--- title: Decomposition of the Diagonal url: https://www.emergentmind.com/topics/decomposition-of-the-diagonal type: topic --- # Decomposition of the Diagonal “Decomposition of the diagonal” is a polysemous term whose central meaning depends on context. In algebraic geometry, it denotes Chow-theoretic or cohomological identities for the class of the diagonal $\Delta_X\subset X\times X$, often with support constraints on the correcting terms, and it is closely tied to universal $\mathrm{CH}_0$-triviality, Abel–Jacobi maps, integral Hodge classes, and obstructions to stable rationality [1005.5621, 2007.07470]. In the geometry of K3 surfaces it also refers to the Beauville–Voisin decomposition of the small diagonal in $S^3$ [1611.10064]. In derived algebraic geometry it appears as an explicit resolution of the diagonal on root stacks [2309.06788]. In homotopy theory it names decompositions of diagonal arrangements [1403.7864]. In quantum information, by contrast, it refers to exact circuit factorizations of diagonal unitary operators [1511.05758, 2403.02109]. ## 1. Algebraic-geometric core: Chow and cohomological decompositions For a smooth projective variety $Y$ of dimension $n$, with diagonal $\Delta_Y\subset Y\times Y$, the basic Bloch–Srinivas pattern is an equality $$ N\Delta_Y=Z_1+Z_2\in CH^n(Y\times Y), $$ with $\operatorname{Supp}(Z_1)\subset D\times Y$, $\operatorname{Supp}(Z_2)\subset Y\times W$, where $D\subset Y$ is a proper closed subset, $W\subset Y$ is a lower-dimensional subset, and $N\neq 0$ [1005.5621]. The corresponding cohomological form is $$ N[\Delta_Y]=[Z_1]+[Z_2]\quad\text{in }H^{2n}(Y\times Y,\mathbb{Z}), $$ with the same support conditions, and an integral cohomological decomposition of the diagonal is the case $N=1$ [1005.5621]. A more general formalism allows an $R$-decomposition of type $(W_1,W_2)$ for a cycle class $Z\in CH^{d_X}(X\times_KX)_R$, meaning $$ Z=Z_1+Z_2\in \mathrm{CH}^{d_X}(X\times_KX)_R $$ with $Z_1$ supported on $W_1\times_KX$ and $Z_2$ supported on $X\times_KW_2$ [2007.07470]. A strict $R$-decomposition is the special case where $Z_1$ is supported on $D\times_KX$ for a divisor $D\subset X$ and $Z_2$ is supported on $X\times_KW_2$ for a $0$-dimensional $W_2\subset X$ [2007.07470]. This formalism is tightly linked to $\mathrm{CH}_0$-theory. For a smooth proper $X/K$, $X$ is universally $(\mathrm{CH}_0)_R$-trivial if and only if $\Delta_X$ admits a strict $R$-decomposition [2007.07470]. Stable rationality implies universally $\mathrm{CH}_0$-triviality for smooth projective varieties; thus, for stably rational $X$, $\Delta_X$ admits a strict Chow decomposition [2007.07470]. If $X$ is rationally chain connected over a perfect field, then $\mathrm{CH}_0(X)_{\mathbb{Q}}$ is universally trivial, hence some nonzero multiple $N\Delta_X$ admits a strict Chow decomposition [2007.07470]. These statements underlie a standard distinction. A decomposition of the diagonal is not merely an equality in a Chow group: the support conditions are the decisive part of the notion, because they force the induced correspondences to factor through lower-dimensional varieties. That factorization is the mechanism behind consequences for torsion, algebraicity, and birational geometry. ## 2. Intermediate Jacobians, Abel–Jacobi maps, and threefolds For a smooth projective threefold $Y$ with $H^{3,0}(Y)=0$, the intermediate Jacobian is $$ J(Y):=H^3(Y,\mathbb{C})/(F^2H^3(Y)\oplus H^3(Y,\mathbb{Z})), $$ and the Abel–Jacobi map on homologically trivial codimension-$2$ cycles is $$ AJ_Y:CH^2(Y)_{hom}\longrightarrow J(Y). $$ Voisin’s analysis organizes the relation between diagonal decompositions and families of $1$-cycles through two properties: (*) asks for a smooth projective variety $B$ and a codimension-$2$ cycle $Z\subset B\times Y$ such that the induced morphism $\phi_Z:B\to J(Y)$ is surjective with rationally connected general fiber; (**) is the corresponding statement for families of cycles in a fixed cohomology class $\alpha\in H^4(Y,\mathbb{Z})$, where the target is the Deligne-twisted torsor $J(Y)_\alpha$ arising from $$ 0\rightarrow J(Y)\rightarrow H^4_D(Y,\mathbb{Z}(2))\xrightarrow{o} Hdg^4(Y,\mathbb{Z})\rightarrow 0. $$ These constructions make the Abel–Jacobi map the intermediary between diagonal decompositions and the geometry of $1$-cycles [1005.5621]. A central theorem states that if $Y$ is a smooth projective threefold with $H^3(Y,\mathcal{O}_Y)=0$ and $Y$ admits an integral cohomological decomposition of the diagonal, then $H^4(Y,\mathbb{Z})$ is generated by classes of algebraic cycles, $H^p(Y,\mathbb{Z})$ has no torsion for any $p$, and $Y$ satisfies condition (*) [1005.5621]. Conversely, assuming those three properties and the existence on $J(Y)$ of a $1$-cycle $\Gamma$ of class $$ \frac{[\Theta]^{g-1}}{(g-1)!