Table of Contents
Fetching ...

Geometric Characterization of Context-Free Intersections via the Inner Segment Dichotomy

Jorge Miguel Silva

TL;DR

A dichotomy is proved for this measure: bounded inner segments imply context-freeness via a finite buffer construction; growing inner segments with pump-sensitive linkages imply non-context-freeness via a finite buffer construction.

Abstract

The intersection of two context-free languages is not generally context-free, but no geometric criterion has characterized when it remains so. The crossing gap (max(i'-i, j'-j) for two crossing push-pop arcs) is the natural candidate. We refute this: we exhibit CFLs whose intersection is CFL despite unbounded-gap crossings. The governing quantity is the inner segment measure: for crossing arcs inducing a decomposition w = P1 P2 P3 P4, it is max(|P2|,|P3|), the length of the longer inner segment between interleaved crossing endpoints. We prove a dichotomy for this measure: bounded inner segments imply context-freeness via a finite buffer construction; growing inner segments with pump-sensitive linkages imply non-context-freeness. The inner segment concept applies to all CFL intersections; the strictness of the resulting characterization depends on the language class. For block-counting CFLs (languages requiring equality among designated pairs of block lengths), the dichotomy is complete: the intersection is CFL if and only if the combined arcs are jointly well-nested. For general CFLs, the CFL direction is unconditional; the non-CFL direction requires pump-sensitive linkages whose necessity is the main open problem, reducing the general CFL intersection problem to a specific property of pump-sensitive decompositions.

Geometric Characterization of Context-Free Intersections via the Inner Segment Dichotomy

TL;DR

A dichotomy is proved for this measure: bounded inner segments imply context-freeness via a finite buffer construction; growing inner segments with pump-sensitive linkages imply non-context-freeness via a finite buffer construction.

Abstract

The intersection of two context-free languages is not generally context-free, but no geometric criterion has characterized when it remains so. The crossing gap (max(i'-i, j'-j) for two crossing push-pop arcs) is the natural candidate. We refute this: we exhibit CFLs whose intersection is CFL despite unbounded-gap crossings. The governing quantity is the inner segment measure: for crossing arcs inducing a decomposition w = P1 P2 P3 P4, it is max(|P2|,|P3|), the length of the longer inner segment between interleaved crossing endpoints. We prove a dichotomy for this measure: bounded inner segments imply context-freeness via a finite buffer construction; growing inner segments with pump-sensitive linkages imply non-context-freeness. The inner segment concept applies to all CFL intersections; the strictness of the resulting characterization depends on the language class. For block-counting CFLs (languages requiring equality among designated pairs of block lengths), the dichotomy is complete: the intersection is CFL if and only if the combined arcs are jointly well-nested. For general CFLs, the CFL direction is unconditional; the non-CFL direction requires pump-sensitive linkages whose necessity is the main open problem, reducing the general CFL intersection problem to a specific property of pump-sensitive decompositions.
Paper Structure (36 sections, 10 theorems, 4 equations, 3 figures)

This paper contains 36 sections, 10 theorems, 4 equations, 3 figures.

Key Result

Theorem 3.3

Let $L\subseteq\Sigma^*$ be a language containing, for infinitely many $n$, a string $w_n$ with a factorization $w_n=P_1 P_2 P_3 P_4$ satisfying: Then $L$ is not context-free.

Figures (3)

  • Figure 1: The pumping obstruction for crossing linkages (\ref{['thm:crossing']}). Top: the string $w_n = P_1 P_2 P_3 P_4$ with crossing linkages $(P_1,P_3)$ and $(P_2,P_4)$. Bottom: since $|vxy| \le p < n \le |P_i|$, the pumpable substring (shown for each case) intersects at most two adjacent segments. Every placement triggers a linkage contradiction (✗): it touches exactly one member of a linked pair without reaching the other.
  • Figure 2: The displacement buffer mechanism (\ref{['thm:bounded-gap']}). Left: the combined stack interleaves entries from $M_1$ (blue) and $M_2$ (red). When $M_1$ pops, its target $(1,\gamma)$ (green) has at most $2k$ crossing $M_2$ entries above it. Right: the $M_2$ entries are displaced into the buffer $\beta$ (finite control), the target is popped, and the displaced entries are restored.
  • Figure 3: The inner segment dichotomy. Top: bounded inner segments ($\max(|P_2|,|P_3|)\le D$); the left-starting crossing arc has span $\le 2D$ and is absorbed into a finite buffer (\ref{['thm:buffer']}). Bottom: growing inner segments ($\max(|P_2|,|P_3|)\to\infty$); the left-starting arc grows unboundedly, the pumping lemma substring $vxy$ cannot straddle both linked segments, and the language is not CFL (\ref{['thm:strengthened']}).

Theorems & Definitions (34)

  • Example 2.1
  • Definition 3.1
  • Remark 3.2: Universal quantification is intentional
  • Theorem 3.3: Pumping incompatibility of crossing linkages
  • proof
  • Corollary 3.4
  • proof
  • Example 3.5: $\{a^n b^n c^n d^n\}$
  • Remark 3.6
  • Example 3.7: $\{a^n b^n c^n\}$
  • ...and 24 more