Chao LiChao Li
Deepleap.ai
lichao@deepleap.aiDeepleap.ai
lichao@deepleap.ai
We present Domain Algebra (DA), an axiomatic framework built around a single semantic operator, @D. In DA, the domain $d$ is not external metadata applied to a relation; it is constitutive of the relation's meaning.我们提出 Domain Algebra(DA),这是一个围绕单一语义算子 @D 建立的公理化框架。在 DA 中,domain $d$ 不是附加在关系上的外部元数据,而是关系意义的构成部分。
The assertion $\langle c, r@d, c' \rangle$ does not mean “$r(c,c')$ holds, annotated with domain $d$”; it means that $r$ under $d$-semantics relates $c$ to $c'$. This operator bridges two mathematical traditions: on the logical side, it induces Heyting semantics in which truth is constructive completion within a domain; on the geometric side, it instantiates Grothendieck-style relativization through a fibered structure $F : D^{op} \to C$, with Lawvere's hyperdoctrine providing the common language of both.断言 $\langle c, r@d, c' \rangle$ 的含义不是“$r(c,c')$ 成立,并带有域 $d$ 的标注”;它的含义是:在 $d$ 语义下的关系 $r$ 将 $c$ 与 $c'$ 联系起来。这个算子连接了两条数学传统:在逻辑侧,它诱导出 Heyting 语义,其中真值是在域内的构造性完成;在几何侧,它通过纤维结构 $F : D^{op} \to C$ 实现 Grothendieck 式相对化,而 Lawvere 的 hyperdoctrine 则提供了统一两侧的共同语言。
We establish ten axioms (A1-A10) and prove four main results: a four-way equivalence between well-foundedness, termination, normalization, and admissibility; precise capture of the self-application-free fragment of $\lambda$-calculus; a categorical characterization of self-application as non-well-founded endomorphism in a CCC; and a proof that the domain lattice forms a Grothendieck fibration with Heyting fibers. Base independence further shows that any structure-preserving realization yields identical computation results.我们建立十条公理(A1-A10),并证明四个主要结果:良基性、终止性、规范化与 admissibility 之间的四重等价;对不含自应用的 $\lambda$ 演算片段的精确刻画;把自应用表述为 CCC 中的非良基端同态的范畴论刻画;以及证明 domain lattice 构成带 Heyting 纤维的 Grothendieck 纤维化。base independence 进一步表明,任何保持结构的实现都会给出相同的计算结果。
Deepleap.ai
lichao@deepleap.aiDeepleap.ai
lichao@deepleap.ai
The relation $r@d$ does not mean “$r$ annotated by $d$”; it means that $r$ is interpreted under $d$-semantics. Domain is constitutive, not supplementary.关系 $r@d$ 的含义不是“给 $r$ 加上域 $d$ 的注释”,而是“在 $d$ 语义下解释的关系 $r$”。domain 是构成性的,而不是附属性的。
A compact index of the adjacent technical papers listed below, spanning knowledge representation, inference architecture, language modeling, and hardware realization.这里给出下方技术论文的紧凑索引,从知识表示、推理架构、语言模型一直延伸到硬件实现。
These cards summarize the current technical papers in the Domain Algebra line, using arXiv id, primary category, and a short abstract-level description.这些卡片汇总了当前 Domain Algebra 方向的技术论文,按 arXiv 编号、主分类和摘要级简介来组织。
Maps domain algebra directly into memristive crossbar topology, where each junction stores a complete domain-scoped ternary assertion rather than a numerical weight, and demonstrates the design on an ICD-11 respiratory disease classification chip with robust simulated behavior.把 domain algebra 直接映射为忆阻交叉阵列拓扑,使每个交叉点存储的是完整的域约束三值断言,而不是数值权重;论文并以 ICD-11 呼吸系统疾病分类芯片展示了这种设计及其稳定的仿真行为。
Proposes a domain-algebraic language model that replaces flat token decoding with three-phase structured generation over a domain lattice, resolving domain, relation, and concept uncertainty under explicit algebraic constraints.提出一种 domain-algebraic 语言模型,用定义在 domain lattice 上的三阶段结构化生成取代扁平 token 解码,在显式代数约束下依次解析域、关系和概念的不确定性。
Argues for representation-computation unity through CDC four-tuples, deriving domain-scoped closure, typed inheritance, and write-time falsification, then implements these principles in a symbolic inference engine validated on ICD-11 and CBT reasoning tasks.通过 CDC 四元组论证表示与计算统一,推出域约束闭包、类型化继承和写入时证伪等机制,并将其实现为一个符号推理引擎,在 ICD-11 与 CBT 推理任务上进行验证。
Establishes a substrate-agnostic computational architecture for explicit-domain reasoning, with query, extend, and bridge operations, reduced search space under domain-scoped pruning, and validation through a PHQ-9 clinical reasoning case study.建立了一种面向显式域推理、与计算基底无关的计算架构,给出 query、extend、bridge 三类操作,通过域约束剪枝缩小搜索空间,并用 PHQ-9 临床推理案例进行了验证。
Introduces the Domain-Contextualized Concept Graph as a modal knowledge-representation framework in which domain becomes part of the assertion itself, giving truth, inference, and conflict checking a structural and computable domain index.提出 Domain-Contextualized Concept Graph 这一模态知识表示框架,使 domain 成为断言本身的一部分,从而让真值、推理和冲突检查都获得结构性的、可计算的域索引。
These papers trace a continuous arc from semantics to implementation. They begin by treating domain as part of the assertion itself, then develop explicit-domain inference structures, symbolic reasoning engines, structured generation for language models, and finally hardware realizations in memristive arrays.这些论文沿着一条从语义到实现的连续技术线展开:先把 domain 视为断言本身的组成部分,再发展显式域推理结构、符号推理引擎、面向语言模型的结构化生成,最后走向忆阻阵列中的硬件实现。
Taken together, they show Domain Algebra not as a single isolated formalism, but as a framework that moves across representation, inference, generation, and device-level embodiment.把这些工作放在一起看,Domain Algebra 就不再只是一个孤立的形式体系,而是一套能够贯穿表示、推理、生成与器件实现的统一框架。