pith. sign in

arxiv: 2601.13946 · v2 · submitted 2026-01-20 · 🧮 math.ST · stat.TH

Topological Criteria for Hypothesis Testing with Finite-Precision Measurements

Pith reviewed 2026-05-16 12:52 UTC · model grok-4.3

classification 🧮 math.ST stat.TH
keywords finite-precision teststopological criteriahypothesis testingF_sigma setsweak topologyconditional independencePolish spacesequicontinuity
0
0 comments X

The pith

A pair of hypotheses admits a consistent finite-precision test exactly when both are F_sigma sets in the weak topology on their union.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes necessary and sufficient topological conditions for distinguishing two statistical hypotheses when observations are recorded only to finite precision. By modeling finite-precision tests as those whose decision regions are open in the sample-space topology, it shows that consistent distinguishability holds if and only if both the null and alternative are countable unions of closed sets under the weak topology on the space of probability measures. This criterion also yields equivalences between different forms of error control and directly implies that conditional independence cannot be consistently tested from finite-precision data on Polish spaces without regularity assumptions. The result strengthens prior impossibility findings and recovers testability under an equicontinuity condition on conditional distributions that is itself testable.

Core claim

The authors prove that for any pair of hypotheses H0 and H1, there exists a consistent finite-precision test if and only if both H0 and H1 are F_sigma sets in the weak topology on W equal to their union. They further establish that finite-sample error control, asymptotic error control, and uniform convergence of errors are equivalent for both classical and finite-precision tests; uniform error control under Hi holds precisely when Hi is closed in W; and uniformly consistent testing with bounded precision is possible whenever H0 and H1 are metrically separated.

What carries the argument

Finite-precision tests, whose decision regions are required to be open sets in the sample-space topology, which allow the reduction of consistent testability to the F_sigma property in the weak topology on the union of the hypotheses.

If this is right

  • Uniform error control under each hypothesis Hi holds if and only if Hi is closed in W.
  • Uniformly consistent testing with bounded precision becomes possible whenever H0 and H1 are metrically separated.
  • Conditional independence is not consistently testable from finite-precision data on Polish sample spaces when the conditioning space has no isolated points.
  • Under an equicontinuity assumption on the family of conditional distributions, consistent finite-precision testing of conditional independence is recovered with uniform error control under the null.
  • The equicontinuity assumption itself admits a finite-precision test, rendering the overall procedure assumption-free in a precise sense.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same topological criterion may classify testability for other composite hypotheses such as goodness-of-fit or independence in infinite-dimensional models.
  • Similar open-region restrictions could be applied to sequential or adaptive testing procedures to obtain analogous F_sigma characterizations.
  • One could first test the equicontinuity assumption with a finite-precision procedure and only proceed to conditional-independence testing when it passes.
  • The framework suggests that measurement-precision limits impose topological regularity requirements on any hypothesis that is to remain testable.

Load-bearing premise

Modeling the limitations of finite-precision recording solely by requiring decision regions to be open sets in the sample-space topology is sufficient to capture the relevant constraints.

What would settle it

A concrete pair of hypotheses H0 and H1 that are not both F_sigma in the weak topology on W yet admit a consistent finite-precision test (or conversely, a pair that are F_sigma yet admit no such test) would falsify the claimed equivalence.

read the original abstract

We establish topological necessary and sufficient conditions under which a pair of statistical hypotheses can be consistently distinguished when i.i.d. observations are recorded only to finite precision. To accommodate finite-precision data, we introduce finite-precision tests: tests whose decision regions are open in the sample-space topology. We first show that, both for classical and finite-precision tests, the existence of such tests with finite-sample error control, asymptotic error control, or uniform convergence of the errors are all equivalent. A pair of null- and alternative hypotheses $H_0$ and $H_1$ admits a consistent finite-precision test if and only if both are $F_\sigma$ in the weak topology on the space of probability measures $W := H_0\cup H_1$. The hypotheses admit uniform error control under $H_i$ if and only if $H_i$ is closed in $W$, and admit uniformly consistent testing with bounded precision under metric separation of $H_0$ and $H_1$. These criteria imply that, without regularity assumptions, conditional independence is not consistently testable from finite-precision data when the conditioning space has no isolated points - strengthening existing impossibility results to Polish sample spaces and showing that even pointwise consistency cannot be obtained. We introduce an equicontinuity assumption on the family of conditional distributions under which we recover consistent finite-precision testability of conditional independence with uniform error control under the null, provided sample spaces are Polish and the conditioning space is locally compact. The equicontinuity assumption is itself a finite-precision-testable hypothesis, so the resulting test for conditional independence is, in a precise sense, assumption-free.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims that a pair of hypotheses H0 and H1 admits a consistent finite-precision test (decision regions open in the sample-space topology) if and only if both are F_σ sets in the weak topology on W = H0 ∪ H1. It establishes equivalence of finite-sample, asymptotic, and uniform error control for both classical and finite-precision tests, shows that uniform error control under Hi holds precisely when Hi is closed in W, and derives uniform consistency under metric separation. These criteria are applied to conditional independence, yielding an impossibility result (no consistent finite-precision test when the conditioning space has no isolated points) that strengthens prior work to Polish spaces, and a recovery result under an equicontinuity assumption on conditional distributions that is itself finite-precision testable.

