Algebraic Geometry: Spaces of Physical Possibility

1 Introduction

1.1 The series and its modular ladder

The present paper is the third module of a seven-paper program, A Math$\to$Physics Representation Library. The program takes as its point of departure the observation that many physical quantities are not merely described by mathematics but are literally obtained as realizations of structured mathematical information [1]. Rather than assert one universal functor from all of mathematics to all of physics — an object we argue in 12.1 is unlikely to be well defined — the program is deliberately modular: each mathematical domain is formalized as its own physical-representation module, each module is defensible on its own terms, and the modules are composed hierarchically so that new structure and new emergent physical content appear at each level.

The ladder proceeds as follows.

• Part I (Foundations). Introduces the category $\mathsf{Dom}$ of mathematical domains, the categories ${\mathsf{Math}}_d$ of structures in each domain, the target $\mathsf{Phys}$ of physical representations, the notion of a representation entry $E=(M,P,\tau ,\sigma )$, the representation prestack and representation stack, the four foundational axioms (Realization, Equivalence, Locality/descent, Decomposition), and the epistemic status system [S]/[H]/[P]. Its central formula is the realization pipeline

  $$\begin{array}{c}{\mathrm{Obs}}_{\alpha }(M)=\mathrm{Obs}({\mathrm{Real}}_{\alpha }(\Phi (M))).\end{array}$$

  (1)

• Part II (Motives, periods, amplitudes). Makes the abstract realization channels ${\mathrm{Real}}_{\alpha }$ of Part I concrete: Betti, de Rham, Hodge, and étale realizations of motives; the period datum $\Pi =(X,D,\omega ,\Gamma )$ with $\mathrm{per}(\Pi )={\int }_{\Gamma }\omega$; the motivic amplitude object ${\mathcal{A}}^{\mathfrak{m}}$ with $\mathrm{per}({\mathcal{A}}^{\mathfrak{m}})=\mathcal{A}$; and the coalgebraic anatomy encoded by the coproduct $\Delta$ and cobracket $\delta$. Observables acquire an internal decomposability into primitive informational pieces.

• Part III (this paper). Supplies the geometric stage. Part II’s motives and periods do not live in a vacuum: a period datum requires $X$ to be an actual algebraic or analytic space, and a family of amplitudes over kinematic space is a variation of Hodge structure over a moduli space. We formalize schemes, moduli spaces and stacks, Hodge theory and Gauss–Manin transport, and positive geometries, and we show that Part II’s per-object pipeline extends to a family pipeline fibered over the geometry.

• Parts IV–VII. Algebraic topology (conserved global information), category theory and homotopy type theory (composition and identity), sheaves/stacks/gauge redundancy/quantum high availability, and finally a synthesis paper recomposing the ladder. Part III sets up Part IV by producing the varieties and moduli whose topological invariants Part IV will study.

1.2 Thesis of Part III

Our thesis, following the primary source’s algebraic-geometry appendix, is compact:

Algebraic geometry is the geometry of physical possibility.

The word “possibility” is meant in a precise, layered sense that the machinery of algebraic geometry makes available:

1. Solutions as points. The set of states satisfying a system of polynomial constraints is an affine variety $V(\mathfrak{a})$; its coordinate ring $R/\mathfrak{a}$ is the algebra of on-shell observables. [S] as algebra; the reading “measurements generate the space of possibilities” is [H].

2. Infinitesimal possibility. Passing from varieties to schemes $\mathrm{Spec}R$ retains nilpotent (infinitesimal, off-shell, ghost-like) directions that varieties discard; possibility acquires a tangent/deformation structure.

3. Families of possibility. A moduli space parametrizes physically distinct configurations; the space itself, its singularities, and its compactification become physical objects (vacua, thresholds, asymptotic sectors).

4. Possibility modulo redundancy. A moduli stack $[X/G]$ remembers the gauge automorphisms (stabilizers) of each configuration, distinguishing genuine physical states from redundant descriptions — the geometric incarnation of Part I’s Equivalence axiom.

