betaFeedback

WorksMathematics

Boij-S"oderberg and Harder-Narasimhan polygons of vector bundles on projective space

Abstract

Eisenbud and Schreyer proved that the cohomology table of a vector bundle FF on Pn\mathbb{P}^n is a unique positive combination of supernatural tables along a chain of root sequences. The slopes of the pieces, weighted by their ranks, form a concave polygon ΨFΨ_F, the Boij-Söderberg polygon, with the same endpoints as the Harder-Narasimhan polygon PFP_F. We prove that PF≤ΨFP_F\leqΨ_F, in any characteristic, by testing subsheaves against sheaves with pure resolutions on thickened lines; the bound is optimal among bounds depending only on the ray of the table. The chain majorizes every other supernatural decomposition, and ΨΨ is monotone under cancellations, hence upper semicontinuous in flat families. The comparison extends to coherent sheaves and, by linear projection, to sheaves of positive dimension. On P2\mathbb{P}^2 the defect is a sum of four cohomological terms; we characterize equality and show that the inequality survives the large volume limit of Bridgeland stability. For a smooth projective variety XX of dimension n≥2n\geq2 in characteristic zero, with HH ample and base-point-free, the mean normalized squared spread G(E)G(\mathcal{E}) of the roots in the cohomology table of every vector bundle E\mathcal{E} satisfies G(E)≥(KX2−2c2(X))⋅Hn−2/((n+1)Hn)G(\mathcal{E})\geq(K_X^2-2c_2(X))\cdot H^{n-2}/((n+1)H^n), a bound independent of E\mathcal{E}. Since G(E)=1G(\mathcal{E})=1 exactly for arithmetically Cohen-Macaulay bundles, when this bound exceeds 11 it is a uniform form of their absence.

Introduction

Throughout, KK is an algebraically closed field of characteristic zero; the results that hold over any algebraically closed field are listed at the end of this introduction. For a coherent sheaf FF on Pn\mathbb{P}^n let γ(F)=(hj(F(t)))0≤j≤n, t∈Z\gamma(F) = (h^j(F(t)))_{0\leq j\leq n,\,t\in\mathbb{Z}} be its cohomology table. Eisenbud and Schreyer [20] proved that the cohomology tables of vector bundles span, up to positive scalars, the 2020 Mathematics Subject Classification. Primary 14J60; Secondary 13D02, 14F06, 14F08.

Key words and phrases. Boij–Söderberg theory, cohomology tables, supernatural bundles, Harder–Narasimhan filtration, Bridgeland stability, Ulrich bundles. cone generated by the tables γz\gamma^z of supernatural bundles, and that every such table has a unique decomposition

γ(F)=∑kqkγzk,qk>0,\gamma(F)=\sum_k q_k\gamma^{z^k},\qquad q_k>0,

along a chain z1>z2>…z^1>z^2>\ldots of root sequences; see [21, 45, 25] for the theory and its extension to coherent sheaves. A root sequence is a strictly decreasing sequence of integers z=(z1>⋯>zn)z=(z_1>\cdots>z_n), and we normalize γz\gamma^z to have Hilbert polynomial ∏j(t−zj)\prod_j(t-z_j), i.e. rank n!n!. We attach to zz its center c(z)=1n∑jzjc(z)=\frac{1}{n}\sum_j z_j and its slope μ(z)=−c(z)−n+12\mu(z)=-c(z)-\frac{n+1}{2}; a supernatural bundle with root sequence zz has slope μ(z)\mu(z). Decreasing slopes correspond to increasing centers. Erman and Sam [23] single out the question whether these numerical decompositions, whose coefficients are rational in general, can be categorified at the level of vector bundles, and they answer it in many cases. Here we take a different route and link the Eisenbud–Schreyer chain decomposition to slope stability, through the following polygon; on P2\mathbb{P}^2 we determine exactly when the Harder–Narasimhan filtration realizes the chain decomposition (Corollary 5.21).

Definition 1.1. The Boij–Söderberg polygon ΨF:[0,rk⁡F]→R\Psi_F:[0,\operatorname{rk} F]\to\mathbb{R} of a vector bundle FF on Pn\mathbb{P}^n is the concave polygon whose edges have slopes μ(zk)\mu(z^k) and horizontal lengths n!qkn!q_k, arranged in decreasing order of slope.

