Paper IXReading edition

Analytic regularity of the local solution map

Reading tools
Reading libraryFind statementsInspect Lean
Switch work

Saved reading

Read side by side

Start reading

Resources & source

Original TeX with line links · Download TeX

The wording and mathematics follow the retained manuscript. This reading format does not add a proof review or extend Lean coverage.

Skip contents and start reading

11 Analytic regularity of the local solution map

The deformation estimates control how far an exact root moves under a small perturbation. We now show that its local dependence on the parameters is analytic, and compute the derivative from the polynomial Jacobian. The analytic argument takes place over a complete normed field. For integral \(p\)-adic data, we first include the coefficients and base point in \({\mathbb Q_p}\); \({\mathbb Z_p}\) itself is not a vector space over \({\mathbb Q_p}\).

Let \(\mathbb K\) be a complete nontrivially normed field. Let \(e,k\in{\mathbb N}\), and write the variables as \((Y,T)\in\mathbb K^e\times\mathbb K^k\). For

\[ F\in\mathbb K[Y_1,\ldots,Y_e,T_1,\ldots,T_k]^e, \qquad a=(y_0,t_0), \]

define the matrices of evaluated formal partial derivatives

\[ J=D_YF(a)\in M_e(\mathbb K),\qquad K=D_TF(a)\in M_{e\times k}(\mathbb K). \]
(11.1)

Rows index equations and columns index variables. We first establish that these formal derivatives are the derivatives of the evaluation map.

Lemma 11.1: Derivative of polynomial evaluation

For \(P\in\mathbb K[X_1,\ldots,X_n]\), polynomial evaluation is analytic at every point. At \(a\in\mathbb K^n\) its strict Fréchet derivative is

\[ v\longmapsto\sum_{j=1}^n (\partial_jP)(a)v_j. \]

For a finite system, its strict derivative is the continuous linear map represented by the matrix of evaluated formal partial derivatives.

Proof of Lemma 11.1

For the derivative statement, use polynomial induction on constants, addition, and multiplication by a variable. Constants have zero derivative. The sum rule agrees with the additivity of formal partial differentiation. For \(PX_j\), the product rule gives

\[ d(PX_j)_a(v)=a_j\,dP_a(v)+P(a)v_j, \]

which agrees with the evaluated formal product rule for every coordinate. The finite product rule for maps to coordinate spaces then assembles the system derivative. Each matrix defines a continuous linear map because the coordinate spaces are finite-dimensional over the complete field. Polynomial analyticity follows from the corresponding polynomial induction in Mathlib and its finite-coordinate assembly.

The derivative bridge and analytic statements are Komorebi.PolynomialAnalyticEvaluation.hasStrictFDerivAt_evalSystem, Komorebi.PolynomialAnalyticEvaluation.analyticAt_evalSystem, and Komorebi.PolynomialAnalyticEvaluation.analyticOnNhd_evalSystem.

Lemma 11.2: The augmented derivative

If \(J\in\operatorname{GL}_e(\mathbb K)\), the polynomial map

\[ \Phi(Y,T)=(F(Y,T),T) \]

has invertible strict derivative at \(a\), with

\[ D\Phi(a)=\begin{pmatrix}J&K\\0&I_k\end{pmatrix},\qquad D\Phi(a)^{-1}=\begin{pmatrix}J^{-1}&-J^{-1}K\\0&I_k\end{pmatrix}. \]
(11.2)

The local inverse supplied by this derivative is analytic near \(\Phi(a)\).

Proof of Lemma 11.2

The lower coordinates of \(\Phi\) are the parameter variables. Differentiating them gives the blocks \(0\) and \(I_k\). Lemma 11.1 identifies the remaining blocks with (11.1). Multiplication of the two matrices in (11.2), in both orders, gives the identity. The resulting linear equivalence is continuous, as is its inverse. Apply the strict inverse-function theorem to \(\Phi\). The analytic inverse theorem applies to the same open partial homeomorphism, because \(\Phi\) is analytic and its derivative is that continuous linear equivalence. It supplies a convergent local power-series representation of this inverse.

