Deformations of the tangent bundle of a projective hypersurface
Insong Choe, Kiryong Chung, Jun-Muk Hwang · 2025
Read Deformations of the tangent bundle of a projective hypersurface by Insong Choe, Kiryong Chung, Jun-Muk Hwang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-20085.txt.
Introduction
Insong Choe, Kiryong Chung, and Jun-Muk Hwang study how a key bundle on a hypersurface can deform in this two thousand twenty-five math paper. A hypersurface is a shape cut out by one equation in projective space. Its tangent bundle captures how the shape bends at each point. The authors compute the space of tiny deformations of that bundle for smooth hypersurfaces of dimension at least three.
They give the space's exact dimension as a clean formula. Every tiny deformation extends to a real one, even when an obstruction space is present. They also show the surrounding moduli space of stable bundles is a rational variety, by building an explicit model of it. Algebraic geometers are the audience.
Catalog record
| CREATOR | Insong Choe, Kiryong Chung, Jun-Muk Hwang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-20085
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Deformations of the tangent bundle of a projective hypersurface Thanks: Jun-Muk Hwang was supported by the Institute for Basic Science (IBS-R032-D1). Insong Choe, Kiryong Chung, Jun-Muk Hwang Abstract . For a nonsingular hypersurface X ⊂ ℙ n , n ≥ 4 , X\subset{\mathbb{P}}^{n},n\geq 4, of degree d ≥ 2 d\geq 2 , we show that the space H 1 ( X , End ( T X ) ) H^{1}(X,\mathrm{End}\,(T_{X})) of infinitesimal deformations of the tangent bundle T X T_{X} has dimension ( n + d − 1 d ) ( d − 1 ) {n+d-1\choose d}(d-1) and all infinitesimal deformations are unobstructed even though H 2 ( X , End ( T X ) ) H^{2}(X,\mathrm{End}\,(T_{X})) can be nonzero. Furthermore, we prove that the irreducible component of the moduli space of stable bundles containing the tangent bundle is a rational variety, by constructing an explicit birational model. MSC2020: 14J60, 14J70, 14D20 Key words: deformation, tangent bundle, projective hypersurface 1. Introduction We work over complex numbers. For a vector bundle E on a variety X , we use the notation h^{i}(X,E):=\dim H^{i}(X,E). Let X\subset{\mathbb{P}}^{n},n\geq 4, be a nonsingular hypersurface of degree d\geq 2. Let T_{X} be its tangent bundle and denote by \mathrm{End}\,(T_{X}) (resp. \mathrm{End}\,_{0}(T_{X}) ) the bundle of endomorphisms (resp. traceless endomorphisms) of T_{X} . Then H^{1}(X,\mathrm{End}\,(T_{X}))=H^{1}(X,\mathrm{End}\,_{0}(T_{X})) (from H^{1}(X,{\mathcal{O}}_{X})=0 ) describes the infinitesimal deformations of T_{X} as vector bundles. In [ P , Theorem 6.1] , Peternell proved that H^{1}(X,\mathrm{End}\,_{0}(T_{X}))\neq 0 , namely, the tangent bundle has nontrivial infinitesimal deformations. This naturally leads to the following questions. Q1 What is the exact value of h^{1}(X,\mathrm{End}\,(T_{X})) ? Q2 Are all elements of H^{1}(X,\mathrm{End}\,(T_{X})) unobstructed? Equivalently, is the deformation space of T_{X} smooth at the point corresponding to T_{X} ? Furthermore, the tangent bundle T_{X} is a stable bundle with respect to the polarization \mathcal{O}(1) on {\mathbb{P}}^{n} by [ PW , Corollary 0.3] . Thus T_{X} represents a point [T_{X}] in the moduli scheme of stable bundles on X . Let {\mathcal{M}} be the