5. Possibility varying analytically. A variation of Hodge structure equips a family with a flat Gauss–Manin connection whose flat sections are the Betti cycles; pairing them against the (non-flat) holomorphic form gives the amplitudes across kinematic space, which obey the Picard–Fuchs differential equations.

6. Possibility with a boundary. A positive geometry carries a canonical form whose boundary residues recursively encode lower-dimensional geometries — a geometric model of physical factorization and singularity structure.

1.3 Contributions

1. A self-contained formalization (framework, schemes) of the algebraic-geometry layer of the representation library: schemes via the functor of points, ideals as constraints, the on-shell algebra $R/\mathfrak{a}$, and the full [S]/[H]/[P]-labelled dictionary of 4, reproduced and organized from the primary source.

2. Four theorems with proofs (8), each of which extends a Part II construction over a geometric family: Gauss–Manin family realization (27), stack-vs-coarse strictness (29), positive-geometry residue/coaction correspondence (31, status [H]), and the derived critical locus as BV on-shell algebra (33).

3. An explicit family realization pipeline (9) generalizing (1) from a single mathematical object to a moduli-indexed family, together with a status-propagation calculus inherited from Part I.

4. Executable Haskell encodings and a best-effort Lean 4 sketch (11) of schemes-as-functors-of-points, quotient stacks with stabilizer data, variations of Hodge structure with Gauss–Manin connections, and positive geometries with recursive residue trees.

5. An explicit accounting (12) of how Part III respects the program’s five standing limitations, and a list of open problems — notably the conjectural identification of positive-geometry residue trees with motivic coaction trees — handed forward to Parts IV–VII.

1.4 Epistemic status discipline

We preserve verbatim the three-level status system of Part I.

Composite labels ([S/H], [H/P]) appear where an entry is standard as pure mathematics but heuristic or speculative in its physical reading; we preserve these composite labels verbatim. This discipline is not decoration: it is the program’s central epistemic-honesty device, and we take pains in 8 to state precisely which of our four theorems are status-[S] (established mathematics applied to a physical reading), which are status-[H] (heuristically plausible, original to this program), and which are status-[P].

2 Mathematical Framework

We recall the ambient framework of Parts I–II in the compressed form we need, then specialize to the algebraic-geometry domain $d=\mathsf{AG}$.

2.1 Domains, structures, and representation entries

For this paper $d=\mathsf{AG}$ and ${\mathsf{Math}}_{\mathsf{AG}}$ is (a chosen full subcategory of) the category of schemes, algebraic spaces, and stacks over a fixed base field $\mathbb{k}$, together with the auxiliary categories of Hodge structures and positive geometries introduced below.

2.2 The realization pipeline

Recall the universal formula (1). In Part II the channels $\alpha$ were the cohomological realizations $\{\text{Betti},\text{de Rham},\text{Hodge},\text{eˊtale}\}$ and the abstract information functor was $\Phi =\mathrm{Mot}$, sending a variety to its motive. The realization pipeline for a single variety $X$ is the chain

Part III’s central move is to let $X$ vary: we replace the single object $X$ by a family $\pi :\mathcal{X}\to S$ over a base scheme (moduli space) $S$ and ask how (2) behaves as we move in $S$. The answer, made precise in 27, is that the realization assembles into a variation of Hodge structure over $S$: the Betti cycles are the flat sections of the Gauss–Manin connection, and the observable is their pairing with the (non-flat) holomorphic form — a period function whose variation is governed by ${\nabla }_{\mathrm{GM}}$ and which satisfies the Picard–Fuchs equations.

2.3 The master diagram, fibered over a base

1 exhibits the family realization pipeline as a commuting diagram. The top row is Part I’s master diagram for a single object; the bottom row is its family version, and the vertical arrows are the fibers of the family over a point $s\in S$.

