Fano Threefolds and Exceptional Collections

• Exceptional collections are ordered sets of vector bundles in D^b(X) that generate the derived category of Fano threefolds.
• The work details methods using braid group actions, mutations, and restrictions to anticanonical K3 divisors for classifying collections in Picard rank one Fano threefolds.
• It connects mirror symmetry and Landau–Ginzburg models with motivic decompositions, offering insights into the arithmetic properties of these varieties.

Exceptional collections for Fano threefolds constitute a foundational structure within the study of derived categories in algebraic geometry, encoding deep information about the geometry and arithmetic of these varieties. They interact with vector bundle theory, semiorthogonal decompositions, the theory of motives, Homological Mirror Symmetry, and the arithmetic of Landau–Ginzburg models. The subject is particularly rigid and explicit on certain classes of Fano threefolds, notably those of Picard rank one with vanishing $h^{1,2}$, as well as the toric Fano threefolds. This article synthesizes the recent classification results, group-theoretic approaches to mutations and spherical twists, geometric constructions of exceptional bundles, and the interplay with mirror symmetry.

1. Classification and Existence of Full Exceptional Collections

A smooth projective variety $X$ admits a full exceptional collection of length $n$ in its bounded derived category $D^b(X)$ precisely if $\mathrm{rk}\, K_0(X) = n = \dim X + 1$ in the threefold case. For Fano threefolds, this criterion, combined with the Iskovskikh–Mori–Mukai classification, isolates four deformation families of Picard rank one and $h^{1,2}=0$: projective 3-space $\mathbb{P}^3$, the quadric threefold $Q_3\subset\mathbb{P}^4$, the Fano $V_5$ (the linear section of the Grassmannian $\mathrm{Gr}(2,5)$), and $V_{22}$ (the Mukai–Umemura threefold) (Nordskova et al., 2024). Each of these admits an explicit full exceptional collection of four vector bundles:

• $\mathbb{P}^3$: $$(\mathcal{O}, \mathcal{O}(1), \mathcal{O}(2), \mathcal{O}(3))$$
• $Q_3$: $$(\mathcal{O}, \mathcal{S}^\vee, \mathcal{O}(1), \mathcal{O}(2))$$, $\mathcal{S}$ the spinor bundle
• $V_5$: $$(\mathcal{O}, \mathcal{U}, \mathcal{U}^\perp(1), \mathcal{O}(1))$$, where $\mathcal{U}$ is the tautological rank 2 bundle
• $V_{22}$: Kuznetsov’s exceptional quadruple of rank 2 bundles

The fundamental theorem states that $D^b(X)$ admits a full exceptional collection of four vector bundles if and only if $X$ is one of these four, up to deformation, equivalently if $X$ is Fano of Picard rank $1$ and $h^{1,2}(X)=0$ (Nordskova et al., 2024).

2. Structure and Generation of the Exceptional Collection Set

On the aforementioned Fano threefolds, every full exceptional collection consists of shifts of vector bundles (Nordskova et al., 2024). The group $B_4 \ltimes \mathbb{Z}^4$, generated by the braid group (mutations) and the shift functor, acts freely and transitively on the set of full exceptional collections of length four. This is established by analyzing the interplay of these collections with anticanonical K3 divisors. Specifically:

• Restriction to a generic smooth anticanonical K3 divisor $i:Y\hookrightarrow X$ sends exceptional bundles $E$ to spherical objects $i^*E\in D^b(Y)$ (in the sense of Seidel–Thomas).
• Mutations correspond to Hurwitz actions on the tuple of spherical twists $(T_{i^* E_0},\dots,T_{i^* E_3})$.
• The product $$T_{i^* E_0}\circ T_{i^* E_1}\circ T_{i^* E_2}\circ T_{i^* E_3}$$ equals the Serre functor, forcing all four-tuples into a single Hurwitz orbit.
• The subgroup generated by the four spherical twists is free of rank four; there are no relations among the $T_{i^* E_j}$ other than $(T_{i^* E_j})^2 = \mathrm{id}$ (Nordskova et al., 2024).

3. Spherical Objects and Spherical Twist Groups on K3 Anticanonical Divisors

The geometric mechanism underlying the above classification is the transfer of categorical data from $D^b(X)$ to $D^b(Y)$, with $Y$ a generic anticanonical K3 divisor in $X$. On a K3 surface $Y$ with $\rho=1$, an object $S\in D^b(Y)$ is spherical if $\omega_Y\otimes S\simeq S$ and $\mathrm{Hom}^i(S, S)=\mathbb{C}$ for $i=0,2$ and $0$ otherwise.

