Zheng–Weihrauch Hierarchy

• The Zheng–Weihrauch hierarchy is an arithmetical classification of real numbers using alternating computable suprema and infima to capture effective numerical complexity.
• It categorizes reals into Σ, Π, and Δ levels, providing a framework to analyze solutions of PDEs, applications of Fekete’s lemma, and various analytic limits.
• This structure connects computability theory with reverse mathematics and Weihrauch degrees, revealing intrinsic barriers in algorithmically solving analytic and variational problems.

The Zheng–Weihrauch hierarchy is an arithmetic hierarchy of real numbers that stratifies degrees of non-computability encountered in computable analysis, particularly through alternated suprema and infima of computable rational sequences. It provides a precise framework for classifying the effective complexity of both numerical and analytic tasks—including limits, infima, boundary value problems, and combinatorial limits—within the context of Turing computability and higher-type reductions. This hierarchy interacts closely with foundational topics such as Fekete’s lemma, the arithmetic complexity of PDE solutions, and the reverse-mathematical classification of mathematical theorems.

1. Formal Definition of the Zheng–Weihrauch Hierarchy

The Zheng–Weihrauch hierarchy (Boche et al., 11 Jan 2026, Boche et al., 2020) is an arithmetical hierarchy of real numbers based on alternations of computable suprema and infima.

• Layer 1 (Σ₁, Π₁, Δ₁):
  • Σ₁ (left-computable, lower semi-computable): $x \in \Sigma_1$ if there exists a computable non-decreasing rational (or real) sequence $(\zeta_n)$ such that $\lim_{n\rightarrow\infty}\zeta_n = x$.
  • Π₁ (right-computable, upper semi-computable): $x \in \Pi_1$ if there exists a computable non-increasing rational (or real) sequence $(\xi_n)$ such that $\lim_{n\rightarrow\infty}\xi_n = x$.
  • $C = \Sigma_1 \cup \Pi_1$ (the weakly computable reals).
  • Δ₁ ($\Sigma_1 \cap \Pi_1$): exactly the Turing-computable real numbers.
• Layer 2 (Σ₂, Π₂, Δ₂):
  • Δ₂ (“recursively approximable”): $x \in \Delta_2$ iff there exists a computable sequence of computable reals $(x_n)$ with $\lim_{n\rightarrow\infty}x_n = x$.
  • Σ₂: $x \in \Sigma_2$ iff there exists a computable double sequence $(\zeta_{n,k})\subseteq\Sigma_1$ such that

  $$x = \sup_{n\in\mathbb{N}} \inf_{k\in\mathbb{N}} \zeta_{n,k}$$ - Π₂: $x \in \Pi_2$ iff there exists a computable double sequence $(\xi_{n,k})\subseteq\Sigma_1$ such that

  $$x = \inf_{n\in\mathbb{N}} \sup_{k\in\mathbb{N}} \xi_{n,k}$$ - Δ₂ ($\Sigma_2 \cap \Pi_2$): the set of reals recursively approximable both ways.

• Higher Layers ($\Sigma_n, \Pi_n, \Delta_n$):

For $n \geq 1$, one alternates sup and inf $n$ times over computable arrays of rationals (or computable reals):

$$\Sigma_n: x = \sup_{m_1} \inf_{m_2} \sup_{m_3} \ldots \Theta r_{m_1, \dots, m_n}$$

$$\Pi_n: x = \inf_{m_1} \sup_{m_2} \inf_{m_3} \ldots \Theta r_{m_1, \dots, m_n}$$

where Δₙ = Σₙ ∩ Πₙ.

This hierarchy is reminiscent of the Kleene–Mostowski hierarchy for sets but applies to real numbers, capturing the arithmetical complexity of limits and solutions to analytic problems.

2. Semicomputability, Weak Computability, and Structural Properties

• Semicomputability refers to real numbers that can be computably approximated from one side; $\Sigma_1$ and $\Pi_1$ are precisely the lower and upper semicomputable reals, respectively (Boche et al., 2020).

• Weakly computable numbers ($C$) are those that are either left- or right-computable. Δ₁ = Σ₁ ∩ Π₁ recovers the computable reals.

• Closure properties:

  • Δₙ is a field—closed under $+,-,\times,\div$—while none of $\Sigma_n, \Pi_n,$ or $C$ has this property. This has implications for which classes solutions to analytic problems land in: for example, the Dirichlet principle yields $\Sigma_1$ energies but not necessarily Δ₁ energies, reflecting the lack of a computable complementary approximation (Boche et al., 11 Jan 2026).

3. Connection to Fekete’s Lemma and Combinatorial Limits

A key insight is that the classes $\Sigma_1$ and $\Pi_1$ coincide exactly with the “Fekete-limit’’ classes:

• $\Sigma_1 = L^+$: limits of computable superadditive sequences of computable reals.
• $\Pi_1 = L^-$: limits of computable subadditive sequences.
• Δ₁ = $\Sigma_1 \cap \Pi_1 = L^+ \cap L^-$.

