MerLean-example

19 Thm 4: Projection Induces Isomorphism

Definition 226 Projection Map \(\pi _*\) at Degree 1
#

Let \(\varepsilon : (\mathbb {F}_2^{m_2}) \to \mathbb {F}_2\) be a linear map (the augmentation of \(F_0\)). The projection \(\pi _*\) at degree 1 is the linear map

\[ \pi _*^{(1)} : \bigl((\mathbb {F}_2^{n_1} \otimes \mathbb {F}_2^{m_2}) \times (\mathbb {F}_2^{m_1} \otimes \mathbb {F}_2^{n_2})\bigr) \; \longrightarrow \; \mathbb {F}_2^{n_1} \]

defined by

\[ \pi _*^{(1)}(x_{10},\, x_{01}) \; =\; (\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(x_{10}), \]

where \(\operatorname {rid}\) denotes the right-unit isomorphism \(V \otimes \mathbb {F}_2 \simeq V\). Concretely, on simple tensors \(\pi _*^{(1)}(b \otimes f,\, 0) = \varepsilon (f) \cdot b\), and \(\pi _*^{(1)}\) is zero on the \(\mathbb {F}_2^{m_1} \otimes \mathbb {F}_2^{n_2}\) component.

Lemma 227 Augmentation Is Invariant Under the Connection

Let \(d_F : \mathbb {F}_2^{n_2} \to \mathbb {F}_2^{m_2}\) be a linear map, let \(\psi \) be a chain automorphism of \(d_F\), and let \(\varepsilon : \mathbb {F}_2^{m_2} \to \mathbb {F}_2\) be a linear map satisfying \(\varepsilon \circ d_F = 0\). If \(\psi \) acts as the identity on the homology of \(d_F\) (i.e. \(\psi .\operatorname {ActsAsIdOnHomology}\)), then

\[ \varepsilon \circ \psi .\alpha _0 = \varepsilon . \]
Proof

Let \(v\) be arbitrary. By extensionality it suffices to show \(\varepsilon (\psi .\alpha _0\, v) = \varepsilon (v)\).

Since \(\psi \) acts as the identity on homology at degree 0, we have \(\psi .\alpha _0\, v - v \in \operatorname {im}(d_F)\). Obtaining \(w\) such that \(d_F(w) = \psi .\alpha _0\, v - v\), we compute:

\[ \varepsilon (\psi .\alpha _0\, v - v) = \varepsilon (d_F(w)) = (\varepsilon \circ d_F)(w) = 0, \]

using the hypothesis \(\varepsilon \circ d_F = 0\). Applying \(\operatorname {map_sub}\) and \(\operatorname {sub_eq_zero}\), we conclude \(\varepsilon (\psi .\alpha _0\, v) = \varepsilon (v)\).

Lemma 228 Chain Map Property of \(\pi _*\) for the Twisted Differential

Let \(d_B\), \(d_F\), and \(\varphi \) be given, and let \(\varepsilon \) satisfy \(\varepsilon \circ d_F = 0\), with \(\varphi \) acting as the identity on homology. For any \(x \in \mathbb {F}_2^{n_1} \otimes \mathbb {F}_2^{m_2}\),

\[ (\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))\bigl(\widetilde{d}_h^0(x)\bigr) = d_B\bigl((\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(x)\bigr), \]

where \(\widetilde{d}_h^0\) is the twisted horizontal differential at bidegree \((1,0)\). That is, \(\pi _*^{(1)}\) intertwines \(\widetilde{d}_h^0\) with \(d_B\).

Proof

We proceed by induction on \(x\) using the tensor product induction principle.

Zero case: Both sides vanish trivially.

Additivity: Both sides are linear, so the identity extends from simple tensors.

