Calculations in Deformation Theory Nathan Owen Ilten UC Berkeley February 22nd, 2012 Deformation Basics Computing Versal Deformations Deformation Basics Computing Versal Deformations The Setup The Setup I Let S = C[x1 , . . . , xd ]. The Setup I Let S = C[x1 , . . . , xd ]. I Consider polynomials f1 , f2 , . . . , fm ∈ S. The Setup I Let S = C[x1 , . . . , xd ]. I Consider polynomials f1 , f2 , . . . , fm ∈ S. I These generate an ideal I in S. The Setup I Let S = C[x1 , . . . , xd ]. I Consider polynomials f1 , f2 , . . . , fm ∈ S. I These generate an ideal I in S. I They also define an affine scheme X ⊂ Cd . The Setup I Let S = C[x1 , . . . , xd ]. I Consider polynomials f1 , f2 , . . . , fm ∈ S. I These generate an ideal I in S. I They also define an affine scheme X ⊂ Cd . Basic idea: we deform X by perturbing the fi . Example I: An A1 Singularity Example I: An A1 Singularity I Take d = 3, f = x 2 + y 2 − z 2 . Example I: An A1 Singularity I Take d = 3, f = x 2 + y 2 − z 2 . Example I: An A1 Singularity I Take d = 3, f = x 2 + y 2 − z 2 . I Perturb this with a parameter t to get e f = x 2 + y 2 − z 2 − t. Example I: An A1 Singularity I Take d = 3, f = x 2 + y 2 − z 2 . I Perturb this with a parameter t to get e f = x 2 + y 2 − z 2 − t. I This cuts out a scheme X ⊂ C3 × C with a natural projection map π : X → C. Example I: An A1 Singularity I Take d = 3, f = x 2 + y 2 − z 2 . I Perturb this with a parameter t to get e f = x 2 + y 2 − z 2 − t. I This cuts out a scheme X ⊂ C3 × C with a natural projection map π : X → C. I The fiber over 0 is just X . The fiber over t 6= 0 is smooth. Example II: A Line in C3 Example II: A Line in C3 I Take d = 3, f1 = x − y , f2 = x − z, f3 = y − z. Example II: A Line in C3 I Take d = 3, f1 = x − y , f2 = x − z, f3 = y − z. I These equations cut out a line x = y = z in C3 . Example II: A Line in C3 I Take d = 3, f1 = x − y , f2 = x − z, f3 = y − z. I These equations cut out a line x = y = z in C3 . I Now perturb: fe1 = x − y + t fe2 = x − z + t fe3 = y − z + t Example II: A Line in C3 I Take d = 3, f1 = x − y , f2 = x − z, f3 = y − z. I These equations cut out a line x = y = z in C3 . I Now perturb: fe1 = x − y + t fe2 = x − z + t fe3 = y − z + t I Something is fishy! The fiber over t 6= 0 is a point! Example II: A Line in C3 I Take d = 3, f1 = x − y , f2 = x − z, f3 = y − z. I These equations cut out a line x = y = z in C3 . I Now perturb: fe1 = x − y + t fe2 = x − z + t fe3 = y − z + t I Something is fishy! The fiber over t 6= 0 is a point! Problem: the relation f2 − f1 = f3 doesn’t lift to a relation among the fei . Lifting Relations Lifting Relations Consider the start of a free resolution of S/I : R F · · · −−−−→ S n −−−−→ S m −−−−→ S −−−−→ S/I −−−−→ 0. Here, F is a matrix whose columns are just the fi . Lifting Relations Consider the start of a free resolution of S/I : R F · · · −−−−→ S n −−−−→ S m −−−−→ S −−−−→ S/I −−−−→ 0. Here, F is a matrix whose columns are just the fi . Consider a ring of deformation parameters T = C[t1 , . . . , te ] and e = S ⊗ T. set S Lifting Relations Consider the start of a free resolution of S/I : R F · · · −−−−→ S n −−−−→ S m −−−−→ S −−−−→ S/I −−−−→ 0. Here, F is a matrix whose columns are just the fi . Consider a ring of deformation parameters T = C[t1 , . . . , te ] and e = S ⊗ T. set S em → S e be a perturbation of F . Let Fe : S Lifting Relations Consider the start of a free resolution of S/I : R F · · · −−−−→ S n −−−−→ S m −−−−→ S −−−−→ S/I −−−−→ 0. Here, F is a matrix whose columns are just the fi . Consider a ring of deformation parameters T = C[t1 , . . . , te ] and e = S ⊗ T. set S em → S e be a perturbation of F . Let Fe : S Definition The relations R lift with respect to Fe subject to equations e :S en → S e m restricting to R such g1 , . . . , gk ∈ T if there exists a R that e ⊂ hgi i Im Fe · R . Example III: Lifting Relations Example III: Lifting Relations I Take d = 4 and consider the matrix F = x1 x3 − x22 x2 x4 − x32 x1 x4 − x2 x3 . Example III: Lifting Relations I Take d = 4 and consider the matrix F = I x1 x3 − x22 x2 x4 − x32 A relation matrix is given by x4 x3 x1 . R = x2 −x3 −x2 x1 x4 − x2 x3 . Example III: Lifting Relations I Take d = 4 and consider the matrix Fe = I x1 x3 − x22 + tx2 x2 x4 − x32 − tx4 x1 x4 − x2 x3 A relation matrix is given by x4 x3 x1 . R = x2 −x3 −x2 I We can perturb F to Fe . . Example III: Lifting Relations I Take d = 4 and consider the matrix Fe = x1 x3 − x22 + tx2 x2 x4 − x32 − tx4 x1 x4 − x2 x3 I A relation matrix is given by x4 x3 e = x2 . x1 R −x3 −x2 + t I We can perturb F to Fe . I e A lifting of R is given by R. . Deformations of Affine Schemes Deformations of Affine Schemes Definition A deformation of X ⊂ Cd over Z = Spec (T /hgi i) consists of a perturbation Fe of F such that the relations lift with respect to Fe , subject to the equations gi . Deformations of Affine Schemes Definition A deformation of X ⊂ Cd over Z = Spec (T /hgi i) consists of a perturbation Fe of F such that the relations lift with respect to Fe , subject to the equations gi . I Fe defines a scheme X ⊂ Cd × Z and a map π : X → Z . Deformations of Affine Schemes Definition A deformation of X ⊂ Cd over Z = Spec (T /hgi i) consists of a perturbation Fe of F such that the relations lift with respect to Fe , subject to the equations gi . I Fe defines a scheme X ⊂ Cd × Z and a map π : X → Z . I X is the total space, Z the base space of the deformation. Deformations of Affine Schemes Definition A deformation of X ⊂ Cd over Z = Spec (T /hgi i) consists of a perturbation Fe of F such that the relations lift with respect to Fe , subject to the equations gi . I Fe defines a scheme X ⊂ Cd × Z and a map π : X → Z . I X is the total space, Z the base space of the deformation. Example For hypersurfaces, arbitrary perturbations are allowed. Induced Deformations Induced Deformations I Consider a deformation of X . Induced Deformations I Consider a deformation of X . I We can induce other deformations of X by applying changes of coordinates to the variables xi and substituting in new deformation parameters for the ti . Induced Deformations I Consider a deformation of X . I We can induce other deformations of X by applying changes of coordinates to the variables xi and substituting in new deformation parameters for the ti . Example Induced Deformations I Consider a deformation of X . I We can induce other deformations of X by applying changes of coordinates to the variables xi and substituting in new deformation parameters for the ti . Example I Consider the deformation given by e f = x 2 + y 2 − z 2 − t. Induced Deformations I Consider a deformation of X . I We can induce other deformations of X by applying changes of coordinates to the variables xi and substituting in new deformation parameters for the ti . Example I Consider the deformation given by e f = x 2 + y 2 − z 2 − t. I Can we induce the deformation given by x 2 + y 2 − z 2 − sz? Induced Deformations I Consider a deformation of X . I We can induce other deformations of X by applying changes of coordinates to the variables xi and substituting in new deformation parameters for the ti . Example I Consider the deformation given by e f = x 2 + y 2 − z 2 − t. I Can we induce the deformation given by x 2 + y 2 − z 2 − sz? I Yes! Substitute t = − 14 s 2 and take the change of coordinates z 7→ (z + 21 s). Versal Deformations Versal Deformations Definition A deformation of X is called (formally) versal if any (infinitesimal) deformation of X may be induced from it. Versal Deformations Definition A deformation of X is called (formally) versal if any (infinitesimal) deformation of X may be induced from it. I e to contain formal If dim Sing X = 0 and we allow Fe and R power series, then X has a formally versal deformation. Versal Deformations Definition A deformation of X is called (formally) versal if any (infinitesimal) deformation of X may be induced from it. I e to contain formal If dim Sing X = 0 and we allow Fe and R power series, then X has a formally versal deformation. Q: How can we compute a versal deformation of X ? Versal Deformations Definition A deformation of X is called (formally) versal if any (infinitesimal) deformation of X may be induced from it. I e to contain formal If dim Sing X = 0 and we allow Fe and R power series, then X has a formally versal