Examples:

• Specker sequences yield noncomputable left-computable $\Sigma_1$-reals.
• Chaitin's $\Omega$ is $\Sigma_1$-complete.
• For nonnegative computable superadditive rational sequences, Fekete's lemma yields computable limit iff the limit is itself a computable real (Boche et al., 2020).

This identifies the Fekete-limit phenomenon (emergence of noncomputable limits from computable rate sequences) with the first layer of the Zheng–Weihrauch hierarchy.

4. Analytical Problems and Placement Within the Hierarchy

Dirichlet Problem on the Disk

The paper (Boche et al., 11 Jan 2026) analyzes the arithmetic complexity of solutions to the Dirichlet problem using both the variational (Dirichlet-principle) and Poisson-integral methods, mapping the resulting values into the Zheng–Weihrauch hierarchy.

Method	Output Class	Explanation
Variational (energy minimizer)	$\Sigma_1$ (left-computable)	The minimal energy $\inf_{u} E(u)$ is always in $\Sigma_1$, and every $\Sigma_1$-number can be realized as the energy for a computable boundary function.
Poisson-integral (pointwise)	At least $C$, at most $\Delta_2$	Pointwise boundary evaluations $f(e^{i\theta})$ for computable $f$ in Sobolev class lie in $\Delta_2$, and for special choices can be as "bad" as any weakly computable number.

• Upper bound for variational energy: Fourier expansion yields a computable, non-decreasing sequence converging to the Dirichlet energy.
• Lower bound: Any $\Sigma_1$ real can be embedded via combinatorial construction in the values of the variational problem.

For the Poisson-integral approach, the worst-case degree is at most $\Delta_2$ and at least $C$ (the set of weakly computable reals). Tighter characterization of this complexity remains open.

Broader significance: Classical analytic procedures (minimization, integration) naturally yield outputs at precise levels of the hierarchy, identifying non-trivial barriers to effective computation of solutions to PDEs and variational problems.

5. Weihrauch Degrees, Reverse Mathematics, and the Hierarchy

The Zheng–Weihrauch hierarchy underpins the classification of theorems by Weihrauch reducibility, which structures the “degree of non-computability” of multi-valued principles between represented spaces. There are precise correspondences:

Reverse Mathematics System	Weihrauch Degree	Zheng–Weihrauch Layer
RCA₀ (computable comprehension)	computable ($0$)	Δ₁
WKL₀ (Weak König’s Lemma)	$C_{2^\mathbb{N}}$ (closed choice)	weakly computable classes
ACA₀ (arithmetical comp.)	lim (limit operation)	transition between Δ₁ and higher layers
ATR₀ (arithmetical transfinite recursion)	$UC_{\mathbb{N}^{\mathbb{N}}}$, $\Sigma^1_1$-separation, comparability of well-orderings (all ≡ in the Weihrauch lattice)	hyperarithmetical, essentially $\Sigma^1_1$-class
Π¹₁‐CA₀ (analytic comprehension)	$C_{\Pi^1_1}$ (choice for countable $\Pi^1_1$ classes)	not yet fully classified

This placement (Kihara et al., 2018) reflects the deep connections between arithmetic hierarchies for sets and functions, higher-type computability theory, and the analysis of effective content in classical theorems (e.g., open determinacy, the perfect tree theorem).

6. Implications, Limitations, and Open Problems

• Precision of Complexity Bounds: For certain analytic operations, especially pointwise evaluation in the Poisson-integral method, the exact layer (Δ₂ vs a potentially smaller subclass) is yet to be determined.
• Hierarchy Completeness: Classical analytic operations (integration, minimization) often occupy the boundaries between layers (e.g., from $\Sigma_1$ but not $\Delta_1$), indicating sharp intrinsic thresholds for effective computation.
• Field Structure and Computable Analysis: The fact that only Δₙ is closed under field operations limits the potential for algorithmic operations in lower classes and explains specific noncomputability phenomena in analysis.
• No Constructive Fekete: The impossibility of uniformly computing Fekete limits from computable input sequences is a direct consequence of the exact $\Sigma_1$ characterization (Boche et al., 2020).
• Broader Significance: The hierarchy highlights how fundamental existence results in analysis mask subtle stratifications of computability, and provides a precise language for describing these barriers.

Open questions include the optimal upper bound for pointwise boundary evaluations, the completeness and separation of higher Δₙ, Σₙ, Πₙ classes, and the relationships between Weihrauch degrees and closure properties under composition in higher-type settings (Boche et al., 11 Jan 2026, Kihara et al., 2018).

• "Arithmetic Complexity of Solutions of the Dirichlet Problem" (Boche et al., 11 Jan 2026)
• "On Effective Convergence in Fekete's Lemma and Related Combinatorial Problems in Information Theory" (Boche et al., 2020)
• "Searching for an analogue of ATR in the Weihrauch lattice" (Kihara et al., 2018)