$\mathcal{X}$

$\underline{\mathrm{Mot}}(\mathcal{X}/S)$

${\mathcal{H}}_{\alpha }$

$\text{period }\Pi (s)$

$S$

$S$

$S$

$(\text{Picard–Fuchs})$

• $\mathcal{X}$→$\underline{\mathrm{Mot}}(\mathcal{X}/S)$$\mathrm{Mot}$
• $\mathcal{X}$↓$S$$\pi$
• $\underline{\mathrm{Mot}}(\mathcal{X}/S)$→${\mathcal{H}}_{\alpha }$${\mathrm{Real}}_{\alpha }$
• $\underline{\mathrm{Mot}}(\mathcal{X}/S)$↓$S$
• ${\mathcal{H}}_{\alpha }$→$\text{period }\Pi (s)$$\mathrm{Obs}={\int }_{\gamma }$
• ${\mathcal{H}}_{\alpha }$↓$S$$\text{fiber}$
• $\text{period }\Pi (s)$↓$(\text{Picard–Fuchs})$
• $S$=$S$
• $S$=$S$
• $S$→$(\text{Picard–Fuchs})$${\nabla }_{\mathrm{GM}}$

3 Schemes as Spaces of Physical Possibility

3.1 From constraints to varieties

Physically, $V(\mathfrak{a})$ is the space of classical states satisfying the imposed equations of motion or conservation constraints, and $R/\mathfrak{a}$ is the algebra of functions (observables) modulo those constraints: an observable that differs from another by an element of $\mathfrak{a}$ takes the same value on every physical state, hence is the same observable on-shell.

3.2 Schemes and infinitesimal possibility

Varieties see only reduced structure: they cannot distinguish $V(x)$ from $V(x^2)$, both being the single point $\{0\}$. Physics, however, frequently needs the infinitesimal or off-shell data that nilpotents encode. Schemes retain this data.

3.3 Projective varieties and scale invariance

Physically a projective variety is a scale-invariant configuration space: physical data considered modulo an overall rescaling. Cross-ratios and other conformal invariants live naturally on projective and Grassmannian geometries; this is the setting in which positive geometries (7) are formulated.

4 The Algebraic Geometry Dictionary

We reproduce and organize the algebraic-geometry library of the primary source (its Appendix C), preserving every status label. The dictionary is the data of a family of representation entries (1) in the domain $d=\mathsf{AG}$; the theorems of 8 pick out the load-bearing rows and prove precise statements about them.

5 Moduli, Stacks, and Gauge-Correct Possibility

5.1 Moduli functors and stacks

The space of “physically distinct configurations” of a given type is a moduli space. We phrase it functorially, which is both the mathematically robust formulation and the one matching our functor-of-points encoding (10).

The essential invariant retained by the stack but forgotten by the coarse space is the automorphism (stabilizer) group of each point.

16 is the algebraic-geometry incarnation of the master formula ${\mathrm{Aut}}_{[X/G]}(x)\simeq {\mathrm{Stab}}_G(x)$ recorded across the whole program. It is the reason a moduli stack is “gauge-correct”: it distinguishes a configuration with no gauge automorphisms from one stabilized by a residual gauge subgroup, even when the two have the same image in the coarse space.

5.2 Gauge redundancy, formally

6 Hodge Theory and Kinematic-Space Amplitudes

We now formalize the machinery that lets Part II’s periods vary over a family, which is the technical heart of 27.

7 Positive Geometries and Geometric Factorization

The residue tree is the object we compare, in 31, with the coaction tree of Part II’s motivic amplitude objects.

8 Main Results

We now state and prove the four theorems announced in 1. Each extends a Part II construction over a geometric family and each carries an explicit status label.

8.1 T1: Gauss–Manin family realization

27 is the precise sense in which Part III “fibers” Part II: the per-object period becomes a period function on the moduli base, constrained by the Picard–Fuchs system, and the previously mysterious “amplitude over kinematic space” becomes a rigorously constrained object.

8.2 T2: Stacks are strictly finer than coarse spaces

29 is the algebro-geometric member of the program’s four-fold “stacks retain gauge information” family: Part I posits it informally, Part III proves it geometrically here, Part V will prove a type-theoretic version (groupoid quotient vs. set truncation), and Part VI will prove the descent-theoretic version. All four are the same fact at increasing rigor.

8.3 T3: Positive-geometry residues and the motivic coaction

8.4 T4: The derived critical locus as BV on-shell algebra

9 The Family Realization Pipeline

We assemble the theorems into the promised generalization of Part I’s pipeline from a single object to a moduli-indexed family, and we track statuses with 2.

38 is exactly the epistemic bookkeeping the program demands: it names precisely which realization channels keep the pipeline standard and which push it into heuristic or speculative territory.

10 Worked Examples

10.1 The Legendre family as a family pipeline

Combining 22 with 36: for $\pi :\mathcal{E}\to S={\mathbb{P}}^1\setminus \{0,1,\infty \}$ the Legendre family, the VHS $\mathcal{H}=R^1{\pi }_{\ast }\underline{\mathbb{Q}}$ has rank $2$, the periods $\Pi (\lambda )$ satisfy the hypergeometric Picard–Fuchs operator, and the monodromy representation ${\pi }_1(S)\to S{L}_2(\mathbb{Z})$ around $\{0,1,\infty \}$ is the branch-cut data. This is a complete status-[S] instance of the entire Part III machinery: a moduli base, a VHS, a Picard–Fuchs equation, and monodromy as physical analytic continuation.

10.2 A quotient stack: $[{\mathbb{A}}^1/{\mathbb{G}}_m]$

Let ${\mathbb{G}}_m$ act on ${\mathbb{A}}^1$ by scaling. The GIT/coarse quotient is $\mathrm{Spec}(\mathbb{k}[x{]}^{{\mathbb{G}}_m})=\mathrm{Spec}\mathbb{k}$, a single point (the only invariant functions are the constants, and the only closed orbit is the origin), so the coarse space records nothing beyond a point. The stack $[{\mathbb{A}}^1/{\mathbb{G}}_m]$, by contrast, has two points: the open orbit, whose objects have trivial automorphisms, and the origin, whose automorphism group is $\mathrm{Aut}(0)\cong {\mathbb{G}}_m$ (by 16, ${\mathrm{Stab}}_{{\mathbb{G}}_m}(0)={\mathbb{G}}_m$). This is the algebraic model of a $U(1)$ gauge theory at a symmetric point: the residual gauge group at the origin is retained by the stack and forgotten by the coarse space, exactly as 29 predicts.

10.3 A derived critical locus: the free scalar

For $X={\mathbb{A}}^1$ with coordinate $\varphi$ and action $S(\varphi )=\frac{1}{2}{m}^2{\varphi }^2$, we have $dS=m^2\varphi d\varphi$, $\mathrm{Crit}(S)=\{\varphi =0\}$, and $R/\mathfrak{a}=\mathbb{k}[\varphi ]/(\varphi )=\mathbb{k}$: the unique on-shell state is $\varphi =0$. The derived critical locus resolves the (here already reduced) intersection by the Koszul complex $\mathbb{k}[\varphi ]\stackrel{\cdot m^2\varphi }{\leftarrow }\mathbb{k}[\varphi ]\xi$ with $\xi$ the antifield in degree $-1$; the $(-1)$-shifted symplectic form pairs $\varphi$ with $\xi$, giving the BV antibracket $\{\varphi ,\xi \}=1$. There are no ghosts, since the free scalar has no gauge symmetry (consistent with 34). This is the smallest nontrivial illustration of 33.

11 Formalization: Haskell and Lean

We accompany the paper with executable Haskell encodings of the four load-bearing structures (scheme as functor of points, quotient stack, variation of Hodge structure, positive geometry) and a best-effort Lean 4 sketch of the same type signatures. The Haskell modules compile with GHC and their main prints demonstrations of 16, 31(1), and the residue-tree construction. Full sources are in src/algebraic-geometry-physical-possibility/ and lean/algebraic-geometry-physical-possibility/.

