If you can distinguish, you can express: Galois theory, Stone--Weierstrass, machine learning, and linguistics
Pith reviewed 2026-05-18 07:51 UTC · model grok-4.3
The pith
Both the Fundamental Theorem of Galois Theory and the Stone-Weierstrass theorem tie a class's power to distinguish objects to its power to express or approximate them.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
Both the Fundamental Theorem of Galois Theory and the Stone-Weierstrass theorem assert that distinguishing power implies expressive power; an elementary theorem directly connects the notions of distinguishing power that appear in each setting, and the same link recurs as a working principle in machine learning and as a foundational assumption in linguistics.
What carries the argument
The elementary theorem that links distinguishing power to expressive power for a class of objects.
If this is right
- Galois correspondences follow once automorphisms are shown to distinguish the relevant subgroups.
- Dense subalgebras follow once they are shown to separate points on a compact space.
- Machine-learning models that separate data classes can be used to construct approximations to target functions.
- Linguistic systems that mark basic distinctions can generate the full range of expressible meanings.
Where Pith is reading between the lines
- The same distinction-to-expression step could be checked in other settings such as differential equations or topological dynamics.
- Feature engineering in data science might be organized by first verifying separation before optimizing expressivity.
- The pattern supplies a template for proving density or completeness results once a separation property is established.
Load-bearing premise
The notions of distinguishing power that appear in Galois theory and in Stone-Weierstrass are close enough that one elementary theorem can connect them without extra assumptions special to fields or to real-valued functions.
What would settle it
A concrete counter-example in which a set of field automorphisms distinguishes every subgroup yet fails to recover the full lattice of intermediate fields, or a subalgebra that separates points yet cannot approximate every continuous function on the compact space.
read the original abstract
This essay develops a parallel between the Fundamental Theorem of Galois Theory and the Stone--Weierstrass theorem: both can be viewed as assertions that tie the distinguishing power of a class of objects to their expressive power. We provide an elementary theorem connecting the relevant notions of "distinguishing power". We also discuss machine learning and data science contexts in which these theorems, and more generally the theme of links between distinguishing power and expressive power, appear. Finally, we discuss the same theme in the context of linguistics, where it appears as a foundational principle, and illustrate it with several examples.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper develops a parallel between the Fundamental Theorem of Galois Theory and the Stone--Weierstrass theorem, framing both as statements that connect the distinguishing power of a class of objects (automorphisms or subalgebras) to their expressive power (fixed fields or dense subalgebras). It supplies an elementary theorem that links the relevant notions of distinguishing power, then extends the theme to machine learning, data science, and linguistics with illustrative examples.
Significance. If the elementary theorem provides a uniform, structure-independent link between distinguishing and expressive power that applies without additional translation lemmas, the manuscript would offer a clean unifying perspective on two classical results and their echoes in modern applied fields. The interdisciplinary discussion is a strength, but the significance is limited by the need for the core connection to be shown to be more than an analogy.
major comments (2)
- [elementary theorem section] The elementary theorem (main body, after the statements of the two classical theorems): the proof appears to require separate verifications that the distinguishing relation behaves identically in the Galois setting (fixed-point subfields) and the Stone--Weierstrass setting (point-separating subalgebras on compact spaces); without an explicit isomorphism or abstraction lemma that avoids invoking field operations or continuity, the claimed single theorem reduces to two parallel but non-identical arguments.
- [machine learning discussion] Application to machine learning (ML section): the claim that the same distinguishing-to-expressive link appears in neural-network universality results is not supported by a precise reduction; the paper does not show that the finite distinguishing sets used in the elementary theorem correspond to the finite-width or finite-sample regimes typical in ML bounds.
minor comments (2)
- [abstract and introduction] The abstract and introduction use the phrase 'elementary theorem' without a forward reference to its precise statement or numbering; adding a label such as 'Theorem 1' would improve readability.
- [linguistics section] Several linguistic examples (final section) are presented informally; a short table listing the distinguishing feature and the corresponding expressive consequence for each example would make the pattern clearer.