Simple tensor case \(x = b \otimes f\): Expanding \(\widetilde{d}_h^0(b \otimes f)\) via twistedDhLin_tmul and simplifying autAtDeg, we obtain

\[ \sum _{b_1, b_0} b(b_1) \cdot d_B(e_{b_1})(b_0) \cdot \varepsilon (\alpha _0^{\varphi (b_1,b_0)} f) \cdot e_{b_0}. \]

For each term, by epsilon_comp_alpha0_eq (when \(d_B(e_{b_1})(b_0) \neq 0\), the chain automorphism \(\varphi (b_1,b_0)\) acts as the identity on homology), we replace \(\varepsilon (\alpha _0^{\varphi (b_1,b_0)} f)\) by \(\varepsilon (f)\). After this substitution the sum becomes

\[ \sum _{b_1, b_0} b(b_1) \cdot d_B(e_{b_1})(b_0) \cdot \varepsilon (f) \cdot e_{b_0}. \]

Proceeding pointwise and applying the basis expansion \(d_B(b) = \sum _{b_1} b(b_1) \cdot d_B(e_{b_1})\), we obtain \(\varepsilon (f) \cdot d_B(b) = d_B(\varepsilon (f) \cdot b)\), which equals \(d_B((\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(b \otimes f))\).

Lemma 229 \(\pi _*\) Maps Cycles to \(\ker (d_B)\)

Under the same hypotheses, for any \(z \in \operatorname {totZ}_1\) (the total 1-cycles of \(B \otimes _\varphi F\)),

\[ \pi _*^{(1)}(z) \; \in \; \ker (d_B). \]
Proof

We must show \(d_B(\pi _*^{(1)}(z)) = 0\). Since \(z \in \operatorname {totZ}_1 = \ker (d_1)\), we have \(d_1(z) = 0\), i.e. \(\widetilde{d}_h^0(z_1) + d_v^0(z_2) = 0\) (as \(d_1 = \operatorname {coprod}(\widetilde{d}_h^0, d_v^0)\)).

Applying \(\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon )\) to this equation:

\[ (\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))\bigl(\widetilde{d}_h^0(z_1) + d_v^0(z_2)\bigr) = 0. \]

The \(d_v^0\)-term vanishes since \(d_v^0 = \operatorname {lTensor}(d_F)\), and \((\operatorname {id} \otimes \varepsilon ) \circ (\operatorname {id} \otimes d_F) = \operatorname {id} \otimes (\varepsilon \circ d_F) = 0\) by hypothesis. Thus:

\[ (\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))\bigl(\widetilde{d}_h^0(z_1)\bigr) = 0. \]

By the chain map property (Lemma 228), this equals \(d_B((\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(z_1)) = d_B(\pi _*^{(1)}(z)) = 0\).

Definition 230 Restriction of \(\pi _*\) to Cycles

The restriction of \(\pi _*^{(1)}\) to \(\operatorname {totZ}_1\), landing in \(\ker (d_B)\):

\[ \pi _*|_{\operatorname {cycles}} : \operatorname {totZ}_1 \; \longrightarrow \; \ker (d_B), \quad z \; \mapsto \; \bigl(\pi _*^{(1)}(z),\; \pi _*^{(1)}(z) \in \ker (d_B)\bigr). \]

This is a well-defined \(\mathbb {F}_2\)-linear map by Lemma 229.

Lemma 231 \(\pi _*\) Maps Boundaries to Zero

Under the same hypotheses,

\[ \operatorname {totB}_1^{Z_1} \; \leq \; \ker (\pi _*|_{\operatorname {cycles}}), \]

i.e. \(\pi _*^{(1)}\) vanishes on the total 1-boundaries \(\operatorname {totB}_1\) (embedded in \(\operatorname {totZ}_1\)).

Proof

Let \(z \in \operatorname {totB}_1^{Z_1}\). Rewriting \(\operatorname {totB}_1^{Z_1}\) as the preimage under inclusion of \(\operatorname {totB}_1 = \operatorname {im}(d_2)\), we obtain \(w\) such that \(d_2(w) = z.\operatorname {val}\). In particular, \(z.\operatorname {val}.1 = d_v^1(w)\) (from the first component of \(d_2\)).

We must show \(\pi _*^{(1)}(z.\operatorname {val}) = 0\). By definition of \(\pi _*^{(1)}\) as \(\operatorname {coprod}(\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ),\; 0)\), it suffices to show

\[ (\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(z.\operatorname {val}.1) = 0. \]

Substituting \(z.\operatorname {val}.1 = d_v^1(w) = \operatorname {lTensor}(d_F)(w)\):

\[ (\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(\operatorname {lTensor}(d_F)(w)) = (\operatorname {rid} \circ \operatorname {lTensor}(\varepsilon \circ d_F))(w) = 0, \]

since \(\varepsilon \circ d_F = 0\) by hypothesis.

Definition 232 Induced Map \(H_1(\pi _*)\) on Homology

The map induced by \(\pi _*\) on degree-1 homology,

\[ H_1(\pi _*) : H_1(B \otimes _\varphi F) = \operatorname {totH}_1 \; \longrightarrow \; \ker (d_B), \]

defined as the quotient lift

\[ H_1(\pi _*) \; =\; \operatorname {liftQ}\bigl(\operatorname {totB}_1^{Z_1},\; \pi _*|_{\operatorname {cycles}},\; \pi _*|_{\operatorname {cycles}}(\operatorname {totB}_1^{Z_1}) = 0\bigr). \]
Lemma 233 Surjectivity of \(H_1(\pi _*)\)

Assume additionally that \(\varepsilon : \mathbb {F}_2^{m_2} \to \mathbb {F}_2\) is surjective. Then

\[ H_1(\pi _*) : \operatorname {totH}_1 \; \twoheadrightarrow \; \ker (d_B) \]

is surjective.

Proof

Since \(\varepsilon \) is surjective, there exists \(f_0 \in \mathbb {F}_2^{m_2}\) with \(\varepsilon (f_0) = 1\).

Let \(v \in \ker (d_B)\). We construct a preimage in \(\operatorname {totH}_1\) mapping to \(v\).

First, we show \(\widetilde{d}_h^0(v \otimes f_0) \in \operatorname {im}(d_v^0)\): since \(v \in \ker (d_B)\) and the connection acts as the identity on homology, for each basis vector \(e_{b_1}\) with \(d_B(e_{b_1})(b_0) \neq 0\), the automorphism \((\varphi (b_1,b_0)).\alpha _0\, f_0 - f_0 \in \operatorname {im}(d_F)\). After expanding via twistedDhLin_tmul, the sum splits into a term \(d_B(v) \otimes f_0 = 0\) (since \(v \in \ker (d_B)\)) and a sum of terms of the form \(e_{b_0} \otimes d_F(g)\), each lying in \(\operatorname {im}(\operatorname {lTensor}(d_F))\).

Obtaining \(y\) with \(d_v^0(y) = \widetilde{d}_h^0(v \otimes f_0)\), we set \(z = (v \otimes f_0,\; y)\). Over \(\mathbb {F}_2\), \(\widetilde{d}_h^0(v \otimes f_0) + d_v^0(y) = 0\), so \(z \in \operatorname {totZ}_1\).

The class \([z] \in \operatorname {totH}_1\) satisfies \(H_1(\pi _*)([z]) = \pi _*|_{\operatorname {cycles}}(z)\). By definition, this equals \((\operatorname {rid} \circ (\operatorname {id} \otimes \varepsilon ))(v \otimes f_0) = \varepsilon (f_0) \cdot v = 1 \cdot v = v\).

Lemma 234 Equal Dimension: \(\operatorname {totH}_1\) and \(\ker (d_B)\)

Assume:

  1. \(\varphi \) acts as the identity on the homology of \(F\),

  2. the augmentation induces an isomorphism \(\bar{\varepsilon } : H_0(F) = (\mathbb {F}_2^{m_2} / \operatorname {im}(d_F)) \xrightarrow {\sim } \mathbb {F}_2\),

  3. \(H_0(B) = 0\), i.e. \(\operatorname {im}(d_B) = \mathbb {F}_2^{m_1}\) (surjectivity).

Then

\[ \dim _{\mathbb {F}_2} \operatorname {totH}_1 \; =\; \dim _{\mathbb {F}_2} \ker (d_B). \]
Proof

By Theorem 212, there is a linear equivalence

\[ e : \operatorname {totH}_1 \; \simeq \; \bigl(\ker (d_B) \otimes H_0(F)\bigr) \; \times \; \bigl(H_0(B) \otimes \ker (d_F)\bigr). \]

Hence \(\dim \operatorname {totH}_1 = \dim (\ker (d_B) \otimes H_0(F)) + \dim (H_0(B) \otimes \ker (d_F))\).

Since \(\operatorname {im}(d_B) = \top \), the cokernel \(H_0(B) = \mathbb {F}_2^{m_1} / \operatorname {im}(d_B)\) is the quotient by \(\top \), which is subsingleton. Therefore \(H_0(B) \otimes \ker (d_F)\) is also subsingleton (as a tensor product with a zero module), and its dimension is \(0\).

Since \(\bar{\varepsilon } : H_0(F) \simeq \mathbb {F}_2\), we have

\[ \dim (\ker (d_B) \otimes H_0(F)) = \dim (\ker (d_B) \otimes \mathbb {F}_2) = \dim (\ker (d_B)), \]

using \(\operatorname {lTensor}(\bar{\varepsilon })\) and the right-unit isomorphism \(\ker (d_B) \otimes \mathbb {F}_2 \simeq \ker (d_B)\).

Combining: \(\dim \operatorname {totH}_1 = \dim \ker (d_B) + 0 = \dim \ker (d_B)\).

Under the following three hypotheses:

  1. the connection \(\varphi \) acts as the identity on the homology of \(F\),

  2. \(\bar{\varepsilon } : H_0(F) \xrightarrow {\sim } \mathbb {F}_2\) (the augmentation \(\varepsilon \) induces an isomorphism on \(H_0(F)\)),

  3. \(\operatorname {im}(d_B) = \top \) (surjectivity of \(d_B\), equivalently \(H_0(B) = 0\)),

and assuming \(\varepsilon \) itself is surjective, the map

\[ H_1(\pi _*) : H_1(B \otimes _\varphi F) \; \xrightarrow {\sim }\; H_1(B) = \ker (d_B) \]

is a bijection (hence a linear isomorphism, since it is a linear map between finite-dimensional \(\mathbb {F}_2\)-vector spaces of equal dimension).

Proof

Surjectivity follows from Lemma 233. By Lemma 234, \(\dim \operatorname {totH}_1 = \dim \ker (d_B)\).

We verify that both spaces are finite-dimensional: \(\operatorname {totH}_1\) is finite-dimensional because by Theorem 212 it is linearly equivalent to a product of tensor products of finite-dimensional spaces; \(\ker (d_B)\) is finite-dimensional as a submodule of a finite-dimensional space.

Since \(H_1(\pi _*)\) is a surjective linear map between finite-dimensional spaces of equal dimension, it is also injective by the rank-nullity theorem (as applied via LinearMap.injective_iff_surjective_of_finrank_eq_finrank). Hence \(H_1(\pi _*)\) is bijective.

Definition 236 \(H_1(\pi _*)\) as a Linear Equivalence

Under the same hypotheses as Theorem 235, the map \(H_1(\pi _*)\) is packaged as a linear equivalence

\[ H_1(\pi _*)^{\simeq } : \operatorname {totH}_1 \; \xrightarrow {\; \sim \; } \ker (d_B), \]

constructed via LinearEquiv.ofBijective applied to the bijection of Theorem 235.

Definition 237 Induced Cohomology Isomorphism \(H^1(\pi ^*)\)

Under the same hypotheses, and assuming \(\operatorname {totH}_1\) and \(\ker (d_B)\) are finite-dimensional, the dual of the isomorphism \(H_1(\pi _*)^{\simeq }\) yields a linear equivalence

\[ H^1(\pi ^*) : \operatorname {Hom}(\ker (d_B),\, \mathbb {F}_2) \; \xrightarrow {\; \sim \; } \operatorname {Hom}(\operatorname {totH}_1,\, \mathbb {F}_2). \]

Explicitly, for \(f \in \operatorname {Hom}(\ker (d_B), \mathbb {F}_2)\), we set \(H^1(\pi ^*)(f) = f \circ H_1(\pi _*)^{\simeq }\), and the inverse sends \(g \mapsto g \circ (H_1(\pi _*)^{\simeq })^{-1}\).

Theorem 238 \(H^1(\pi ^*)\) Is an Isomorphism

Under the same hypotheses as Theorem 235, the cochain map \(\pi ^*\) induces an isomorphism on degree-1 cohomology:

\[ H^1(\pi ^*) : H^1(B) \; \xrightarrow {\; \sim \; } H^1(B \otimes _\varphi F). \]
Proof

The map \(H^1(\pi ^*)\) is defined as \(H^1(\pi ^*) = H_1(\pi _*)^{\simeq ,\operatorname {dual}}\), the dual of the linear equivalence \(H_1(\pi _*)^{\simeq }\). Since \(H_1(\pi _*)^{\simeq }\) is a linear equivalence (and in particular bijective), its dual \(H^1(\pi ^*)\) is also a linear equivalence, and its underlying linear map is bijective. This follows directly from LinearEquiv.bijective applied to \(H^1(\pi ^*)\).