The concept of anima between homotopic topology and Neoplatonism — v2, all module results complete
An exploratory project combining mathematical computations on Clausen's new Weil-Moore anima with a terminological-historical and philosophical analysis of the concept anima — carried out on a local AI ecosystem without cloud connectivity.
Dustin Clausen's paper Weil-Moore anima (arXiv:2605.11950, May 2026) introduces a concept whose name is already remarkable: anima — Latin for soul. In Remark 1.1, Clausen explains the naming with two independent motivations, the second of which explicitly references the Neoplatonic tradition: an anima is what remains of a CW-complex once one removes its worldly (point-set) incarnation.
That was the starting point of this project. The questions were: Is this structurally substantial or mere terminological decoration? And, as an independent mathematical thread: what are the concrete homotopy groups of the archimedean Weil-Moore anima Xℂ, and what general principle underlies them?
Both were investigated using local language models — without cloud connectivity, on a Mac Studio M3 Ultra with 256 GB unified memory. The models received the complete primary texts as context. An adversarial verification round independently checked all central mathematical claims.
Primary texts used: Clausen, Weil-Moore anima (2026) · Scholze, Lectures on Condensed Mathematics (2026) · Scholze, Lectures on Analytic Geometry (2026) · Beilinson, Topological ε-factors (2007) · Clausen–Mathew, Hyperdescent and étale K-theory
The search for the origin of the concept anima leads directly to Beilinson (2007). In Section 1.1 of his paper on topological ε-factors he writes: “The homotopy world is a (yet unnamed) animation of the homotopy category that hovers inbetween the latter and highly non-canonical model categories.” The words “yet unnamed” are decisive: Beilinson describes the complete ∞-categorical framework of homotopy theory and calls it intuitively “animation”, without giving it a proper name. Clausen and Scholze give it that name.
In §3.8 of the same paper, “animation” appears in a second, more precise sense: it is the process by which a numerical identity (the Euler characteristic formula) is replaced by a fuller homotopy-theoretic structure — the formula “now lives in an ∞-category rather than just a number”. This is the same movement that Clausen's condensed anima performs: an abstract structure is “animated” by being placed in its natural ∞-categorical context.
“An anima is just a homotopy type, but viewed as an object in its natural ambient ∞-category rather than its 1-categorical shadow (the classical homotopy category). Why the name ‘anima’? There is a two-fold motivation. First, one has the following intuition described by Beilinson in [4]. A set is static in the sense that its points are stuck where they are: they cannot move around within the set. But if we have an anima A, with underlying set π0A, then a point of A is free to move around within its connected component: as Beilinson says, the points of A are naturally ‘animated’. Second, an anima is something like a soul. If we have, say, a CW complex, then its associated anima is what remains of the CW complex when it is stripped of its worldly (point-set) incarnation.”
— Clausen, Weil-Moore anima (2026), Remark 1.1
An ordinary anima is a homotopy type, viewed as an object in its ∞-category rather than the 1-categorical homotopy category. A condensed anima additionally carries topological structure — independently of the homotopy-theoretic one. The key example: as condensed anima, K(ℝ, n) ≠ K(ℝδ, n) and K(ℝ, n) ≠ *, even though ℝ as a topological space is contractible. The condensed structure preserves the topological information of ℝ as an abelian topological group and makes it cohomologically accessible.
In §3.8 of the same paper, “animation” appears in a second, more precise sense:
“One can view ε†(F) as an animation of Kashiwara’s characteristic cycle CC(F); therefore the identification [RΓ(X,F)] = tr ε(F) of 4.1 becomes an animation of the Dubson–Kashiwara formula.”
— Beilinson, Topological ε-factors (2007), §3.8
Here “animation” is the process by which a numerical identity (the Euler characteristic formula) is replaced by a fuller homotopy-theoretic structure — the formula “now lives in an ∞-category rather than just a number”. This is the same movement Clausen’s condensed anima makes: an abstract structure is “animated” by placing it in its natural ∞-categorical context.
For Xℂ one has π1 = Wℂ = ℂ×. The maximal compact subgroup of the identity component is K° = S1 = U(1). Lemma 7.11 gives:
The algebraic core: Homcont(S1, ℝ/ℤ) = ℤ (classical result: continuous characters of the circle). Then Sym2ℤ(ℤ) = ℤ·(e⊗e) ≅ ℤ. The Pontryagin dual: ℤ∨ = Hom(ℤ, ℝ/ℤ) = ℝ/ℤ.
For Xℝ, the group Gal(ℂ/ℝ) = ℤ/2 acts on K° = S1 by z ↦ z̄ = z-1. On Λ = ℤ this induces negation n ↦ -n. The decisive step: on Sym2ℤ(ℤ) this action is trivial, since the non-trivial element acts as (-1)·(n⊗n) = (-n)⊗(-n) = n⊗n. Squaring cancels the sign.