Simulated Author's Rebuttal
We thank the referee for the careful reading and constructive comments on our essay. The report correctly identifies the core theme and the intended scope of the interdisciplinary discussion. Below we respond point by point to the major comments, indicating where we will revise the manuscript to strengthen the presentation while preserving the essay's character as a unifying perspective rather than a technical treatise.
read point-by-point responses
-
Referee: [elementary theorem section] The elementary theorem (main body, after the statements of the two classical theorems): the proof appears to require separate verifications that the distinguishing relation behaves identically in the Galois setting (fixed-point subfields) and the Stone--Weierstrass setting (point-separating subalgebras on compact spaces); without an explicit isomorphism or abstraction lemma that avoids invoking field operations or continuity, the claimed single theorem reduces to two parallel but non-identical arguments.
Authors: We appreciate this observation and agree that the uniformity can be made more explicit. The elementary theorem is formulated purely in terms of an abstract distinguishing relation (a binary relation between collections of functions and elements of the domain) together with the definition of expressive power as the smallest structure closed under the relevant operations. The proof itself uses only this relation and the closure property, without invoking addition, multiplication, or continuity. Nevertheless, to eliminate any appearance of separate verifications, we will insert a short abstraction lemma that isolates the common set-theoretic content of the distinguishing relation before applying it to the two classical settings. This lemma will not require an isomorphism between fields and function algebras, but will simply record the shared logical structure. revision: yes
-
Referee: [machine learning discussion] Application to machine learning (ML section): the claim that the same distinguishing-to-expressive link appears in neural-network universality results is not supported by a precise reduction; the paper does not show that the finite distinguishing sets used in the elementary theorem correspond to the finite-width or finite-sample regimes typical in ML bounds.
Authors: We agree that the ML discussion is illustrative rather than a formal embedding. The section invokes the elementary theorem to explain why the ability of a network family to distinguish finite sets of points or features is linked to its capacity to approximate arbitrary continuous functions, drawing on known universality results. It does not claim a direct quantitative correspondence between the finite distinguishing sets of the theorem and the finite-width or finite-sample regimes in specific ML bounds. In revision we will add a clarifying paragraph that explicitly labels the connection as conceptual motivation, cites representative universality theorems that rely on point-separation or finite-sample distinguishability, and notes that a rigorous quantitative translation would constitute a separate research project. revision: partial
Circularity Check
No significant circularity; derivation externally grounded in classical theorems
full rationale
The paper's central claim is an interpretive parallel between the Fundamental Theorem of Galois Theory and Stone-Weierstrass, linked by a newly stated elementary theorem on distinguishing vs. expressive power. No equations or definitions in the provided abstract or description reduce a derived quantity to a fitted input or self-citation by construction. The elementary theorem is presented as an independent connecting result rather than a renaming or ansatz smuggled from prior author work. Machine learning and linguistics sections apply the theme but do not serve as load-bearing derivations for the mathematical core. The reasoning remains self-contained against external benchmarks of the two classical theorems.
Axiom & Free-Parameter Ledger
axioms (2)
- standard math Fundamental Theorem of Galois Theory
- standard math Stone-Weierstrass theorem
Lean theorems connected to this paper
-
IndisputableMonolith/Foundation/AbsoluteFloorClosure.leanabsolute_floor_iff_bare_distinguishability echoes?
echoesECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.
both the Stone–Weierstrass theorem and the Fundamental Theorem of Galois Theory can be interpreted as statements of the form distinguish⇔express
-
IndisputableMonolith/Foundation/RealityFromDistinction.leanreality_from_one_distinction echoes?
echoesECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.
If you can distinguish, then you can express (and conversely)
What do these tags mean?
- matches
- The paper's claim is directly supported by a theorem in the formal canon.
- supports
- The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
- extends
- The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
- uses
- The paper appears to rely on the theorem as machinery.
- contradicts
- The paper's claim conflicts with a theorem or certificate in the canon.
- unclear
- Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.