Its endpoints are (0,0)(0,0) and (rk⁡F,deg⁡F)(\operatorname{rk} F,\deg F), the same as those of the Harder–Narasimhan polygon PFP_F, whose edges have the slopes of the Harder–Narasimhan factors and lengths their ranks. The two polygons record different kinds of information: PFP_F is defined by subsheaves, ΨF\Psi_F by cohomology. Our main result compares them.

Theorem A (Theorem 3.6). Let FF be a vector bundle on Pn\mathbb{P}^n, over an algebraically closed field of any characteristic. Then deg⁡F′≤ΨF(rk⁡F′)\deg F'\leq\Psi_F(\operatorname{rk}F') for every subsheaf F′⊂FF'\subset F; equivalently, PF≤ΨFP_F\leq\Psi_F. In particular, a vector bundle whose cohomology table is a multiple of a single supernatural table is semistable.

Theorem A gives an upper bound for the slopes of all subsheaves in terms of the table alone. It is optimal among bounds that depend only on the ray of the table: every table has a multiple that is the table of a direct sum of supernatural bundles with P=ΨP=\Psi. For a fixed table, integrality gives a sharper bound (Remark 3.9(1)). Moreover a saturated subsheaf of a supernatural bundle with the same slope, and its quotient, are again supernatural with the same roots (Proposition 3.8), so supernatural bundles are Gieseker semistable. Theorem A yields VHN(F)≤Vc(F)V_{HN}(F)\leq V_c(F) for the variances of the Harder–Narasimhan slopes and of the Boij–Söderberg slopes (Corollary 3.7); equivalently, for n≥2n\geq2, the gap moment G(F)G(F), the mean over the chain of a normalized spread of the roots (§ 3.4), satisfies G(F)≥1+12n+1∑iwiδ(Qi)G(F)\geq1+\frac{12}{n+1}\sum_i w_i\delta(Q_i), where QiQ_i are the Harder–Narasimhan factors, with weights wi=rk⁡Qi/rk⁡Fw_i=\operatorname{rk}Q_i/\operatorname{rk}F and normalized discriminants δ(Qi)≥0\delta(Q_i)\geq0 (Corollary 3.12). Always G(F)≥1G(F)\geq1, with equality exactly for split bundles. The proof rests on a formula for the concave conjugate

ΛF(λ)=∑kn!qk(μ(zk)−λ)+,ΨF(m)=min⁡λ(mλ+ΛF(λ)).\Lambda_F(\lambda)=\sum_k n!q_k\bigl(\mu(z^k)-\lambda\bigr)_+,\qquad\Psi_F(m)=\min_\lambda\bigl(m\lambda+\Lambda_F(\lambda)\bigr).

For a root sequence zz comparable with every element of the chain, ΛF(μ(z))\Lambda_F(\mu(z)) is the divided difference of truncated Euler characteristics

ϕz(γ(F))=(n−1)!∑r=1n∑j<r(−1)jhj(F(zr))∏i≠r(zr−zi).\phi_z(\gamma(F))=(n-1)!\sum_{r=1}^{n}\frac{\sum_{j<r}(-1)^j h^j(F(z_r))}{\prod_{i\ne r}(z_r-z_i)}.

By the theorem of Eisenbud, Fløystad, Weyman and Eisenbud, Schreyer on pure resolutions [19, 20], there is a sheaf T, set-theoretically supported on a general line LL (in general on a thickening of LL), whose minimal free resolution is pure, with twists z1,…,znz_1,\ldots,z_n. For a saturated subsheaf F′⊂FF' \subset F we may choose LL disjoint from the loci where F′F' and F/F′F/F' are not locally free; then F′⊗T⊂F⊗TF' \otimes T \subset F \otimes T, and

χ(F′⊗T)≤h0(F′⊗T)≤h0(F⊗T).\chi(F' \otimes T) \le h^0(F' \otimes T) \le h^0(F \otimes T).

Riemann–Roch turns the left-hand side into a positive multiple of deg⁡F′−μ(z)rk⁡F′\deg F' - \mu(z)\operatorname{rk} F', and the hypercohomology spectral sequence of FF tensored with the resolution bounds the right-hand side by the same multiple of ϕz(γ(F))\phi_z(\gamma(F)). This spectral sequence argument is the one by which Eisenbud and Schreyer prove the nonnegativity of their pairing between Betti tables and cohomology tables [20 Thm. 3.1], and the bound by h0(F⊗T)h^0(F \otimes T) is the sheaf form of an inequality in their proof; here the sheaf TT has one-dimensional support, and h0(F⊗T)h^0(F \otimes T) replaces the vanishing of F⊗TF \otimes T.