The augmented Jacobian is identified in Komorebi.PolynomialJointSystem.jacobian_augmented. The continuous inverse block and its restriction are Komorebi.PolynomialAnalyticBlocks.upperEquiv and Komorebi.PolynomialAnalyticBlocks.inverse_solved_zero. Analyticity of the same constructed inverse is Komorebi.PolynomialAnalyticInverse.analyticAt_inverse.

Theorem 11.3: Analytic local root graph

Suppose \(F(a)=0\) and \(J\in\operatorname{GL}_e(\mathbb K)\). There are open neighborhoods \(t_0\in U\subseteq\mathbb K^k\) and \(y_0\in V\subseteq\mathbb K^e\), and a map \(\psi:\mathbb K^k\to\mathbb K^e\), such that

\[\begin{gathered} \psi(t_0)=y_0,\qquad \psi(U)\subseteq V,\qquad \psi\text{ is analytic at every point of }U,\\ F(y,t)=0\quad\Longleftrightarrow\quad y=\psi(t) \qquad(t\in U,\ y\in V). \end{gathered}\]
(11.3), (11.4)

At the base parameter, the strict Fréchet derivative is

\[ \boxed{D\psi(t_0)=-J^{-1}K.} \]
(11.5)

In particular, \(\psi\) is differentiable on \(U\).

Proof of Theorem 11.3

Let \(g\) be the analytic local inverse of \(\Phi\) supplied by Lemma 11.2. Since \(F(a)=0\), \(\Phi(a)=(0,t_0)\) and \(g(0,t_0)=a\). Define

\[ \psi(t)=\pi_Yg(0,t), \]

where \(\pi_Y\) projects to the solved coordinates. This gives the base value and analyticity at \(t_0\). The chain rule and (11.2) give

\[ D\psi(t_0)v =\pi_Y\begin{pmatrix}J^{-1}&-J^{-1}K\\0&I_k\end{pmatrix} \binom{0}{v} =-J^{-1}Kv. \]

To obtain the root equivalence on a product neighborhood, we choose the neighborhoods as follows. The source of the open partial homeomorphism defined by \(\Phi\) is open around \(a\); its inverse is \(g\). Choose open sets \(V\) and \(U_0\) containing \(y_0\) and \(t_0\) such that \(V\times U_0\) lies in that source. Restrict \(U_0\) further by the preimage of an open analytic inverse neighborhood under \(t\mapsto(0,t)\), the open locus where this same \(\psi\) is analytic, and the interior of \(\psi^{-1}(V)\). All of these sets contain \(t_0\). For the last set, use the continuity of \(\psi\) at \(t_0\) and \(\psi(t_0)=y_0\in V\). Their intersection is an open neighborhood \(U\) with \(\psi(U)\subseteq V\).

If \(t\in U\), \(y\in V\) and \(F(y,t)=0\), then \(\Phi(y,t)=(0,t)\). The left inverse identity forces \(g(0,t)=(y,t)\), hence \(y=\psi(t)\). Conversely, the right inverse identity \(\Phi(g(0,t))=(0,t)\) says both that its solved equations vanish and that its parameter coordinates are exactly \(t\). Thus \(g(0,t)=(\psi(t),t)\) and \(F(\psi(t),t)=0\). Analyticity at every point of \(U\) implies differentiability there.

This is Komorebi.PolynomialAnalyticImplicit.exists_analytic_implicit. Differentiability is also included explicitly in the statement contract Komorebi.PolynomialRegularityContracts.actual_padic_analytic_root_graph.

Theorem 11.4: Regularity of a supplied root selection

Under the hypotheses of Theorem 11.3, let \(h:\mathbb K^k\to\mathbb K^e\) be continuous at \(t_0\), with \(h(t_0)=y_0\). Suppose \(F(h(t),t)=0\) for every \(t\) in some neighborhood of \(t_0\). Then \(h\) and \(\psi\) agree in a neighborhood of \(t_0\). Consequently \(h\) is analytic there as a germ and has strict derivative \(Dh(t_0)=-J^{-1}K\).

Proof of Theorem 11.4

Continuity at \(t_0\) places \(h(t)\) inside \(V\) for all sufficiently close parameters. Intersect this neighborhood with \(U\) and the neighborhood on which the root equation holds. Equation (11.4) gives \(h(t)=\psi(t)\) throughout the intersection. Analyticity and strict differentiability at a point are unchanged by agreement on a neighborhood.

