Contribution
The paper extends the single-system programme of papers I–III (LIT-343, LIT-335, LIT-299) to collections of systems. It proposes a category Sys whose arrows are “physically meaningful” translations of physical quantities, with local languages giving a functor L : Sys → Loc
It proposes rules (Def 3.2) for realising Sys in a category M(Sys) of topoi with left-exact functors as arrows. The rules are:
- a pull-back of physical quantities along the triple (τ(j), φ(j), β(j));
- a pull-back of propositions, φ(j)⁻¹(τ(j)(K)).
It then tests the rules on classical physics, where they work, and on quantum theory, where they work for disjoint sums and fail for tensor-product composites.
Key insight
In classical physics a composite’s state space is a categorical product S₁ × S₂. Its projections pull a constituent’s quantities back to the composite exactly, so “the energy of particle 1” means the same function in both descriptions.
In the quantum topos there is no such projection Σ_{H₁⊗H₂} → Σ_{H₁}. The context category V(H₁⊗H₂) contains many contexts that are not of the form V₁⊗V₂ (“operator entanglement”). The best available pull-back of δ̆(A₁) therefore agrees with δ̆(A₁⊗1̂) only on contexts coming from V(H₁). Elsewhere it approximates A₁ inside V_W, the largest H₁-part of the context. So the composite’s topos “knows” more contexts than any translation from a constituent can reach.
Assumptions
- The collection of systems. Sys is small. It is closed under sub-systems, composites and disjoint sums. All its systems share the ground types Σ and R and differ only in their function symbols (§2.1).
- Disjoint sums and composites. For disjoint sums, F_{S₁⊔S₂}(Σ,R) ≅ F_{S₁} × F_{S₂} (2.1, 2.12). For composites no such decomposition is postulated; only the translations A₁ ↦ A₁⋄1 exist.
- Axioms. ⊔ and ⋄ are associative and commutative up to isomorphism (2.9–2.10, 2.13–2.14). They satisfy distributivity (S₁⊔S₂)⋄S ≅ (S₁⋄S)⊔(S₂⋄S) (2.16), a trivial system 1 with S⋄1 ≅ S (2.17), and an optional empty system 0 (2.18–2.19). Coherence axioms are “very plausible” and not checked. Bifunctoriality holds only if arrows are restricted to sub-system and composite arrows (§2.2.4).
- Every system has a trivial quantity. Each language contains I with ∀s̃₁ ∀s̃₂ I(s̃₁) = I(s̃₂) (2.8).
- Arrows between topoi. The arrows of M(Sys) are left-exact functors, chosen so that monics, and hence propositions, can be pulled back (§3.1.2).
- Quantum setting. Separable Hilbert spaces. For disjoint sums, only block-diagonal operators Â₁⊕Â₂ are used (fn. 20). For composites, H₁⊗H₂ with the translation Â₁ ↦ Â₁⊗1̂.
Key results
- Classical physics (§2.4).
- Construction. Every arrow j gives a map σ(j) : Σ_{S₁} → Σ_S, the inclusion into a disjoint union or the projection from a product. Pulling back along it, σ(L(j))(f) = f ∘ σ(j) (2.33), makes the square (2.30) commute.
- Trivial and empty systems. The trivial system goes to {∗} (2.26) and the empty system to ∅ (2.27).
- Strength. This is exact and elementary.
- Rules (Def 3.2).
- Objects and quantities. The state object, the quantity-value object, quantities as arrows and propositions as sub-objects are as in I–III.
- States. Truth objects instead of microstates (Rule 3). The unit of M(Sys) is Sets (Rule 4, (3.14)).
- Arrows. Each arrow gets a triple (τ(j), φ(j), β(j)). The pull-back is φ(L(j))(μ) = β(j) ∘ τ(j)(μ) ∘ φ(j) (3.5). The target condition, which “may, or may not” hold, is φ(L(j))(A_{φ,S}) = [L(j)(A)]_{φ,S₁} (3.18).
- Conjectures. For a sub-system, φ(j) is monic (Rule 6a). It is also conjectured that R_{S₁} ≅ τ(j)(R_S) and that φ(j) is epic for epic j (Rule 6b).
- Geometric morphisms from functors (Thm 4.1, cited from Mac Lane–Moerdijk). A functor ϕ : C → D induces ϕ* : Sets^{D^op} → Sets^{C^op}, F ↦ F ∘ ϕ^op. This is left exact, with left adjoint ϕ_!.