deformation. Q: How can we compute a versal deformation of X ? A: Using Macaulay2 and the package VersalDeformations. Deformation Basics Computing Versal Deformations Basic Features of the Package VersalDeformations Basic Features of the Package VersalDeformations Input: Basic Features of the Package VersalDeformations Input: A matrix F containing the equations of X . Basic Features of the Package VersalDeformations Input: A matrix F containing the equations of X . Output: Basic Features of the Package VersalDeformations Input: A matrix F containing the equations of X . Output: I A basis for TX1 , the space of deformations over Spec C[t]/t 2 . Basic Features of the Package VersalDeformations Input: A matrix F containing the equations of X . Output: I A basis for TX1 , the space of deformations over Spec C[t]/t 2 . I A basis for TX2 , which contains obstructions to lifting deformations. Basic Features of the Package VersalDeformations Input: A matrix F containing the equations of X . Output: I A basis for TX1 , the space of deformations over Spec C[t]/t 2 . I A basis for TX2 , which contains obstructions to lifting deformations. I A formally versal deformation of X (more details later). Basic Features of the Package VersalDeformations Input: A matrix F containing the equations of X . Output: I A basis for TX1 , the space of deformations over Spec C[t]/t 2 . I A basis for TX2 , which contains obstructions to lifting deformations. I A formally versal deformation of X (more details later). Basic approach: iteratively lift deformations in TX1 to larger and larger base spaces. Computational Example I: Our A1 Singularity Computational Example I: Our A1 Singularity I F = ( x 2 + y 2 − z 2 ). Computational Example I: Our A1 Singularity I F = ( x 2 + y 2 − z 2 ). I Any first order deformation can be induced from F + t1 · 1 = ( x 2 + y 2 − z 2 + t1 ) with t12 = 0. Output of Command “versalDeformation” Output of Command “versalDeformation” The command “versalDeformation” outputs four lists FL, RL, GL, and CL where Output of Command “versalDeformation” The command “versalDeformation” outputs four lists FL, RL, GL, and CL where I FL is a list of matrices whose sum is the perturbation matrix Fe of a versal deformation. Output of Command “versalDeformation” The command “versalDeformation” outputs four lists FL, RL, GL, and CL where I I FL is a list of matrices whose sum is the perturbation matrix Fe of a versal deformation. e of R. RL is a list of matrices whose sum is a lifting R Output of Command “versalDeformation” The command “versalDeformation” outputs four lists FL, RL, GL, and CL where I I I FL is a list of matrices whose sum is the perturbation matrix Fe of a versal deformation. e of R. RL is a list of matrices whose sum is a lifting R GL is a list of matrices whose sum G contains the equations cutting out the versal base space. Output of Command “versalDeformation” The command “versalDeformation” outputs four lists FL, RL, GL, and CL where I I I FL is a list of matrices whose sum is the perturbation matrix Fe of a versal deformation. e of R. RL is a list of matrices whose sum is a lifting R GL is a list of matrices whose sum G contains the equations cutting out the versal base space. These matrices solve the “deformation equation” e tr + C · G = 0 (Fe · R) where C is the sum of the list CL. Computational Example II: The Cone over the Rational Normal Curve of Deg. 3 Computational Example II: The Cone over the Rational Normal Curve of Deg. 3 Take F = x1 x3 − x22 x2 x4 − x32 x1 x4 − x2 x3 . Computational Example II: The Cone over the Rational Normal Curve of Deg. 4 Computational Example II: The Cone over the Rational Normal Curve of Deg. 4 Take F to be the transpose of x1 x3 − x22 x2 x4 − x32 x3 x5 − x42 x1 x4 − x2 x3 x2 x5 − x3 x4 x1 x5 − x2 x5 . Computational Example II: The Cone over the Rational Normal Curve of Deg. 4 Take F to be the transpose of I x1 x3 − x22 x2 x4 − x32 x3 x5 − x42 x1 x4 − x2 x3 x2 x5 − x3 x4 x1 x5 − x2 x5 . The base space has two components, C3 and C meeting in a point. Total Spaces Over the Components Total Spaces Over the Components The components come from two ways of writing the