The formal result is Komorebi.PolynomialAnalyticSelection.continuous_selection_regularity. Its formulation derives \(F(a)=0\) from the root equation near the base parameter and the specified base value.

Corollary 11.5: Regularity throughout an open parameter set

Let \(U\subseteq\mathbb K^k\) be open, and let \(h\) be continuous on \(U\). Suppose that for every \(t\in U\),

\[ F(h(t),t)=0, \qquad J_t=D_YF(h(t),t)\in\operatorname{GL}_e(\mathbb K). \]

Then \(h\) is analytic at every point of \(U\) and its strict derivative is

\[ Dh(t)=-J_t^{-1}D_TF(h(t),t)\qquad(t\in U). \]

Proof of Corollary 11.5

At each \(t\in U\), openness turns continuity on \(U\) into continuity at \(t\) and makes the root equation hold on a neighborhood of \(t\). Apply Theorem 11.4 at \((h(t),t)\).

This is Komorebi.PolynomialAnalyticSelection.continuous_selection_analyticOnNhd. The pointwise invertibility assumption is part of this corollary.

Corollary 11.6: Integral coefficients and centers

Let \(p\) be any prime. Suppose \(F\in{\mathbb Z_p}[Y,T]^e\), \(a\in{\mathbb Z_p}^{e+k}\), and \(F(a)=0\). Let \(\iota:{\mathbb Z_p}\hookrightarrow{\mathbb Q_p}\) be the inclusion. If \(\iota(D_YF(a))\in\operatorname{GL}_e({\mathbb Q_p})\), the polynomial system obtained by including the original coefficients in \({\mathbb Q_p}\) has the analytic local root graph of Theorem 11.3 at \(\iota(a)\). The strict derivative at the base is

\[ -\iota(D_YF(a))^{-1}\iota(D_TF(a)). \]

The open neighborhoods and the output map are in the ambient spaces over \({\mathbb Q_p}\).

Proof of Corollary 11.6

Coefficient inclusion commutes with evaluation and formal partial differentiation. Consequently both Jacobian blocks after inclusion are the entrywise inclusions of the original blocks, and the original exact root remains an exact root. Apply Theorem 11.3 to the system with included coefficients.

This is Komorebi.PolynomialRegularityIntegral.exists_analytic_integral_implicit. It requires no invertibility modulo \(p\). In particular, it includes solved Jacobians singular modulo \(p\) but invertible over the fraction field.

Example 11.7: The derivative sign and a nonlinear control

For \(F(Y,T)=Y+T\), the exact branch is \(\psi(T)=-T\) and \(J=K=1\). The derivative is \(-1\). Uniqueness of the derivative rules out \(+1\) over \({\mathbb Q_p}\), including \(p=2\). The nonlinear polynomial \(F(Y,T)=Y+Y^2+T\) has \(F(0,0)=0\) and \(J=K=1\) there. Applying the formal theorem to this polynomial constructs an analytic local root graph with derivative \(-1\). Thus the example checks the existence conclusion as well as the derivative formula. The declarations and the deliberately rejected positive-sign statement are in the regularity control and statement-contract files.

The results cover arbitrary finite \(e\) and \(k\), including zero. The local inverse machinery explicitly handles a zero-dimensional domain. The analytic theorem does not assign a uniform numerical radius to its neighborhoods. The earlier weighted lifting and joint deformation theorems retain their quantitative radii and error bounds, while the present statements establish regularity of the local solution germ. A direct Lean identification with each historical subtype-valued chart is an additional interface task. General persistence at a fraction-field singular solved block is a separate problem.

Source and conversion record

Source SHA-256: 32f48dcc51d918d65be6dc827f31b05bc9870decb20dc596fed9276940244ffd

Unsupported prose commands stop the build. The mathematical expression sequence is checked against the source, with manuscript macros expanded for display.

Conversion records

Notation in Paper IX

Look up a symbol without leaving this chapter. Select text before opening help to search for it.

Search this work’s notation.

    Symbols can have different meanings in different passages. Read each result in its own context.

    Open the full notation index