In the notation of [20], ϕz\phi_z is a positive multiple of the pairing with the Betti table of TT (see the paragraph after Lemma 3.5); for a categorified form of the Eisenbud–Schreyer functionals, as a single pairing between derived categories, see [17]. On P2\mathbb{P}^2 the test sheaf can be taken to be O(a−b)L(a)\mathcal{O}_{(a-b)L}(a), with resolution O(b)→O(a)\mathcal{O}(b) \to\mathcal{O}(a), for a line LL and a>ba>b, and the argument is elementary (§ 5).

Theorem A has consequences in three directions. On Pn\mathbb{P}^n it gives structural properties of the cone of cohomology tables: the chain decomposition is extremal among all decompositions into supernatural tables, and Ψ\Psi is monotone under the cancellations that occur in extensions and in flat families (Theorems B and C). On P2\mathbb{P}^2 the defect in Theorem A is an explicit sum of four cohomological terms, the equality PF=ΨFP_F=\Psi_F is characterized, and the comparison persists in the large volume limit of Bridgeland stability (Theorems D and E). Finally, a projection to Pn\mathbb{P}^n carries the comparison to sheaves of lower dimension on a larger projective space (Theorem F) and to vector bundles on a polarized variety (Theorem G). The last case was our starting point: after [1], Eisenbud and Schreyer asked about the Boij–Söderberg cone of cohomology tables of vector bundles on the surfaces constructed there (§ 8.2), and Theorem G gives a lower bound, the same for all vector bundles on (X,H)(X,H), for the gap moment of their tables.

Structural results on Pn\mathbb{P}^n. Let C\mathcal{C} be the set of nonnegative rational combinations of supernatural tables on Pn\mathbb{P}^n. Every γ∈C\gamma\in\mathcal{C} has a unique chain decomposition, and Ψγ\Psi_\gamma, Λγ\Lambda_\gamma are defined as above. Applied to supernatural bundles, the inequality behind Theorem A says that ϕz(γy)≥n!(c(z)−c(y))+\phi_z(\gamma^y) \ge n!(c(z)-c(y))_+ for all root sequences y,zy,z, with equality if yy and zz are comparable. So, for λ∈1nZ\lambda\in\frac{1}{n}\mathbb{Z}, Λγ(λ)\Lambda_\gamma(\lambda) is the minimum of the linear functionals ϕz\phi_z with μ(z)=λ\mu(z)=\lambda, attained at the root sequences comparable with the chain (Proposition 4.2). Two consequences follow.

Theorem B (Theorem 4.3). Let γ∈C\gamma\in\mathcal{C}, and let γ=∑jqj′γyj\gamma=\sum_j q'_j\gamma^{y_j} be any decomposition with qj′≥0q'_j \ge0 and arbitrary root sequences yjy_j. Then the distribution of centers of the chain decomposition, with masses qkq_k, majorizes the distribution of the centers c(yj)c(y_j), with masses qj′q'_j. Equivalently, Ψγ\Psi_\gamma is the largest slope polygon of any decomposition of γ\gamma, and for every convex function ff the map γ↦∑kqkf(c(z(k)))\gamma\mapsto\sum_k q_k f(c(z^{(k)})) is concave on C\mathcal{C}.

Theorem B makes precise the sense in which the chain decomposition is distinguished among all decompositions of γ\gamma into supernatural tables. That it is the natural one was suggested in [8], where the termwise order on root sequences is characterized by the existence of nonzero homomorphisms between supernatural sheaves. Equivalently, for m∈Qm \in\mathbb{Q}, Ψγ(m)\Psi_\gamma(m) is the largest degree of a summand of γ\gamma in C\mathcal{C} of rank mm (Proposition 4.5).

