Skip to content
Shadow Theory

Observable Quotients and Exact Projected Dynamics: Closure, Memory, Minimal Dynamical Completion, and Effective Field Operators

Nonlinear dynamical descent

Jeremy RodgersSection 9 of 14Original PDF ↗Publication DOI ↗

Nonlinear dynamical descent

The criterion PLQ=0PLQ=0 has an exact nonlinear counterpart, obtained by replacing the linear quotient map by a surjective submersion and the linear generator by a smooth vector field.

Assumption 8.1.

SS and TT are smooth manifolds, p:S→Tp:S\to T is a smooth surjective submersion (so Dps:TsS→Tp(s)TDp_{s}:T_{s}S\to T_{p(s)}T is surjective for every s∈Ss\in S, and pp is onto), and XX is a smooth vector field on SS.

The hypothesis that pp be a surjective submersion is not cosmetic. It guarantees (i) that TT is a genuine smooth manifold and p−1(t)p^{-1}(t) an embedded submanifold for each tt, and (ii) the existence of smooth local sections, which is what makes the smoothness argument in Theorem 8.2 work. We make no claim about arbitrary set-theoretic quotients: a quotient of a manifold by an arbitrary equivalence relation need not be a smooth manifold at all, and nothing below should be read as asserting otherwise.

Theorem 8.2 (Nonlinear projectability criterion).

Under Assumption 8.1, there exists a smooth vector field YY on TT with Dp∘X=Y∘p,i.e.DpsXs=Yp(s)  for every s∈S,(8.1) Dp\circ X=Y\circ p , \qquad\text{i.e.}\qquad Dp_{s}X_{s}=Y_{p(s)}\ \ \text{for every }s\in S,\tag{8.1} if and only if DpDp applied to XX is constant along every fiber: p(s)=p(s′) ⟹ DpsXs=Dps′Xs′in Tp(s)T.(8.2) p(s)=p(s')\ \Longrightarrow\ Dp_{s}X_{s}=Dp_{s'}X_{s'} \qquad\text{in }T_{p(s)}T .\tag{8.2} When it exists, YY is unique.

Proof. Necessity. Suppose YY exists and p(s)=p(s′)=:tp(s)=p(s')=:t. Then by (8.1), DpsXs=Yt=Dps′Xs′Dp_{s}X_{s}=Y_{t}=Dp_{s'}X_{s'}, and both sides lie in the same tangent space TtTT_{t}T, so (8.2) holds.

Sufficiency. Assume (8.2). For t∈Tt\in T, use surjectivity of pp to pick s∈p−1(t)s\in p^{-1}(t) and define Yt:=DpsXs ∈TtT.(8.3) Y_{t}:=Dp_{s}X_{s}\ \in T_{t}T .\tag{8.3} By (8.2) this is independent of the chosen s∈p−1(t)s\in p^{-1}(t), so Y:T→TTY:T\to TT is a well-defined section of TTTT: Yt∈TtTY_{t}\in T_{t}T by construction. Equation (8.1) holds by definition.

Smoothness. Fix t0∈Tt_{0}\in T and s0∈p−1(t0)s_{0}\in p^{-1}(t_{0}). Because pp is a submersion, the local normal form for submersions [8, Thm. 4.12] provides an open neighbourhood W∋t0W\ni t_{0} in TT and a smooth local section σ:W→S\sigma:W\to S with p∘σ=idWp\circ\sigma=\mathrm{id}_{W} and σ(t0)=s0\sigma(t_{0})=s_{0}. On WW we then have, from (8.3) applied with s=σ(t)∈p−1(t)s=\sigma(t)\in p^{-1}(t), Y∣W=Dp∘X∘σ,(8.4) Y|_{W}=Dp\circ X\circ\sigma ,\tag{8.4} which is a composition of smooth maps (σ\sigma smooth, XX smooth as a section of TSTS, Dp:TS→TTDp:TS\to TT smooth as the tangent map of a smooth map), hence smooth on WW. Different choices of section give the same YY on the overlap by fiber constancy (8.2), so these local formulas patch to a globally well-defined smooth field. Since t0t_{0} was arbitrary, YY is smooth on TT.

Uniqueness. If YY and Y′Y' both satisfy (8.1), then for any t∈Tt\in T pick s∈p−1(t)s\in p^{-1}(t) (possible by surjectivity); then Yt=DpsXs=Yt′Y_{t}=Dp_{s}X_{s}=Y'_{t}. ◻

Remark 38 (The nonlinear analogue of PLQ≠0PLQ\neq0).

Failure of (8.2) says: there are two source states s,s′s,s' with the same readout p(s)=p(s′)p(s)=p(s') whose projected velocities DpsXs≠Dps′Xs′Dp_{s}X_{s}\neq Dp_{s'}X_{s'} differ. This is verbatim the phenomenon of Theorem 6.4, where pp is the linear map PP, the fibers are the affine subspaces x0+HQx_{0}+\mathcal{H}_{Q}, and the projected velocity at x0+z0x_{0}+z_{0} is PL(x0+z0)=PLP x0+PLQ z0PL(x_{0}+z_{0})=PLP\,x_{0}+PLQ\,z_{0}, which is fiber-constant precisely when PLQ=0PLQ=0. Thus the linear closure criterion is the linearization, and \href\href{#eq:fiberconst}{(8.2)} its exact generalization: the flow descends to the observable manifold exactly when equal present readouts force equal projected velocities. Nothing in Theorem 8.2 produces a memory kernel; in the nonlinear setting the exact reduction identity of Theorem 6.2 has no elementary counterpart, and the theorem should be read as supplying only the sharp criterion for descent, not a reduction formula.

Back to top ↑