From Corollary 3.7: H3(BWℂ; ℝ/ℤ) = H4(Bℂ×; ℤ). Since Bℂ× ≃ ℂP∞ and H4(ℂP∞; ℤ) = ℤ, the k-invariant space of Xℂ is canonically isomorphic to ℤ. The k-invariant κ, classifying the second Postnikov step, is the generator — the precise condensed analogue of the classical Hopf element η.
| Classical S2 | Condensed Xℂ | |
|---|---|---|
| π1 | trivial | Wℂ = ℂ× |
| π2 | ℤ | ℝ/ℤ = ℤ̂ |
| k-invariant space | H4(ℂP∞; ℤ) = ℤ | H3(Bℂ×; ℝ/ℤ) = ℤ |
| k-invariant | Generator η (Hopf) | Generator κ |
| Euler char. χ | 2 | 2 (Example 8.10) |
| Cohom. dim. | 2 | 2 (Theorem 8.1) |
The most striking pattern in the computations is the transformation ℤ ↦ ℝ/ℤ through Pontryagin dualisation. The principle operates on the unstable homotopy groups πn(S2), not the stable stem groups πns: π2(S2) = ℤ (unstable) ⟶ ℝ/ℤ = π2(Xℂ). The stable stem π2s = ℤ/2 would have predicted ℤ/2 — incorrect.
ℝδ (the real numbers with discrete topology) is as an abstract abelian group a ℚ-vector space. ℚ-vector spaces are injective as ℤ-modules. It follows that Hn(K(ℝδ, 1); ℤ) = 0 for all n ≥ 1, and analogously all higher Ext groups vanish. K(ℝδ, n) is cohomologically trivial — contractible. By contrast, K(ℝ, n) with the genuine topology carries rich cohomology. Forgetting the topological structure through discretisation destroys everything.
Clausen writes in Remark 1.5: “The question of how to calculate the higher homotopy groups of XQ is tantalizing… we do not investigate this in detail.” For π3, the Serre spectral sequence for the Postnikov fibration
yields E2-terms E2p,q = Hp(Bℂ×; Hq(Kcond(ℝ/ℤ,2); ℝ/ℤ)). For q = 0 and q = 2 terms contribute; for q > 2 the cohomology Hq vanishes (since Kcond(ℝ/ℤ,2) is 2-connected). From Lemma 4.12: π3(Xℂ) = (H4(Xℂ(2),ˆ; ℝ/ℤ))∨, where Xℂ(2) is the 2-stage Postnikov tower. The duality principle and comparison with π3(S2) = ℤ suggest π3(Xℂ) = ℝ/ℤ. A complete proof requires the condensed cohomology of B2(S1).
The canonical functor X ↦ |X|, which forgets the condensed structure and returns the “bare anima”, is not fully faithful: distinct morphisms between condensed anima can induce the same morphisms between the underlying anima. This is a category-theoretic result with a consequence: the bare anima |X| is preserved as a structural skeleton across different condensed incarnations of X and is not uniquely determined by them. The condensed S1 and |S1| are different as objects — Scholze: “S1 is a physical circle, while |S1| is some ghostly appearance of a point with an internal automorphism (the anima ∗/ℤ).”
All three central mathematical claims were independently checked by Module B2 (deepseek-r1:70b, chain-of-thought). All three were confirmed.
Three philosophical parallels were examined. The goal in each case: to distinguish structural isomorphism from terminological coincidence, and to name the break points at least as precisely as the parallels.
In Plotinus (Enneads IV.8), ψυχη is what remains of a being when τò υλη (matter) is removed — the principle of self-motion (αυτοκíνητον) that preserves the essential structure. In Clausen–Scholze, the anima is what remains of a CW-complex when point-set topology (“worldly incarnation”) is removed. The structural parallel: in both cases there is an operation of “stripping” that reduces a rich concrete structure to a more abstract one — and the more abstract one carries its own name.
Beilinson's “animated points” — points free to move within their connected component — structurally corresponds to the Plotinian principle of self-motion.
Break point: The Plotinian soul exists before material embodiment as a metaphysical entity; the anima is an abstraction arrived at by removing the topology. The ontological direction is reversed.
Iamblichus disputes Plotinus' thesis of an “undescended part” of the soul. For Iamblichus, the soul descends completely into matter. The mathematical analogue would be: there is no “undescended remainder” — the bare anima |X| would be completely determined by the condensed structure X.
But this is precisely false. Since the functor X ↦ |X| is not fully faithful, there is a permanent “Plotinian remainder”: different condensed anima can have the same bare anima |X|. |X| survives every condensed incarnation. Concretely: |S1| remains the same anima, regardless of what condensed structure one places on S1.