Let ei,je_{i,j} be the table with a single entry 11 in row jj and column ii. We call the tables ei,j+ei+1,je_{i,j}+e_{i+1,j}, 0≤j≤n0 \le j \le n, cancellation vectors. If 0→F′→F→F′′→00 \to F' \to F \to F'' \to0 is an exact sequence, the long exact cohomology sequences give

γ(F′)+γ(F′′)−γ(F)=∑t∑j=0n−1rk⁡(∂tj)(ej,t+ej+1,t),\gamma(F')+\gamma(F'')-\gamma(F)=\sum_t\sum_{j=0}^{n-1}\operatorname{rk}(\partial^j_t)(e_{j,t}+e_{j+1,t}),

References

  1. [1]C. Anghel, Smooth counterexamples to the Eisenbud–Schreyer–Weyman Ulrich existence problem, arXiv:2609.02718.
  2. [2]C. Anghel, Polarized varieties without arithmetically Cohen–Macaulay bundles and section rings without graded maximal Cohen–Macaulay modules in characteristic zero, arXiv:2609.15589.
  3. [3]C. Anghel, The recent counterexamples to the Eisenbud–Schreyer–Weyman Ulrich problem (2003) and to Hochster’s small Cohen–Macaulay conjecture (1973): a survey, with a generalization to arbitrary dimension, arXiv:2610.30027.
  4. [4]D. Arcara, A. Bertram, Bridgeland-stable moduli spaces for K-trivial surfaces, J. Eur. Math. Soc. 15 (2013), 1–38.arxiv.org/abs/0708.2247
  5. [5]D. Arcara, A. Bertram, I. Coskun, J. Huizenga, The minimal model program for the Hilbert scheme of points on P² and Bridgeland stability, Adv. Math. 235 (2013), 580–626.arxiv.org/abs/1203.0316
  6. [6]A. Bayer, Polynomial Bridgeland stability conditions and the large volume limit, Geom. Topol. 13 (2009), 2389–2425.
  7. [7]A. A. Beilinson, Coherent sheaves on Pⁿ and problems of linear algebra, Funct. Anal. Appl. 12 (1978), 214–216.DOI
  8. [8]C. Berkesch, D. Erman, M. Kummini, S. V. Sam, Poset structures in Boij–Söderberg theory, Int. Math. Res. Not. IMRN (2012), no. 22, 5132–5160.
  9. [9]F. A. Bogomolov, Holomorphic tensors and vector bundles on projective varieties, Izv. Akad. Nauk SSSR Ser. Mat. 42 (1978), 1227–1287.DOI
  10. [10]M. Boij, J. Söderberg, Betti numbers of graded modules and the multiplicity conjecture in the non-Cohen–Macaulay case, Algebra Number Theory 6 (2012), 437–454.arxiv.org/abs/0803.1645
  11. [11]T. Bridgeland, Stability conditions on triangulated categories, Ann. of Math. (2) 166 (2007), 317–345.
  12. [12]T. Bridgeland, Stability conditions on K3 surfaces, Duke Math. J. 141 (2008), 241–291.DOI
  13. [13]M. Casanellas, R. Hartshorne, Stable Ulrich bundles, with an appendix by F. Geiss and F.-O. Schreyer, Internat. J. Math. 23 (2012), 1250083.
  14. [14]I. Coskun, J. Huizenga, G. Smith, Stability and cohomology of kernel bundles on Pⁿ, Michigan Math. J. 75 (2025), 173–198.DOI
  15. [15]L. Costa, R. M. Miró-Roig, Instanton bundles vs Ulrich bundles on projective spaces, Beitr. Algebra Geom. 62 (2021), 429–439.DOI
  16. [17]D. Eisenbud, D. Erman, Categorified duality in Boij–Söderberg theory and invariants of free complexes, J. Eur. Math. Soc. 19 (2017), 2657–2695.DOI
  17. [18]D. Eisenbud, D. Erman, F.-O. Schreyer, Filtering free resolutions, Compositio Math. 149 (2013), 754–772.DOI
  18. [19]D. Eisenbud, G. Fløystad, J. Weyman, The existence of equivariant pure free resolutions, Ann. Inst. Fourier 61 (2011), 905–926.DOI
  19. [20]D. Eisenbud, F.-O. Schreyer, Betti numbers of graded modules and cohomology of vector bundles, J. Amer. Math. Soc. 22 (2009), 859–888.arxiv.org/abs/0712.1843
  20. [21]D. Eisenbud, F.-O. Schreyer, Cohomology of coherent sheaves and series of supernatural bundles, J. Eur. Math. Soc. 12 (2010), 703–722.arxiv.org/abs/0902.1594