- Quantum disjoint sum (§4.2).
- The functor. m : V(H₁) → V(H₁⊕H₂), V ↦ V ⊕ ℂ1̂_{H₂}, which is order-preserving. μ* is the induced inverse image.
- Lemma 4.2. δ(Â₁⊕Â₂){V₁⊕V₂} = δ(Â₁){V₁} ⊕ δ(Â₂){V₂}. The proof is correct: projectors in V₁⊕V₂ split, so inner daseinisation of each spectral projector splits. It is applied with V₂ = ℂ1̂{H₂}, where δ(Â₂) = max sp(Â₂)·1̂ (4.21), even though ℂ1̂ is not an object of V(H₂). The formula is still right.
- Spectra. Σ_{V⊕ℂ1̂} ≅ Σ^{H₁}V ∪ {λ₀}, where λ₀ is the one extra character that is 1 on 0̂⊕1̂. φ(i) is the inclusion, and β(i) is the evident isomorphism μ*ℝ^⪰ ≅ ℝ^⪰{H₁} (4.20).
- Result. The pull-back of δ̆(⟨A₁,A₂⟩) is δ̆(A₁) (4.24), so (3.18) holds for i₁ and i₂. I checked this; it follows from Lemma 4.2 and the definitions.
- Quantum composite (§4.3).
- The functor. n : V(H₁⊗H₂) → V(H₁), W ↦ V_W, the largest V with V⊗ℂ1̂ ⊆ W. This is well defined: V_W⊗1̂ = W ∩ (B(H₁)⊗1̂). V(H₁) adjoins ℂ1̂_{H₁}, and Σ and ℝ^⪰ are extended to it (a single character; ℝ).
- The pull-back. τ(p₁) = ν*, φ(p)W(λ) = λ|{V_W} (4.40) and β(p)(α)(W′) = α(V_{W′}) (4.39). The resulting arrow is, at each W, the Gel’fand transform of δ(Â₁)_{V_W} ⊗ 1̂ (4.44)–(4.45). It agrees with δ̆(A₁⋄1) at every W of the form V₁⊗ℂ1̂ (fn. 27).
- Where it fails. “It is relatively easy to show” that δ(Â₁⊗1̂)W ≠ δ(Â₁){V_W} ⊗ 1̂ in general (4.46). Hence φ(L(p))(δ̆(A₁)) ≠ δ̆(A₁⋄1) (4.47), and (3.18) fails. “There appears to be no operator B̂” whose arrow equals the pull-back.
- Status. Neither (4.46) nor the non-existence claim is proved. Fn. 25 adds that it is open whether δ(Â₁⊗1̂)W = δ(Â₁){V₁}⊗1̂ even for W = V₁⊗V₂ with V₂ non-trivial.
- Authors’ summary. The scheme holds for classical physics and “quantum physics ‘almost’ does” (§5).
- Speculation (§5).
- A “gros topos”. A single topos Sh(M(Sys)) for all systems of a theory-type, if M(Sys) is a site. It is posed, not attempted.
- Interpolating functors. An “interpolating chain” of functors between a neo-realist topos and an instrumentalist one.
- Quantum gravity. A claim that quantum gravity must be built in a topos U ≠ Sets.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Systems form a category Sys, symmetric-monoidal under ⋄ and under ⊔, with arrows given by language translations | assertion / axioms (“not cast in stone”) | §2.2.4. Coherence is not checked; bifunctoriality holds only for restricted arrows |
| C2 | Classical physics satisfies the rules exactly, via pull-back along σ(j) | proof (elementary) | §2.4, (2.33), (3.9)–(3.10) |
| C3 | Propositions pull back along any triple whose functor preserves monics | proof (standard topos facts) | §3.1.2 |
| C4 | The unit of M(Sys) is Sets, because V(ℂ) has one object | argument inconsistent with the series’ definition of V(H), which excludes ℂ1̂ | Rule 4, (4.9); see corrections |
| C5 | Quantum disjoint sums satisfy (3.18): the pull-back of δ̆(A₁⊕A₂) is δ̆(A₁) | proof (checked) | Lemma 4.2, (4.18)–(4.24) |
| C6 | For composites, the natural pull-back of δ̆(A₁) equals δ(Â₁)_{V_W}⊗1̂ at each W, agreeing with δ̆(A₁⋄1) exactly on contexts V₁⊗ℂ1̂ | proof (construction checked) | (4.29)–(4.45), fn. 27 |
| C7 | This pull-back differs from δ̆(A₁⋄1) in general, and no operator B̂ has it as δ̆(B) | assertion (“relatively easy to show”; “there appears to be no”) | (4.46)–(4.47), §4.3 |
| C8 | The failure reflects “operator entanglement”: V(H₁⊗H₂) has contexts not of the form V₁⊗V₂ | informal argument (interpretation) | §4.3; the premise about contexts is true, and the causal link is a gloss |
| C9 | The translation is “as good as possible” | assertion | §4.3, “Comments on these results” |
| C10 | Quantum gravity should be formulated in a topos other than Sets | assertion (programmatic) | §5 |
Method
Categorical axiomatics (monoidal categories, translations of local languages after Bell), geometric morphisms induced by functors between base categories (Mac Lane–Moerdijk VII), and the daseinisation machinery of papers II–III.