Scholze's quote provides the most vivid mathematical model for the philosophical distinction: “S1 is a physical circle, while |S1| is some ghostly appearance of a point with an internal automorphism (the anima */ℤ).” (Scholze, Lectures on Analytic Geometry, Lecture XI)
S1 as condensed anima is the “materially incarnated” version — with topological structure, physically tangible. |S1| as bare anima is the “pure soul” — only the homotopy-theoretic essence, π1 = ℤ and nothing more. This is a mathematically precise instance of the Plotinian soul/matter distinction.
Damascius' apophatic method: every attempt to determine the One positively collapses. The mathematical analogue: K(ℝδ, n) ≃ * — discretisation (removal of all topological structure) leads to the point.
Break point 1 (intention): For Damascius, apophatics is an approach to the Absolute; the collapse is not the goal but a pointer to the limits of positive determinability. For Clausen–Scholze, K(ℝδ, n) ≃ * is a negative result — a warning signal against information loss, not an approach to something.
Break point 2 (universality): Damascius' method is universally applicable to all positive determinations. Mathematical discretisation is selective: it removes topological structure, but leaves algebraic structure (the group structure of ℝ) intact. “Mathematical apophatics” would only be possible selectively — not universally in the Damascian sense.
On one point the two main models of the project disagree — and this disagreement is itself a result.
The conflict is not arbitrary. deepseek-r1:70b works with chain-of-thought reasoning and tends toward formal rigour; the break points (selective vs. universal; approach vs. negative result) are from this perspective weighty enough to reject the analogy. qwen3:235b-a22b has access to the full Neoplatonism context and evaluates the structural operation itself as a genuine similarity, even if the framework conditions differ.
Both assessments are internally consistent. The question of whether structural agreement in the operation (stripping → collapse) constitutes genuine resonance even when the intention and scope differ is a methodological question this project cannot answer definitively. The conflict marks exactly the point where a formal framework for structural analogies between mathematics and philosophy is lacking.
The two project threads converge on one point: the condensed anima is an object that on two independent levels navigates between “too much” and “too little structure.”
Topologically: between the concrete CW-complex (too many non-canonical details) and the abstract homotopy type (too little, because the topology of the coefficient groups is lost). The condensed anima keeps both separate and independent — this is mathematically the content of the non-collapse property K(ℝ, n) ≠ K(ℝδ, n) ≠ *.
Arithmetically: between BWK (wrong cohomological dimension, archimedean asymmetries) and full Galois cohomology (similar defects). The Weil-Moore anima XK repairs both by adding higher homotopy groups whose structure reflects the compact/discrete duality principle.
Pontryagin duality ℤ ↔ ℝ/ℤ is precisely this transition: the discrete, abstract homotopy group is “animated” through condensation — it receives a compact, topologically rich structure. This is the mathematical content of Beilinson's “animation” and Clausen's “anima”, and it is structurally the same movement that Plotinus describes with the transition from matter to soul — with the important caveat of the reversed ontological direction.
Does the compact/discrete duality principle hold for all n and all archimedean Weil-Moore anima? This requires the complete calculation of the Postnikov towers of Xℂ, in particular the condensed cohomology groups H*(Kcond(ℝ/ℤ, n); ℝ/ℤ). Case 3 of the principle (finite, not self-dual) does not occur in the known range and remains untested.
How deep does Clausen and Scholze's knowledge of the Neoplatonic tradition run? The formulations “worldly incarnation” and “ghostly appearance” are too specific to be coincidental. An investigation of earlier manuscripts and correspondence could reveal further connections.
Is there a formal framework that describes structural analogies between mathematical objects and philosophical concepts precisely — beyond metaphorical similarity, but without categorical fallacies? The model conflict in ADV-4 shows that without such a framework, the answer to “what constitutes structural resonance” depends on the evaluation criteria of the respective model. This is the actual methodological limit of this project.
The project ran entirely locally on a Mac Studio M3 Ultra (256 GB unified memory, 80-core GPU, macOS Sequoia) with Ollama 0.24.0. Models received the complete primary texts as context.
| Model | Role | Parameters |
|---|---|---|
| qwen3:235b-a22b | Terminological archaeology (A1), Neoplatonism analysis (A3), Synthesis | 46K tokens input · num_predict 18K · thinking mode active |
| deepseek-r1:70b | Postnikov computations (B1), π3 problem + adversarial (B2) | 65K context · chain-of-thought · 65K tokens input |
| DSV4-Flash (GGUF) | Planned for full-text ingestion; not used | 153 GB Q4 · GPU memory error at 99K-token KV-cache |
The execution was not without problems: a system crash (GPU memory error with too-large KV-cache), a timeout from too small num_predict with active thinking mode (qwen3 consumed the entire token budget for internal reasoning tokens without visible output — solution: num_predict: 18000), and several syntax errors through mathematical notation in Python f-strings. All problems were diagnosed and resolved. Total model runtime: approximately 4 hours.
All raw results are at /Volumes/CLAUDE-DATA/Anima-Project/ — five .md files with the complete model responses.