irreducible component of the moduli scheme containing [T_{X}] . Q3 What is the geometric nature of the algebraic variety {\mathcal{M}} ? Our goal is to study these three questions. The answer to Q1 is the following, to be proved in Section 2 . Theorem 1 . Let X\subset{\mathbb{P}}^{n},n\geq 4, be a nonsingular projective hypersurface of degree d\geq 2. Then h^{1}(X,{\rm End}_{0}(T_{X}))={n+d-1\choose d}(d-1) . The obstruction to realizing an infinitesimal deformation in H^{1}(X,{\rm End}(T_{X})) as an actual deformation of T_{X} lies in H^{2}(X,{\rm End}(T_{X}))=H^{2}(X,T_{X}\otimes\Omega_{X}). Thus the answer to Q2 would have been straightforward if H^{2}(X,T_{X}\otimes\Omega_{X})=0. However, this group does not always vanish as the following theorem says, which is proved in Section 3 . Theorem 2 . Let X\subset{\mathbb{P}}^{n},n\geq 4, be a nonsingular projective hypersurface of degree d\geq 2. (i) H^{2}(X,T_{X}\otimes\Omega_{X})=0 if n\geq 6 . (ii) When n=4 , h^{2}(X,T_{X}\otimes\Omega_{X})=\begin{cases}0,&\text{if}\ d=2,3\\ 45,&\text{if}\ d=4\\ 224,&\text{if}\ d=5.\end{cases} Consequently, Q2 is a subtler question for some values of n and d . Nevertheless, we answer Q2 affirmatively in Section 4 as follows for all n\geq 4 and d\geq 2 by an explicit construction of deformations of T_{X} , which also gives a partial answer to Q3 . Theorem 3 . Let X\subset{\mathbb{P}}^{n},n\geq 4, be a nonsingular projective hypersurface of degree d\geq 2. Then there is an affine space {\mathscr{A}} with a base point o\in{\mathscr{A}} and a torsionfree sheaf \mathbb{E}\to X\times{\mathscr{A}} which satisfy the following properties. (i) \dim{\mathscr{A}}={n+d-1\choose d}(d-1) . (ii) The restriction \mathbb{E}|_{X\times\{o\}} is isomorphic to the tangent bundle T_{X} . (iii) There is a Zarisky open subset {\mathscr{A}}^{\circ}\subset{\mathscr{A}} with o\in{\mathscr{A}}^{\circ} such that for any \alpha\in{\mathscr{A}}^{\circ} , the restriction \mathbb{E}|_{X\times\{\alpha\}} is a vector bundle of determinant \wedge^{n-1}(\mathbb{E}|_{X\times\{\alpha\}})\cong K^{-1}_{X} . (iv) There is a Zariski open subset {\mathscr{A}}^{\rm faithful}\subset{\mathscr{A}}^{\circ} with o\in{\mathscr{A}}^{\rm faithful} such that for any \alpha,\alpha^{\prime}\in{\mathscr{A}}^{\rm faithful} , the two vector bundles \mathbb{E}|_{X\times\{\alpha\}} and \mathbb{E}|_{X\times\{\alpha^{\prime}\}} are isomorphic only when \alpha=\alpha^{\prime} . (v) There is a Zariski open subset {\mathscr{A}}^{\rm stable}\subset{\mathscr{A}}^{\circ} with o\in{\mathscr{A}}^{\rm stable} such that for any \alpha\in{\mathscr{A}}^{\rm stable} , the restriction \mathbb{E}|_{X\times\{\alpha\}} is a stable vector bundle. Consequently, the moduli space {\mathcal{M}} is nonsingular at [T_{X}] and it is a rational variety. What happens when n=3 and X is a surface? A construction analogous to {\mathscr{A}} and \mathbb{E} of Theorem 3 works even when n=3 . However, the statement of Theorem 1 does not hold when n=3 . In fact, by a general theory, the moduli space {\mathcal{M}} at the point [T_{X}] has dimension at least h^{1}(X,T_{X}\otimes\Omega_{X})-h^{2}(X,T_{X}\otimes\Omega_{X}) . One can check (see Proposition 11 ) that when n=3 and d\geq 3 , h^{1}(X,T_{X}\otimes\Omega_{X})-h^{2}(X,T_{X}\otimes\Omega_{X})=\frac{1}{3}(d-1)(7d^{2}-5d-3), which is strictly bigger than \dim{\mathscr{A}}=\frac{1}{2}(d-1)(d^{2}+3d+2) . Thus one needs new ideas to study Q1 , Q2 and Q3 for surfaces in {\mathbb{P}}^{3} . We would like to comment on the cases missing in Theorem 2 . When n=4 , we expect for all d\geq 4, (1.1) h^{2}(X,T_{X}\otimes\Omega_{X})=(11d+1){d-1\choose 3}. This was confirmed for the Fermat hypersurfaces for 4\leq d\leq 25 by using the Macaulay2 program( [ GS ] ). Also this agrees with Theorem 2 (ii) for d=4,5 . We have not been able to verify this for d\geq 6 . When n=5 , similarly we checked for Fermat hyperfaces of degree d that h^{2}(X,T_{X}\otimes\Omega_{X})=1 for d=3 and h^{2}(X,T_{X}\otimes\Omega_{X})=0 for 4\leq d\leq 25 . To close the introduction, let us mention that the third-named author erroneously asserted in [ Hw , page 77] that the tangent bundle of an irreducible Hermitian symmetric space cannot be deformed. This contradicts Theorem 3 for the quadric hypersurface, which is an irreducible Hermitian symmetric space of type IV. The flaw was in the limiting argument in the proof of [ Hw , Proposition 2.1] . Consequently, the main result Theorem 1.2 of [ Hw ] is incorrect. It is expected that the deformation space of the tangent bundle of an irreducible Hermitian symmetric space, other than projective space, has positive dimension (see [ P , Question 1.6] ). It would be interesting to find an explicit birational model of the deformation space, generalizing Theorem 3 . 2. Dimension of H^{1}(X,T_{X}\otimes\Omega_{X}) In this section, we present a number of cohomological computations associated to the tangent bundle T_{X} of a smooth projective hypersurface X\subset{\mathbb{P}}^{n} , including a proof of Theorem 1 . First we recall the following result, Bott’s formula (for example, [ OSS , page 8] ). Proposition 4 . h^{i}({\mathbb{P}}^{n},\Omega^{j}_{{\mathbb{P}}^{n}}(k))=\begin{cases}{n+k-j\choose k}{k-1\choose j},&\text{if}\ i=0,0\leq j\leq n,k>j\\ {j-k\choose-k}{-k-1\choose n-j},&\text{if}\ i=n,0\leq j\leq n,k<j-n\\ 1,&\text{if}\ k=0,i=j\\ 0,&\text{otherwise}.\end{cases} From this, we can show: Lemma 5 . For 0\leq i\leq n-2 and d\geq 2 , we have \dim H^{i}({\mathbb{P}}^{n},\mathrm{End}\,T_{{\mathbb{P}}^{n}}(-d))=0 . Proof. This follows from Proposition 4 combined with the Euler sequence tensored by \Omega_{{\mathbb{P}}^{n}}(-d) : 0\to\Omega_{{\mathbb{P}}^{n}}(-d)\to\Omega_{{\mathbb{P}}^{n}}(1-d)^{\oplus(n+1)}\to\mathrm{End}\,T_{{\mathbb{P}}^{n}}(-d)\to 0. ∎ Recall the following from [ F , Satz 8.11] . Proposition 6 . For a nonsingular projective hypersurface X\subset{\mathbb{P}}^{n} and each 0<i<n-1 , we have \begin{cases}H^{i}(X,\Omega_{X}^{j}(k))=0,&\text{if}\ i+j\neq n-1,k\neq 0\\ H^{i}(X,\Omega_{X}^{j})=0,&\text{if}\ i+j\neq n-1,i\neq j\\ H^{i}(X,\Omega_{X}^{i})={\mathbb{C}},&\text{if}\ 2i\neq n-1.\end{cases} Lemma 7 . For a nonsingular projective hypersurface X\subset{\mathbb{P}}^{n},n\geq 4 of degree d\geq 2 , the following holds. (i) h^{0}(X,\Omega_{X}(d))={n+d-1\choose d}(d-1) . (ii) h^{i}(X,\Omega_{X}(k))=0 for 0<i<n-2 and k\neq 0 . (iii) h^{i}(X,T_{{\mathbb{P}}^{n}}(-d)|_{X})=0 for i=0,1,2 . (iv) h^{1}(X,T_{X}(-d))=1 . Proof. From the vanishing h^{1}(X,{\mathcal{O}}_{X})=0 and the cotangent sequence 0\to{\mathcal{O}}_{X}\to\Omega_{{\mathbb{P}}^{n}}|_{X}(d)\to\Omega_{X}(d)\to 0, we have h^{0}(X,\Omega_{X}(d))=h^{0}(X,\Omega_{{\mathbb{P}}^{n}}|_{X}(d))-1. Combining Proposition 4 and the following sequence