  21. [22]D. Eisenbud, F.-O. Schreyer, Resultants and Chow forms via exterior syzygies, with an appendix by J. Weyman, J. Amer. Math. Soc. 16 (2003), 537–579.DOI
  22. [23]D. Erman, S. V. Sam, Supernatural analogues of Beilinson monads, Compositio Math. 152 (2016), 2545–2562.DOI
  23. [24]D. Erman, S. V. Sam, Questions about Boij–Söderberg theory, Proc. Sympos. Pure Math. 95, Amer. Math. Soc., 2017, 285–304; arXiv:1606.01867.
  24. [25]G. Fløystad, Boij–Söderberg theory: introduction and survey, in: Progress in Commutative Algebra 1, de Gruyter, Berlin, 2012, 1–54.
  25. [27]G. H. Hardy, J. E. Littlewood, G. Pólya, Inequalities, Cambridge University Press, Cambridge, 1934.
  26. [28]J. Herzog, M. Kühl, On the Betti numbers of finite pure and linear resolutions, Comm. Algebra 12 (1984), 1627–1646.
  27. [29]J. Herzog, H. Srinivasan, Bounds for multiplicities, Trans. Amer. Math. Soc. 350 (1998), 2879–2902.DOI
  28. [30]G. Horrocks, Vector bundles on the punctured spectrum of a local ring, Proc. London Math. Soc. (3) 14 (1964), 689–713.
  29. [31]D. Huybrechts, M. Lehn, The geometry of moduli spaces of sheaves, 2nd ed., Cambridge Math. Library, Cambridge Univ. Press, 2010.
  30. [32]S. B. Iyengar, L. Ma, M. E. Walker, Lim Ulrich sequences and Boij–Söderberg cones, Forum Math. Sigma 11 (2023), e115.
  31. [33]M. Jardim, Instanton sheaves on complex projective spaces, Collect. Math. 57 (2006), 69–91.
  32. [34]J. Karamata, Sur une inégalité relative aux fonctions convexes, Publ. Math. Univ. Belgrade 1 (1932), 145–148.
  33. [35]A. Langer, Semistable sheaves in positive characteristic, Ann. of Math. (2) 159 (2004), 251–276.DOI
  34. [36]J. Lo, Z. Qin, Mini-walls for Bridgeland stability conditions on the derived category of sheaves over surfaces, Asian J. Math. 18 (2014), no. 2, 321–344.arxiv.org/abs/1103.4352
  35. [37]A. F. Lopez, On the existence of smooth varieties of any dimension carrying no Ulrich bundles for some very ample line bundle, arXiv:2609.06678v3.arxiv.org/abs/2609.06678
  36. [38]E. Macrì, B. Schmidt, Lectures on Bridgeland stability, in: Moduli of Curves, Lect. Notes Unione Mat. Ital. 21, Springer, Cham, 2017, 139–211; arXiv:1607.01262.arxiv.org/abs/1607.01262
  37. [39]E. Miller, D. E. Speyer, A Kleiman–Bertini theorem for sheaf tensor products, J. Algebraic Geom. 17 (2008), 335–340.arxiv.org/abs/math/0601202
  38. [40]D. Mumford, Abelian varieties, Tata Inst. Fund. Res. Stud. Math. 5, Oxford University Press, London, 1970.
  39. [41]C. Okonek, M. Schneider, H. Spindler, Vector bundles on complex projective spaces, Progress in Mathematics 3, Birkhäuser, Boston, 1980.
  40. [42]J. Pine, Existence of Ulrich bundles on polarized surfaces and weakly Ulrich bundles on polarized varieties in positive characteristic, arXiv:2610.28456v3.arxiv.org/abs/2610.28456
  41. [43]J. Pine, ACM, almost Ulrich, and Ulrich vector bundles arising from Frobenius, arXiv:2610.28464v1.arxiv.org/abs/2610.28464
  42. [45]F.-O. Schreyer, D. Eisenbud, Betti numbers of syzygies and cohomology of coherent sheaves, Proc. ICM Hyderabad 2010, Vol. II, 586–602.arxiv.org/abs/1102.3559
  43. [46]S. S. Shatz, The decomposition and specialization of algebraic families of vector bundles, Compositio Math. 35 (1977), 163–187.

Paper details

vlnc:W_kcetphpd
Version
v2 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents