Integral Regulators for Higher Chow Complexes

ny Z\in Z^p(X,n) may be lifted to c^p(X,n), and we may add degenerate cycles to any element of c^p(X,n) to force it into Z^p_{\infty,n}(X,n). The (well-defined) map given by this process is an isomorphism, and we shall tacitly equate Z^p(X,n) and Z^p_{\infty,n}(X,n) in what follows.

For each integer l \leq n-1, define h^l\colon \square^{n+1}\rightarrow\square^n by

and for Z\in Z^p(X,n), define H^l(Z):=(-1)^{l}\big(h^l\big)^{-1}(Z)\in Z^p(X,n+1). If l\geq n, set H^l(Z)=0. The map

stabilizes in any degree and so defines an endomorphism \phi\colon Z^p(X,\bullet)\rightarrow Z^p(X,\bullet), which is visibly homotopic to the identity.

To determine its image, write (for any Z\in Z^p(X,n))

and

where the notation means that we pull back (the equations defining) \partial^{\infty}_{n-k+1}Z via h^l (or h^{l-1}): \square^n \to \square^{n-1}. Thus we have

In particular, for Z\in Z^p_{\infty,i}(X,\bullet), we have \big(d\circ H^l+H^l\circ d\big)Z=0 for l \leq n-i-2. For l=n-i-1, we find

which belongs to Z^p_{\infty,i-1}. Applying \big(\mathrm{Id}-\big(\partial\circ H^{l+1}+H^{l+1}\circ\partial\big)\big) then maps Z' to some Z''\in Z^p_{\infty,i-2}, and so forth until finally we reach Z^p_{\infty,0}(X,n)=\mathscr{N}^p(X,n). Since all the \partial\circ H^l + H^l\circ \partial are zero on Z^p_{\infty,0}, \phi|_{\mathscr{N}} gives the identity on normalized cycles.

We have thus constructed a morphism \phi\colon Z^p(X,\bullet)\to \mathscr{N}^p(X,\bullet), whose composition with the inclusion \mathscr{N}^p\hookrightarrow Z^p is homotopic to (resp. equal to) the identity on Z^p (resp. \mathscr{N}^p); thus \phi and the inclusion are both quasi-isomorphisms.

et F(z)={\rm i} z-1, G(z)=-\tfrac{(1+z)(1+3z)}{(1+{\rm i} z)(1-2z)}, and H(z)=\tfrac{{\rm i} z-1}{3+z}. Then we have Z=(F(z),G(z),H(z))_{z\in\mathbb{P}^1}\in Z^2(\text{pt},3); but for all \varepsilon>0, Z \notin Z^2_{\varepsilon}(\text{pt},3). More precisely, for any \epsilon>0, we have \dim_{\mathbb{R}}\big(Z\cap T^{\epsilon}_{z_1}\cap T^{\epsilon}_{z_2}\cap T^{\epsilon}_{z_3}\big)=0, not -1 (i.e., empty) as required for a proper-analytic intersection. To see this, we need to find a value of z for which \mathrm{arg}(F), \mathrm{arg}(G), \mathrm{arg}(H) equal to \pi-\epsilon. Such a value is given by z=\tan(\epsilon).

onsider the projection (\mathbb{C}^*)^n\rightarrow\big(S^1\big)^n \cong (\mathbb{R}/2\pi\mathbb{Z})^n defined by \big(r_1e^{{\rm i}\epsilon_1},\dots,r_ne^{{\rm i}\epsilon_n}\big)\mapsto(\epsilon_1,\dots,\epsilon_n), whose fibers are T^{\epsilon_1}_{z_1}\cap\cdots\cap T^{\epsilon_n}_{z_n}. There is also a natural 2^n:1 map \big(S^1\big)^n\rightarrow\big(\mathbb{P}_{\mathbb{R}}^1\big)^n by taking slopes: (\epsilon_1,\dots,\epsilon_n)\mapsto(\tan\epsilon_1,\dots,\tan\epsilon_n). The composite map \Theta^n \colon (\mathbb{C}^*)^n\rightarrow\big(S^1\big)^n\rightarrow\big(\mathbb{P}_{\mathbb{R}}^1\big)^n is real algebraic, sending (x_1+{\rm i} y_1,\dots,x_n+{\rm i} y_n)\mapsto(y_1/x_1,\dots, y_n/x_n).

Now let Z\in\mathscr{N}^p(X,n) be given. Set Z^*:=\bar{Z}\cap(X\times (\mathbb{C}^*)^n), and let \tilde{Z^*} be its resolution of singularities. The intersections of Z^* with the fibers of \Theta^n_X\colon X\times(\mathbb{C}^*)^n\rightarrow X\times \big(\mathbb{P}_{\mathbb{R}}^1\big)^n are {Z^*}\cap\big(X\times \big\{T^{-\epsilon_1}_{z_1}\cap\cdots\cap T^{-\epsilon_n}_{z_n}\big\}\big). Write \Theta_Z for the composition of \tilde{Z}^*\to X\times (\mathbb{C}^*)^n with \Theta^n_X. The set of \underline{\epsilon} for which these intersections are good is the complement of the non-flat locus \Delta \subset\big(\mathbb{P}^1_{\mathbb{R}}\big)^n of \Theta_Z (see [No]). Since the flat locus of an algebraic map is Zariski open, \Delta \subset \big(\mathbb{P}^1_{\mathbb{R}}\big)^n is a real subvariety, which is proper by dimension considerations. (That is to say, if all the fibers had real dimension >2\dim _{\mathbb{C}}(Z^*)-n, then Z^* would have real dimension >2\dim _{\mathbb{C}}(Z^*), which is a contradiction.)

Therefore the preimage \tilde{\Delta} of \Delta in \big(S^1\big)^n is real analytic. By the form of the inequalities in B_{\varepsilon}, we know that we can choose an \varepsilon>0 such that B^n_{\varepsilon}\cap\tilde{\Delta}=\varnothing. (This follows from [LR, Theorem 1] for \tilde{\Delta}, and the fact that all derivatives of e^{-1/x} limit to 0 at 0.) This means that Z intersects X\times \big(T^{\epsilon_1}_{z_1}\cap\cdots\cap T^{\epsilon_n}_{z_n}\big) properly \forall\,\underline{\epsilon}\in B^{\bullet}_{\varepsilon}, as desired.

Repeating the argument for X\times(\mathbb{C}^*)^i\times\big(\mathbb{P}_{\mathbb{C}}^1\big)^{n-i} and X\times(\mathbb{C}^*)^i\times(\{0,\infty\})^k\times\big(\mathbb{P}_{\mathbb{C}}^1\big)^{n-i-k}, we pick the minimum of the required values of \varepsilon, so that Z intersects X\times \big(T^{\epsilon_1}_{z_1}\cap\cdots\cap T^{\epsilon_i}_{z_i}\big) and X\times \big(T^{\epsilon_1}_{z_1}\cap\cdots\cap T^{\epsilon_i}_{z_i}\cap\partial^k\Box^n\big) properly \forall\, i,k,\underline{\epsilon}\in B^{\bullet}_{\varepsilon}, which means Z\in \mathscr{N}^p_{\varepsilon}(X,n).

ccording to [KLM, equations (5.2), (5.3), (5.4)], generally we have (for \mathcal{R}^{\underline{\epsilon}}_{\Box,n} \in (C^n_{\mathscr{D}}(Y_{I,f})), without the loss of generality we take I=\{1,\dots,n\}^c):

where for the \delta in the last line, we only consider the components mapping into C^n_{\mathscr{D}}(Y_{I,f}). This tells us that

for any n; thus each component of \mathbb{D}\mathcal{R}^{\underline{\epsilon}}_{\Box} is 0, and so \mathcal{R}^{\underline{\epsilon}}_{\Box}\in\operatorname{Ker}(\mathbb{D}) is a 0-cocycle.

n any given n-“face” (\cong \big(\mathbb{P}^1\big)^n) we have

Noting that D\mathcal{R}^{\epsilon}_{z_i}=-2\pi{\rm i}(\delta_{(z_i)},\delta_{(z_i)},0)=:-2\pi{\rm i} \Delta_{(z_i)} (which commutes with other triples) and D\mathcal{S}^{\epsilon,\epsilon'}_{z_i}=\mathcal{R}^{\epsilon}_{z_i}-\mathcal{R}^{\epsilon'}_{z_i}, we can rewrite this expression by applying the telescoping method and rearranging the order of the summation. Denoting (-1)^{k-1}_l := \begin{cases} (-1)^{k-1}, & l>k, \\ (-1)^{k-2}, & l<k, \end{cases}

This tells us

holds on each n-face of \big(\mathbb{P}^1\big)^n. Thus \mathbb{D}\mathcal{S}^{\underline{\epsilon},\underline{\epsilon'}}_{\Box}=\mathcal{R}^{\underline{\epsilon}}_{\Box}-\mathcal{R}^{\underline{\epsilon'}}_{\Box} holds.

or Z\in\mathscr{N}^p_{\epsilon}(X,n), \partial Z only lives in X\times\partial^{\infty}_n\Box^n by definition. Thus in (?), the only relevant term in the braces is \mathcal{S}_{n-1}^{\{\epsilon_1,\ldots,\epsilon_{n-1}\},\{\epsilon_1 ',\ldots,\epsilon_{n-1} '\}}. Therefore Proposition 6.4 implies at once that \mathcal{R}^{n,\underline{\epsilon}}_{\varepsilon} (Z)- \mathcal{R}^{n,\underline{\epsilon '}}_{\varepsilon}(Z) = D \mathcal{S}^{\underline{\epsilon},\underline{\epsilon'}}_{\varepsilon}(Z) + \mathcal{S}^{\underline{\epsilon},\underline{\epsilon'}}_{\varepsilon}(\partial Z), so that \mathcal{R}_{\varepsilon}^{\bullet,\underline{\epsilon}}\sim \mathcal{R}_{\varepsilon}^{\bullet,\underline{\epsilon}'} as claimed.

ince \mathscr{N}^p_{\varepsilon}(X,\bullet)\subset\mathscr{N}^p(X,\bullet), we have H_n\big(\mathscr{N}^p_{\varepsilon}(X,\bullet)\big) maps to H_n(\mathscr{N}^P(X,\bullet))= {\rm CH}^P(X,n) for every \varepsilon, hence there exists a natural map \lim\limits_{\overrightarrow{\varepsilon}}H_n\big(\mathscr{N}^p_{\varepsilon}(X,\bullet)\big)\rightarrow {\rm CH}^p(X,n). Since \bigcup\mathscr{N}^p_{\varepsilon}(X,\bullet)=\mathscr{N}^p(X,\bullet), this map is surjective. To show it is injective, consider \xi\in {\rm CH}^p(X,n), and \tilde{\xi}, \tilde{\xi}' be two representations of \xi in the following sequence:

We need to show that \tilde{\xi} and \tilde{\xi}' will eventually merge at some \epsilon, that is to say, \bigcup\partial\mathscr{N}^p_{\varepsilon}(X,n+1)=\partial\mathscr{N}^p(X,n+1), which directly comes from the property of normalized cycle and \bigcup\mathscr{N}^p_{\varepsilon}(X,\bullet)=\mathscr{N}^p(X,\bullet).