Given such $S$, the spherical twist $T_S$ is the corresponding autoequivalence. For any full exceptional collection $(E_0,\dots,E_3)$ on $X$, the restrictions $(i^*E_0, \dots, i^*E_3)$ generate a subgroup $G\subset\mathrm{Aut} D^b(Y)$ free of rank four (Nordskova et al., 2024). This freeness reflects the topological structure of the configuration of spherical objects; no relations hold among the $T_{i^*E_j}$ except the involutivity involutive property.

4. Toric Fano Threefolds and Exceptional Line Bundle Collections

Bernardi–Tirabassi and subsequent authors proved Bondal's conjecture for all smooth toric Fano threefolds: $D^b(X)$ admits a full strongly exceptional collection of line bundles, constructed explicitly via Frobenius pushforwards and combinatorics of the fan (Uehara, 2010, Anderson, 2023). There are 18 such isomorphism classes, as classified by Batyrev; for 16 of these, the existence of a full strong collection of line bundles is established by explicit construction and by the cellular and diagonal resolutions of the diagonal (Anderson, 2023). The two highest Picard rank cases, $X_{17} = \mathrm{Bl}_{(p,\ell)}\mathbb{P}^3$ and $X_{18} = \mathrm{Bl}_{p,q,r,s}\mathbb{P}^3$, present obstructions: no ordering of candidate line bundles achieves the required vanishing.

For all toric Fano threefolds except these two, the direct summands of a sufficiently high Frobenius push-forward $(F_{m*}\mathcal{O}_X)$ form such a collection, and the birational geometry between maximal models and their descendants enables the construction for all lower-rank cases (Uehara, 2010, Bernardi et al., 2010). For explicit examples (e.g. $\mathbb{P}^3$ and $\mathbb{P}^2\times\mathbb{P}^1$), see Table 1 in (Anderson, 2023).

5. Exceptional Collections on Prime Fano Threefolds and Mukai Bundles

In the prime Fano case (Picard rank one, genus $g\ge 6$), the existence and uniqueness of certain exceptional vector bundles, known as Mukai bundles, is affirmed via Lazarsfeld's construction, Brill–Noether theory on K3 surfaces, and stability theory (Bayer et al., 2024).

• Each prime Fano threefold $X$ possesses a Mukai bundle $U_r$ of rank $r$, $c_1(U_r) = -H$ (with $H$ the anticanonical class).
• A semi-orthogonal decomposition $$D^b(X) = \langle U_r, \mathcal{O}_X, \mathcal{O}_X(H), U_r^\vee(H) \rangle$$ yields a full exceptional collection of length four (Bayer et al., 2024).
• For certain even-genus cases, the residual category is a Calabi–Yau (Kuznetsov) component, and deformations connect the nontrivial residual categories of del Pezzo threefolds and their associated higher-genus Fano threefolds (Kuznetsov et al., 2023).

6. Mirror Symmetry, Landau–Ginzburg Models, and Motivic Aspects

For Fano threefolds, both the length and precise structure of maximal exceptional collections are often predicted by Homological Mirror Symmetry and the properties of Landau–Ginzburg models. Przyjalkowski establishes a tight correspondence between the number of ordinary double points in non-central fibers of an LG model and the maximal length of semiorthogonal exceptional blocks in $D^b(X)$ (Przyjalkowski, 23 Jan 2026). In the toric setting, mirror LG superpotentials give rise to a phase map whose image in Pic$(X)$ recovers the full strongly exceptional collection, with monodromy matching Hom-space structures (“M-alignment”) (Jerby, 2016).

Motivic consequences are encoded by Marcolli–Tabuada: if $D^b(X)$ admits a full exceptional collection, then the Chow motive $M(X)$ decomposes as a sum of tensor powers of the Lefschetz motive; for a Fano threefold with $H^\text{odd}(X) = 0$, the length of any full collection matches the rank of $H^{\text{even}}(X)$ (Marcolli et al., 2012).

7. Further Phenomena and Open Problems

For many non-toric, non-prime, or singular Fano threefolds, the structure of full exceptional or semiorthogonal collections is subtler. Known obstructions arise for higher Picard rank toric examples and may also appear for certain degenerations of low-genus Fano threefolds (Przyjalkowski, 23 Jan 2026, Anderson, 2023). Open problems include:

• The existence of full exceptional collections (possibly of vector bundles, not just line bundles) for all smooth Fano threefolds.
• The extension of the “M-aligned” mirror correspondence to higher Picard rank and non-toric cases.
• The classification of residual (Kuznetsov) components in various geometric settings and their connections to Calabi–Yau categories.

The theory of exceptional collections for Fano threefolds thus exhibits remarkably rigid, explicit, and often mirror-symmetric phenomena, tightly linking algebraic and symplectic geometry, representation theory, and arithmetic.