Concepts
- Sys. The category of systems. An arrow j : S₁ → S is a translation L(j) of S’s quantities into S₁’s.
- Disjoint sum ⊔ and composite ⋄. The two monoidal structures on Sys. The unit of ⋄ is the trivial system 1; the unit of ⊔ is the empty system 0.
- Topos realisation. A triple (ρ_{φ,S}, L(S), τ_φ(S)) per system, plus a triple (τ(j), φ(j), β(j)) per arrow.
- Translation representation φ(L(j)). The pull-back of arrows Σ_S → R_S to arrows Σ_{S₁} → R_{S₁}.
- Operator entanglement. The authors’ name for the fact that V(H₁⊗H₂) is richer than contexts of the form V₁⊗V₂.
- Augmented context category V(H)*. V(H) with ℂ1̂ readmitted as a bottom element.
- Gros topos. A single topos of sheaves over a site of topoi, housing every system of a theory-type.
Connections
- Papers I–III (LIT-343, LIT-335, LIT-299). This paper uses I’s languages and translations and III’s δ̆ and ℝ^⪰. It repeats II’s truth objects as Rule 3.
- Isham–Butterfield 1998 (LIT-325). Its sequels are cited ([7]–[11]) for the physical meaning of contexts. Nothing on composites is taken from them.
- Abramsky–Brandenburger (LIT-016), THEORY-012, LIT-277, LIT-278. No relation.
- Composites without Bell. The paper treats composite quantum systems but never Bell non-locality, product measurement covers, or no-signalling. Those are exactly what LIT-016 adds for composites (measurement covers, locality as factorisability).
- Where the failure sits. The composite failure here concerns translating a constituent’s quantity arrows into the composite topos. It does not concern contextuality or non-local correlations. The “entanglement” in the title of §4.3 is about the context poset of H₁⊗H₂, not about any state.
- THEORY-017 (a tensor factorisation is extra data). A related point. Here the factorisation H₁⊗H₂ is given, and the constituent quantities translate faithfully only on the sub-poset of contexts V₁⊗ℂ1̂ that the factorisation picks out. Everything else in V(H₁⊗H₂) is invisible to the constituent. This is the topos-side face of THEORY-017’s claim that parts are supplied from outside the space.
- Mereology entries. None in the record engages physical composition in these terms. I checked no specific entry for a direct link, and none is claimed.
Bearing on the record
- The map’s role (§0, row 8: instantiate “their presheaf … as the Topos of Bricks”).
- This is the one paper in the series about building a whole from parts. Its result is a caution: the programme’s quantum topoi do not compose along ⊗ in the way its rules require (C6–C7). Only disjoint sums compose cleanly (C5).
- If the “Topos of Bricks” is meant to assemble a concept system from brick-wise topoi, or to translate a sub-model’s concept operators into a larger model, this paper predicts a mismatch. The larger model has contexts that mix the parts (not of the form V₁⊗V₂), and on those contexts translated quantities are not the larger model’s own quantities.
- If instead the Topos of Bricks is a single presheaf topos over one context category, this paper is not needed. The owner’s documents should say which is meant before citing it.
- The map’s §2 item 5 (
𝒜^∞, “stable across every adequate model’s concept algebra”). The paper’s own open question, a “gros topos” housing all systems of a theory-type (§5), has the same shape as a cross-model object. It is posed, not solved. Citing this paper for a cross-model topos would overstate it. - The rest of §2.1. Nothing here bears on ultrafilters, Kochen–Specker, or topes.
