The Operator Derivative in Continuous Stochastic Calculus

2 The Representation Layer

This chapter develops the cylindrical structure, IBP formula, Leibniz rule, and the calculus layer. The main theorem: cylindrical reduction implies Leibniz [1, Theorem 5.4].

2.1 Cylindrical Structure

The integration-by-parts formula on cylindrical functionals is the mechanism that reduces \(D\) to partial differentiation.

Definition 24 Cylindrical structure (unbounded)
✓

Coordinates \(\xi _i\), representers \(\kappa _i\), coordinate derivatives \(\partial _i\), an IBP formula \(\langle F, \delta u \rangle = \sum _i \langle (\partial _i F) \cdot \kappa _i, u \rangle \) on cylindricals, and a closure axiom [1, Definition 5.1].

Definition 25 Cylindrical structure (bounded)
✓
#

The bounded version of the cylindrical structure, with the same IBP formula and a closure axiom extending Leibniz to all of \(L^2(\Omega )\) [1, Definition 5.1].

Theorem 26 IBP representation (unbounded)
✓

For cylindrical \(F\): \(DF = \sum _i (\partial _i F) \cdot \kappa _i\). Derived from the IBP formula + adjoint identity + nondegeneracy [1, Theorem 5.2].

Proof ▶

Show \(\langle DF - \sum _i (\partial _i F)\kappa _i, u \rangle = 0\) for all \(u \in \mathrm{dom}(\delta )\) using adjoint + IBP, then apply density.

Theorem 27 IBP representation (bounded)
✓
#

For cylindrical \(F\): \(DF = \sum _i (\partial _i F) \cdot \kappa _i\). The bounded version [1, Theorem 5.2].

Proof ▶

Same inner-product testing argument as the unbounded case.

2.2 Leibniz Rule

Definition 28 Leibniz condition (unbounded)
✓

\(D(FG) = F \cdot DG + G \cdot DF\) for \(F, G \in \mathrm{dom}(D)\) with \(FG \in \mathrm{dom}(D)\) [1, Definition 5.3].

Definition 29 Leibniz condition (bounded)
✓
#

\(D(FG) = F \cdot DG + G \cdot DF\) for all \(F, G \in L^2(\Omega )\) [1, Definition 5.3].

Theorem 30 Cylindrical Leibniz (unbounded)
✓
#

Leibniz holds on cylindrical \(F, G\). Derived from IBP representation + ordinary product rule for coordinate derivatives [1, proof of Theorem 5.4, step (b)].

Proof ▶

Expand \(D(FG)\), \(DG\), \(DF\) via IBP rep, apply coordinate Leibniz, recombine sums.

Theorem 31 Cylindrical implies Leibniz (unbounded)
✓

Cylindrical structure implies full Leibniz. Step 1 (algebraic Leibniz on cylindricals) is proved; Step 2 (closure to \(\mathbb {D}^{1,4}\)) uses the closure axiom [1, Theorem 5.4].

Proof ▶

Apply leibniz_closure to the proved cylindrical Leibniz.

Theorem 32 Cylindrical implies Leibniz (bounded)
✓
#

The bounded version of Theorem 5.4 [1].

Proof ▶

Apply leibniz_closure to the proved cylindrical_leibniz_on_class.

2.3 Product Rule and Chain Rule

Definition 33 Product rule
✓
#

\(\delta (F \cdot u) = F \cdot \delta (u) - \langle DF, u \rangle _H\) for \(F \in \mathrm{dom}(D)\) [1, Theorem 5.5].

Theorem 34 Leibniz iff product rule
✓

Leibniz \(\Leftrightarrow \) Product Rule. Both directions proved by inner-product testing [1, Theorem 5.5].

Proof ▶

Forward: test \(\langle \delta (Fu) - F\delta u + \mathrm{pip}(DF, u), G \rangle = 0\) for all \(G\) using Leibniz + adjoint. Backward: analogous.

Theorem 35 Chain rule for squares
✓
#

\(D(F^2) = 2F \cdot DF\), derived from Leibniz with \(G = F\) [1].

Proof ▶

Apply Leibniz with \(G = F\), simplify \(2 \cdot \mathrm{smul}(F, DF)\).

Theorem 36 Chain rule for powers
✓
#

\(D(F^n) = n F^{n-1} \cdot DF\) for \(n \geq 1\), by induction on Leibniz [1].

Proof ▶

Induction on \(n\): base case from D_const, step from Leibniz + inductive hypothesis.

2.4 Itô Correction

Definition 37 Intrinsic bracket (unbounded)
✓

\(\langle Y \rangle ^{(D,\Pi )} := \mathrm{pip}(\Pi DY, \Pi DY)\). The intrinsic quadratic variation [1, Definition 6.1].

Theorem 38 Itô decomposition (unbounded)
✓

\(\delta (\varphi '(Y) \cdot \Pi DY) = \varphi '(Y) \cdot \delta (\Pi DY) - \varphi ''(Y) \cdot \langle Y \rangle \). Proved from Leibniz \(\Rightarrow \) product rule + chain rule [1, Theorem C].

Proof ▶

Apply product rule (from Leibniz), substitute the chain rule for \(\varphi '\), identify the correction via pip_smul.

Theorem 39 Main pipeline
✓
#

From PRP + isometry + cylindrical structure + smooth calculus: both Clark–Ocone and the Itô decomposition hold. This captures the paper’s entire contribution as a formal implication [1].

Proof ▶

Conjunction of clark_ocone_unbounded and operator_ito_decomposition_unbounded.

2.5 Stochastic Volatility

Theorem 40 Leibniz for stochastic volatility (unbounded)
✓

If \(D_M F = \sigma ^{-1} \cdot D^W F\) and \(D^W\) satisfies Leibniz, then \(D_M\) satisfies Leibniz. The transfer formula + Brownian Leibniz + smul algebra [1, Theorem D].

Proof ▶

Apply the transfer formula, Brownian Leibniz, distribute smul, commute scalars.

Theorem 41 Gubinelli base-invariance
✓
#

The pip-ratio \(\mathrm{pip}(DF, u) / \mathrm{pip}(DY, u)\) is independent of the driving process. Cross-multiplication form proved from transfer + pip_smul + mul algebra [1, Theorem F].

Proof ▶

Factor through pip_smul, cancel volatility via commutativity.