Significance. If the central topological characterization holds, the work supplies clean necessary-and-sufficient conditions linking hypothesis testing under finite precision to F_σ and closed sets in the weak topology, together with a concrete, testable regularity condition that restores consistent testing for conditional independence on Polish spaces. The explicit use of Polish-space properties and the avoidance of circularity via the equicontinuity hypothesis are strengths that make the results falsifiable and potentially useful for clarifying the scope of existing impossibility theorems.

major comments (2)
  1. [Abstract and §3] Abstract and §3 (main characterization): the necessity direction of the iff statement—that consistent finite-precision tests exist only when both hypotheses are F_σ—relies on the fact that open decision regions in the sample space induce F_σ sets under the weak topology; the manuscript should explicitly verify that this direction does not require additional separation axioms beyond the Polish assumption already stated for W.
  2. [§4] §4 (conditional-independence application): the claim that conditional independence is not consistently testable from finite-precision data when the conditioning space has no isolated points is presented as a strengthening of prior results; the argument appears to use the F_σ criterion directly, but the manuscript should confirm that the no-isolated-points condition indeed prevents both H0 and H1 from being F_σ simultaneously in the relevant weak topology.
minor comments (2)
  1. [§2] The definition of finite-precision tests via open decision regions is introduced early but the precise sample-space topology (product topology on the observation space) should be stated explicitly in the first paragraph of §2 to avoid ambiguity when the observations are real-valued.
  2. [Throughout] Notation W := H0 ∪ H1 is used throughout; a single sentence clarifying that the weak topology is the relative topology induced from the space of all probability measures would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We are grateful to the referee for the positive assessment and the detailed suggestions for improvement. We have revised the manuscript to address the two major comments, providing the requested explicit verifications. Below we respond point by point.

read point-by-point responses
  1. Referee: [Abstract and §3] Abstract and §3 (main characterization): the necessity direction of the iff statement—that consistent finite-precision tests exist only when both hypotheses are F_σ—relies on the fact that open decision regions in the sample space induce F_σ sets under the weak topology; the manuscript should explicitly verify that this direction does not require additional separation axioms beyond the Polish assumption already stated for W.

    Authors: We thank the referee for pointing this out. In the revised manuscript, we have added a paragraph in Section 3 immediately following the statement of the main theorem. This paragraph explicitly notes that since the sample space is Polish, the space of probability measures equipped with the weak topology is also Polish (hence metrizable and separable), and the preimages under the map from measures to their finite-precision observations are open in the sample space topology, which directly implies they are F_σ in the weak topology without invoking any separation axioms beyond those already ensured by the Polish assumption on W. This clarification confirms the necessity direction holds under the stated assumptions. revision: yes

  2. Referee: [§4] §4 (conditional-independence application): the claim that conditional independence is not consistently testable from finite-precision data when the conditioning space has no isolated points is presented as a strengthening of prior results; the argument appears to use the F_σ criterion directly, but the manuscript should confirm that the no-isolated-points condition indeed prevents both H0 and H1 from being F_σ simultaneously in the relevant weak topology.

    Authors: We agree that an explicit confirmation strengthens the presentation. In the revised version of Section 4, we have expanded the proof of the impossibility result to include a direct argument showing that if the conditioning space has no isolated points, then in the weak topology on the union W = H0 ∪ H1, at least one of H0 or H1 fails to be F_σ. Specifically, we construct a sequence of measures converging weakly to a limit point that would require the set to contain a non-F_σ subset if both were F_σ, leveraging the fact that without isolated points, the conditional distributions can approximate any weak limit without being countable unions of closed sets. This confirms the no-isolated-points condition prevents simultaneous F_σ membership. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivations use external topological facts

full rationale

The central characterization (consistent finite-precision test exists iff both hypotheses are F_sigma in the weak topology on W) follows from equivalences between finite-sample/asymptotic/uniform error control (shown via standard arguments for open decision regions) plus known separation properties of F_sigma sets in Polish spaces under weak convergence. No self-definitional loops, fitted inputs renamed as predictions, or load-bearing self-citations appear. The equicontinuity assumption is stated to be independently finite-precision testable. All steps rest on external measure-theoretic and topological results rather than reducing to the paper's own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The results rest on the standard assumption that sample spaces are Polish (to guarantee the weak topology behaves well) and on the new definition of finite-precision tests as open decision regions. No free parameters are fitted; the equicontinuity assumption is introduced as a domain restriction that is itself testable.

axioms (2)
  • domain assumption Sample spaces are Polish spaces
    Invoked to ensure the weak topology on probability measures is metrizable and to support F_sigma characterizations.
  • ad hoc to paper Finite-precision tests have open decision regions in the sample-space topology
    This modeling choice is introduced in the paper to represent finite-precision recording.
invented entities (1)
  • finite-precision test no independent evidence
    purpose: A hypothesis test whose decision regions are required to be open sets to model finite-precision observations
    New concept introduced to accommodate the finite-precision constraint; no independent empirical evidence is provided beyond the modeling motivation.

pith-pipeline@v0.9.0 · 5604 in / 1382 out tokens · 47837 ms · 2026-05-16T12:52:49.628570+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.