- ML practice. It carries nothing for ML practice.
- For filing. Tags as proposed.
quantum-foundationsfirst.mereologyis justified by the paper’s subject (composition and individuation of systems).mathematicsandlogiccover the categorical and linguistic machinery.philosophy-of-sciencecovers the realism and instrumentalism discussion and idealised systems.anthology-candidateis not proposed.
Limitations
- The rules are provisional. The rules and axioms are self-described as “partly ‘experimental’”, and several (Rule 6, (3.20), the epic conjecture) are conjectures.
- The main negative result is not proved. Neither (4.46) nor “no B̂” is shown, and fn. 25 leaves open an easier case.
- The trivial-system derivation contradicts the series’ own convention (C4).
- The composite realisation does not quite fit the rules. It needs an augmented category V(H₁)*, and several notational slips obscure the direction of the arrows (see corrections).
- Only two arrow types are treated for quantum theory (disjoint-sum inclusions and composite projections). Genuine sub-systems (Rule 6a) are not realised for any quantum example.
- Physically thin. There is no dynamics and no states on composites (how T^ψ for an entangled ψ relates to the constituents’ truth objects is not discussed). Nothing is said about locality or about how partial trace appears.
Open questions
- Prove (4.46), and decide whether any B̂ realises the pulled-back arrow, i.e. whether the failure is a theorem.
- Is there a different context category, or a different choice of functor V(H₁⊗H₂) → V(H₁)* or of β, under which tensor-product composites satisfy (3.18)?
- How do truth objects of entangled states restrict to the constituents? What is the topos analogue of the partial trace?
- Can M(Sys) be made a site so that a “gros topos” exists for quantum systems (§5)?
Corrections to the seeded skim
- Seeded from metadata. The seed’s summary is accurate as far as it goes. The text adds the finding: the scheme works for disjoint sums, and composites fail the paper’s own commutativity condition (3.18); the seed does not say so.
- The trivial system is internally inconsistent. Rule 4 postulates that the trivial system’s topos is Sets. It derives this (4.9) from “exactly one abelian subalgebra of B(ℂ) … namely ℂ itself”. But paper II excludes the trivial algebra ℂ1̂ from V(H) by definition (LIT-335, §2.1.1), and this paper says so itself (§4.3: “The trivial algebra ℂ1̂_{H₁} is not an object in the category V(H₁)”).
- Consequence. Under the series’ convention V(ℂ) is empty, and the presheaf topos over the empty category is the degenerate one-object topos, not Sets. Rule 4 holds only if ℂ1̂ is readmitted, as the paper does ad hoc elsewhere with the augmented category V(H₁)*.
- This is my observation; the paper does not note it.
- Notational slips that could mislead a reader checking the construction:
- Wrong direction. (4.19) calls μ* “the functor … τ_φ(j) : τ_φ(S₁) → τ_φ(S)”, which is the wrong direction; μ* goes from the sum’s topos to S₁’s.
- Composition order. (4.23) writes the pull-back as φ(i) ∘ μ*(δ̆) ∘ β_φ(i), the reverse of the order in (3.5) and (3.17).
- Undefined symbol. The geometric morphism for the composite is said to be “induced by π” when the functor defined is n (4.31).
- Target mismatch. The functor used for the composite has target V(H₁)*, not V(H₁). So τ_φ(p₁) is not literally a functor out of τ_φ(S₁) = Sets^{V(H₁)^op} as Rule 5(c) requires.
- Tag proposal: add
mereology, whose blurb is “composition, individuation of systems”. The paper is about sub-systems, composites and what counts as a system (§2.1–§2.2.3), and that is its main subject after quantum theory. Addphilosophy-of-sciencefor §2.1 and §5, on idealised systems and neo-realist versus instrumentalist interpretations. Dropcontextuality: the paper does not treat contextuality as such. The context category appears only as V(H₁⊗H₂) being richer than V(H₁)×V(H₂), which is not something a browser of contextuality would come for. Keeplogic, for translations of local languages. - Identifiers check against the arXiv listing: journal-ref J. Math. Phys. 49:053518 (2008); DOI 10.1063/1.2883826; “38 pages, no figures”. arXiv v1 is 7 Mar 2007.
published: 2008-05-01is the journal issue (flagged only).