11.1 Scheme as functor of points

Following 10 we model a scheme by its functor of points $\mathsf{CRing}\to \mathsf{Set}$, and the on-shell algebra $R/\mathfrak{a}$ by a coordinate ring with a distinguished constraint ideal:

-- A commutative ring presented by generators and a constraint ideal.
data CoordRing = CoordRing { generators :: [String], relations :: [Poly] }

-- A scheme exposed via its functor of points R |-> Hom(Spec R, X).
newtype Scheme = Scheme { pointsOver :: CoordRing -> [RPoint] }

-- The on-shell algebra R/I: functions modulo the ideal of constraints.
-- (idealGens is the record accessor for an Ideal's generating polynomials.)
onShell :: CoordRing -> Ideal -> CoordRing
onShell r i = r { relations = relations r ++ idealGens i }

11.2 Quotient stack with stabilizer data (16)

-- A finite group action, enough to compute stabilizers concretely.
data GAction g x = GAction { act :: g -> x -> x, elements :: [g] }

-- Stabilizer of a point: {g | g . x == x}. This is Aut_{[X/G]}(x).
stabilizer :: (Eq x) => GAction g x -> x -> [g]
stabilizer ga x = [ g | g <- elements ga, act ga g x == x ]

-- A quotient stack remembers, for each point, its automorphism group.
data QuotientStack g x = QuotientStack
  { action    :: GAction g x
  , autGroup  :: x -> [g]        -- x |-> Stab_G(x) = Aut_{[X/G]}(x)
  }

The demonstration in Main.hs builds $[\mathbb{Z}/4\curvearrowright \mathbb{Z}/4]$ (rotation of a square) and prints the automorphism group at the fixed point, exhibiting 16 concretely: the coarse quotient forgets it, the stack does not.

11.3 Variation of Hodge structure with Gauss–Manin connection

-- A VHS over a base: a fiber vector space, a Hodge filtration, and a
-- flat (Gauss-Manin) connection whose FLAT SECTIONS are the Betti cycles
-- (a basis against which the non-flat holomorphic form is measured).
data VHS base = VHS
  { fiber        :: base -> [Double]      -- local system fiber
  , hodgeFiltr   :: base -> [[Double]]    -- F^p, decreasing filtration
  , gaussManin   :: Connection base       -- flat connection nabla
  }

-- Derive the Picard-Fuchs operator by expressing the iterated
-- Gauss-Manin derivatives (nabla^j [omega]) in the finite-rank
-- de Rham basis and reading off the linear dependence. It
-- ANNIHILATES the period function Pi(s) -- the pairing of a flat
-- cycle with the NON-flat form -- which is not a flat section.
derivePicardFuchs :: VHS base -> ODE
derivePicardFuchs = linDep . gmDerivatives

11.4 Positive geometry and the recursive residue tree

-- A positive geometry: a region, its canonical form, and its boundary
-- strata (themselves positive geometries) -- the recursion of Def 7.1.
data PositiveGeometry = PositiveGeometry
  { canonicalForm :: DiffForm
  , boundaries    :: [PositiveGeometry]
  }

-- The residue tree of Def 7.4 / Theorem 8.5(1): root = canonical form,
-- children = residue trees of the boundary strata.
data Tree a = Node a [Tree a]

residueTree :: PositiveGeometry -> Tree DiffForm
residueTree pg = Node (canonicalForm pg)
                      (map residueTree (boundaries pg))

11.5 Lean sketch

The Lean 4 file records idiomatic type signatures for the same structures (scheme via a functor of points, quotient stack with a stabilizer field, VHS with Griffiths transversality, positive geometry with the residue axiom) and states 29 as a signature with a sorry placeholder. It is a best-effort sketch, not build-gated:

structure QuotientStack (X G : Type) [Group G] [MulAction G X] where
  autGroup : X -> Subgroup G            -- x |-> Stab_G(x) = Aut_{[X/G]}(x)
  coarseMap : X -> Quotient (MulAction.orbitRel G X)

-- T2: a nontrivial stabilizer forces the coarse map to lose information.
theorem stack_finer_than_coarse
    {X G : Type} [Group G] [MulAction G X] (x : X)
    (h : MulAction.stabilizer G x /= (bot : Subgroup G)) :
    True := by trivial   -- best-effort placeholder; see paper Thm 8.3

12 Discussion

12.1 How Part III respects the standing limitations

The program carries five standing limitations from Part I; we discharge each for the algebraic-geometry domain.

1. Not every analogy is a physical theory. We enforce this with the status labels of dict1, dict2 and, sharply, with 31: its part (1) is [S] and its part (2) is [H], and we refuse to collapse them.

2. Motives are not literally “the functions of” an object. We used $\mathrm{Mot}$ only through its realizations ${\mathrm{Real}}_{\alpha }$ (Hodge, Betti, de Rham), which are standard functors; the “functional essence” reading remains a Part II [H] slogan and is never invoked as a premise in a proof here.

3. No single universal functor $\mathrm{Math}\to \mathrm{Phys}$. Our pipeline is domain-local: it is a functor on the algebraic-geometry domain ${\mathsf{Math}}_{\mathsf{AG}}$, fibered over a base $S$, not a global functor. 36 is explicitly indexed by a family and a channel $\alpha$.

4. Not every observable is a period; elliptic/modular/non-Tate structures intervene. 22 already exhibits an elliptic period, and 31(2) is flagged [H] precisely because non-Tate geometries obstruct a naive polylogarithmic coaction (Part II, T4). Corollary 28 bounds only the order of the Picard–Fuchs operator, not the transcendental type of its solutions.

5. Gauge redundancy $\ne$ physical symmetry. 18 states this explicitly; 29 makes it geometric (stabilizers are gauge automorphisms, retained by the stack, not new physical states).

12.2 Open problems handed forward

• Positive geometry $=$ motivic coaction (31(2)). The sharpest open problem of this domain: construct, for all amplituhedron-type geometries, a functor from the poset of boundary strata to the comodule of motivic periods intertwining ${\Delta }_{\mathrm{geo}}$ with the motivic coproduct $\Delta$. Partial results exist for one-loop integrands and polygon/associahedron geometries; the general case is open (“Hopf algebra of the amplituhedron”).

• Stacky path integrals. Making 30 quantitative — the precise measure on ${\overline{\mathcal{M}}}_g$ weighted by $1/|\mathrm{Aut}|$ — connects Part III to Part IV’s TQFT functors and Part VI’s stacks-and-descent treatment.

• Derived enhancement of the whole dictionary. 33 upgrades one row to [S] via derived geometry; systematically deriving the tangent/cotangent-complex rows of dict2 (BV/BRST for gauge theories) is a natural next step toward Part VI.

12.3 How Part III seeds Part IV

Part IV (algebraic topology: conserved global information) takes the varieties and moduli produced here and studies their topological invariants stable under continuous deformation: the (co)homology $H_{\bullet }(X)$ underlying the Betti realization of 27, the characteristic classes of the bundles in dict1, and the bordism/TQFT functors that make realization functorial. In particular, the monodromy representation of 28 is a topological (local-system) datum, and the automorphism data of 29 is what Part IV’s groupoid-cardinality/TQFT weightings act on. Part III thus supplies the geometric substrate; Part IV supplies its topology.

13 Conclusion

We have formalized the algebraic-geometry layer of the Math$\to$Physics Representation Library as the geometry of physical possibility, preserving the program’s [S]/[H]/[P] status discipline throughout. The central technical contribution is the extension of Part II’s per-object realization pipeline to a family pipeline fibered over a moduli base: Gauss–Manin transport realizes periods as the pairing of flat Betti cycles with the non-flat holomorphic form, giving period functions constrained by Picard–Fuchs equations (27), moduli stacks retain the gauge automorphisms that coarse spaces forget (29), positive geometries encode factorization through recursive residues that conjecturally match the motivic coaction (31), and derived critical loci realize the on-shell algebra $R/\mathfrak{a}$ at the BV level, upgrading a heuristic dictionary entry to a theorem (33).

In the program’s compressed slogans:

mathematics is the syntax of physical representation;
motives are functional semantics;
periods are numerical measurements;
coalgebras are decomposition laws.

Part III adds a fifth clause: geometry is the space in which those measurements vary. The moduli space is the stage, its singularities are the thresholds, its compactification supplies the asymptotic sectors, and its stackiness is the gauge redundancy — a single algebro-geometric object carrying, in its points, its automorphisms, and its variation of Hodge structure, the entire apparatus of physical possibility. Part IV will endow this stage with its conserved topological invariants.