},\qquad g=\dim J(Y), $$ one recovers an integral cohomological decomposition of the diagonal [1005.5621]. In the same circle of ideas, a cohomological decomposition $$ N[\Delta_Y]=[Z]+[Z']\quad\text{in }H^6(Y\times Y,\mathbb{Z}), $$ with $\operatorname{Supp}(Z)\subset D\times Y$ and $\dim W\leq 1$ for the support of $Z'$, implies that $N$ annihilates the torsion of $H^p(Y,\mathbb{Z})$ for all $p$, annihilates $Z^4(Y)$, forces $H^i(Y,\mathcal{O}_Y)=0$ for $i>1$, and yields a cycle $Z\subset J(Y)\times Y$ with $\phi_Z=N\,Id_{J(Y)}$ [1005.5621]. Cubic threefolds provide the main explicit class of examples. For a smooth cubic threefold $Y\subset \mathbb{P}^4$, Iliev–Markushevich and Tikhomirov proved that the Abel–Jacobi map $$ \phi_{5,1}:M_{5,1}\longrightarrow J(Y) $$ from a desingularization of the Hilbert scheme of degree $5$, genus $1$ curves is surjective with general fiber $\mathbb{P}^5$, and Voisin proved that for a general cubic threefold the analogous map $$ \phi_{6,1}:M_{6,1}\longrightarrow J(Y) $$ for degree $6$ elliptic curves is surjective with rationally connected general fiber [1005.5621]. These results supply the families required for (*) and (**), and they feed directly into the proof that degree $4$ integral Hodge classes on suitable fibrations into cubic threefolds are algebraic [1005.5621]. ## 3. Stable rationality, degeneration methods, and positive characteristic In later work the integral decomposition of the diagonal became a birational invariant and an obstruction to stable rationality. Over algebraically closed fields of positive characteristic, this obstruction has been extended from Hodge-theoretic intermediate Jacobians to $\ell$-adic cohomology and algebraic representatives. For a smooth projective threefold $X$ over an algebraically closed field $k$, if $\Delta_X$ admits a strict cohomological $\mathbb{Z}$-decomposition with respect to $H^\bullet(-,\mathbb{Z}_\ell)$, then $H^1(X,\mathbb{Z}_\ell)=0$, $H^{2i}(X,\mathbb{Z}_\ell(i))$ is $\mathbb{Z}$-algebraic for all $i$, $T_\ell\lambda^2$ and $T_\ell\phi^2_{X/k}$ are isomorphisms, $\mathrm{Tors}\,H^\bullet(X,\mathbb{Z}_\ell)=0$, $\mathrm{Ab}^2_{X/k}$ admits a universal codimension-$2$ cycle, and the minimal class $$ \frac{[\Theta_X]^{g-1}}{(g-1)!}\in H^{2g-2}(\mathrm{Ab}^2_{X/k},\mathbb{Z}_\ell(g-1)) $$ is $\mathbb{Z}$-algebraic [2007.07470]. A partial converse gives necessary and sufficient conditions for a strict cohomological $\mathbb{Z}_\ell$-decomposition in terms of precisely these data [2007.07470]. A key replacement for the complex intermediate Jacobian is the second algebraic representative $\mathrm{Ab}^2_{X/K}$. Under geometric rational chain connectedness or stable rationality, there exists a canonical symmetric $K$-isogeny $$ \Theta_X:\mathrm{Ab}^2_{X/K}\longrightarrow \widehat{\mathrm{Ab}^2_{X/K}}, $$ constructed from miniversal or universal codimension-$2$ cycles; in characteristic $0$ it agrees with the principal polarization on $J^3(X)$ induced by Hodge theory [2007.07470]. This transfer of the polarization package is what allows Voisin-type obstructions to be reformulated in positive characteristic. The quartic double solid with nodes is the main test case. In characteristic greater than two, a desingularization of a very general quartic double solid with seven nodes satisfies conditions (1)–(4) and (6) of the $\ell$-adic criterion but fails condition (5): it has no universal codimension-$2$ cycle class [2007.07470]. This produces stably irrational unirational examples detected by the diagonal obstruction even when classical invariants do not detect irrationality [2007.07470]. Recent degeneration arguments have pushed the obstruction further. For a very general degree-$d$ hypersurface or a very general $(2,d)$ complete intersection, Fiammengo–Lüders reduce the existence of a decomposition of the diagonal to the corresponding question for cubic hypersurfaces together with Voisin’s essential dimension condition. In particular, if a very general degree-$d$ hypersurface $H\subset \mathbb{P}^{n+1}$ does not admit a decomposition of the diagonal and satisfies $\mathrm{essdim}(H)=\dim(H)$, then a very general $(2,d)$ complete intersection in $\mathbb{P}^{n+3}$ does not admit a decomposition of the diagonal; similarly, under the cubic input, a very general degree-$d$ hypersurface in $\mathbb{P}^{n+2}$ does not admit a decomposition of the diagonal [2510.07111]. Using a recent result on cubic threefolds, they also obtain a new proof that a very general complex quartic $4$-fold and a very general complex $(2,3)$-complete intersection $4$-fold do not admit a decomposition of the diagonal and are therefore not retract rational [2510.07111]. ## 4. The small diagonal of a K3 surface For a smooth projective K3 surface $S$ over $\mathbb{C}$, the small diagonal is $$ \Delta_S=\{(x,x,x)\mid x\in S\}\subset S^3, $$ viewed as a class in $CH^4(S^3)$. Beauville and Voisin proved the exact identity $$ \begin{aligned} [\Delta_S] &= p_{12}^*[\Delta_{S}] \cdot p_3^*(o_S)\;+\;p_{23}^*[\Delta_{S}] \cdot p_1^*(o_S)\;+\;p_{13}^*[\Delta_{S}] \cdot p_2^*(o_S)\\ &\quad -\;p_1^*([S])\cdot p_2^*(o_S)\cdot p_3^*(o_S)\;-\;p_1^*(o_S)\cdot p_2^*([S])\cdot p_3^*(o_S)\;-\;p_1^*(o_S)\cdot p_2^*(o_S)\cdot p_3^*([S]), \end{aligned} $$ where $o_S\in CH_0(S)$ is the canonical $0$-cycle of degree $1$ given by the class of any point lying on a rational curve on $S$ [1611.10064]. This identity has two classical consequences. For divisor classes $D,D'\in \mathrm{Pic}(S)$, $$ D\cdot D'=\deg(D\cdot D')\,o_S\in CH_0(S), $$ and $$ c_2(S)=24\,o_S. $$ Thus, although $CH_0(S)$ is not equal to $\mathbb{Z}$ for a K3 surface, the subring generated by divisor classes behaves as though all $0$-cycles coming from divisor intersections collapse to the canonical class [1611.10064]. Bazhov gave a new proof of this decomposition under the hypotheses $\mathrm{Pic}(S)=\mathbb{Z}\cdot L$, $L$ very ample, and $L^2=2g-2>4$, replacing Beauville–Voisin’s use of one-parameter families of elliptic curves by an explicit projective-geometric construction in the embedding $S\hookrightarrow \mathbb{P}^g$ [1611.10064]. The proof uses codimension-$2$ linear sections, special multiplicity loci, universal incidence cycles $\mathrm{II2}$ and $\mathrm{II3}$ in $(\mathbb{P}^g)^2$ and $(\mathbb{P}^g)^3$, and a decomposition of their restrictions to $S\times S$ and $S^3$. In this setting, “decomposition of the diagonal” refers not to the big diagonal in $S\times S$ that appears in rationality questions, but to the small diagonal in $S^3$ and the multiplicative structure of the Chow ring. ## 5. Categorical and homotopical variants A different usage occurs for root stacks. For the $r$-th root stack $$ \sqrt[r]{(L,s)/X} $$ of a line bundle with section cutting out a divisor $D$, an explicit resolution of the diagonal on $$ X_{D,r}\times_X X_{D,r} $$ is constructed from equivariant kernels $\tau_{m,n,X}$ and the associated Fourier–Mukai endofunctors $\overline{\tau}_{m,n,X}$ [2309.06788]. The key identity $$ \overline{\tau}_{0,r-1,X}\cong L\theta_X^{r*}R\theta_{X*}^{r} $$ identifies the projector onto the “bulk” subcategory, while the successive differences are built from functors $\triangleright_{m,X}\circ\triangleleft_{i,X}$ corresponding to $\mu_r$-weight pieces [2309.06788]. The resulting semi-orthogonal decomposition is $$ D^b_{\mathrm{coh}(\sqrt[r]{(L,s)/X})} = \Big\langle \mathcal{D}_{r,D}^{\,i-r+1},\dots,\mathcal{D}_{r,D}^{\, -1}, L\theta_X^{r*}D^b_{\mathrm{coh}(X)}, \mathcal{D}_{r,D}^{\, 0},\dots,\mathcal{D}_{r,D}^{\, i-1} \Big\rangle, $$ where the $\mathcal{D}_{r,D}^{\,m}$ are the images of fully faithful functors from $D^b_{\mathrm{coh}(D)}$ [2309.06788]. Here the phrase denotes a Fourier–Mukai resolution of $\Delta_*\mathcal{O}$, not a Bloch–Srinivas decomposition in the Chow group. An analogous shift of meaning appears in topology. For a simplicial complex $K$ on $[m]$ and a space $X$, the partially diagonal subspaces $$ A_\sigma(X)=\{(x_1,\dots,x_m)\in X^m\mid x_i=x_j\text{ for all }i,j\in [m]\setminus \sigma\} $$ assemble into the diagonal arrangement $$ A_K(X)=\bigcup_{\sigma\in K}A_\sigma(X)\subset X^m. $$ If $2(\dim K+1)