

The paper also acknowledges practical limitations. Catalan numbers grow super‑exponentially, so direct computation of C_m for large m can cause integer overflow; the authors suggest using logarithmic approximations or arbitrary‑precision libraries. Moreover, the choice of α affects compression efficiency and may need to be adapted to the statistical properties of the source data; an adaptive estimation scheme is proposed as future work. In summary, the work demonstrates how a classic combinatorial sequence can be harnessed to create an elegant, efficient, and theoretically rich encoding of natural numbers, opening avenues for both practical data compression and deeper investigations in formal language and recursion theory.