equations of X: Total Spaces Over the Components The components come from two ways of writing the equations of X: x1 x2 x3 x4 rk ≤1 x2 x3 x4 x5 Total Spaces Over the Components The components come from two ways of writing the equations of X: x1 x2 x3 x4 rk ≤1 x2 x3 x4 x5 x1 x2 x3 rk x2 x3 x4 ≤ 1 x3 x4 x5 Total Spaces Over the Components The components come from two ways of writing the equations of X: x1 x2 x3 x4 rk ≤1 x2 + s1 x3 + s2 x4 + s3 x5 x1 x2 x3 rk x2 x3 x4 ≤ 1 x3 x4 x5 Total Spaces Over the Components The components come from two ways of writing the equations of X: x1 x2 x3 x4 rk ≤1 x2 + s1 x3 + s2 x4 + s3 x5 x1 x2 x3 rk x2 x3 + s4 x4 ≤ 1 x3 x4 x5 Further Features of the Package VersalDeformations Further Features of the Package VersalDeformations I Can calculate TX1 , TX2 , and normal modules for projective X in good situations. Further Features of the Package VersalDeformations I Can calculate TX1 , TX2 , and normal modules for projective X in good situations. I Can calculate versal deformations for projective X . Further Features of the Package VersalDeformations I Can calculate TX1 , TX2 , and normal modules for projective X in good situations. I Can calculate versal deformations for projective X . I Can calculate local (multigraded) Hilbert schemes. Further Features of the Package VersalDeformations I Can calculate TX1 , TX2 , and normal modules for projective X in good situations. I Can calculate versal deformations for projective X . I Can calculate local (multigraded) Hilbert schemes. I Can lift deformations in given tangent direction to a one-parameter family. A Toric Fano Threefold A Toric Fano Threefold Let X be the projective subscheme of P8 cut out by xi+1 xi−1 − xi y0 1≤i ≤6 y02 y02 1≤i ≤3 xi xi+3 − y1 y2 − where P8 has coordinates x1 , . . . , x6 , y0 , y1 , y2 . A Toric Fano Threefold Let X be the projective subscheme of P8 cut out by xi+1 xi−1 − xi y0 1≤i ≤6 y02 y02 1≤i ≤3 xi xi+3 − y1 y2 − where P8 has coordinates x1 , . . . , x6 , y0 , y1 , y2 . A Toric Fano Threefold (cont.) A Toric Fano Threefold (cont.) The versal base space of X has three components. A Toric Fano Threefold (cont.) The versal base space of X has three components. X admits smoothings to three different kinds of Fano threefolds. A Toric Fano Threefold (cont.) The versal base space of X has three components. X admits smoothings to three different kinds of Fano threefolds. Similar calculations + lots of hard work can be used to classify all smoothings of Gorenstein Fano toric threefolds of degree ≤ 12. References Jan Arthur Christophersen and Nathan Owen Ilten. Toric degenerations of low degree Fano threefolds. arXiv:1202.0510v1 [math.AG], 2012. Nathan Owen Ilten. VersalDeformations — a package for computing versal deformations and local Hilbert schemes. arXiv:1107.2416v1 [math.AG], 2011. Jan Stevens. Computing versal deformations. Experiment. Math., 4(2):129–144, 1995. First Order Deformations First Order Deformations Definition Let TX1 be the set of isomorphism classes of deformations of X with base space Spec C[t]/t 2 . First Order Deformations Definition Let TX1 be the set of isomorphism classes of deformations of X with base space Spec C[t]/t 2 . TX1 may be computed as the cokernel of J : S d → HomS (I , S/I ) ⊂ (S/I )m ∂fi where J is the Jacobian matrix ∂x . j ij First Order Deformations Definition Let TX1 be the set of isomorphism classes of deformations of X with base space Spec C[t]/t 2 . TX1 may be computed as the cokernel of J : S d → HomS (I , S/I ) ⊂ (S/I )m ∂fi where J is the Jacobian matrix ∂x . j ij If dim Sing(X ) = 0, then dimC TX1 < ∞. First Order Deformations (cont.) First Order Deformations (cont.) I Choose φi ∈ Hom(S m , S) i = 1, . . . , e which represent a basis of TX1 . First Order Deformations (cont.) I Choose φi ∈ Hom(S m , S) i = 1, . . . , e which represent a basis of TX1 . I Set T = C[t1 , . . . , te ] with maximal ideal m = ht1 , . . . , te i. First Order Deformations (cont.) I Choose φi ∈ Hom(S m , S) i = 1, . . . , e which represent a basis of TX1 . I Set T = C[t1 , . . . , te ] with maximal ideal m = ht1 , . . . , te i. em → S e be the perturbation of F = F 0 defined by Let F 1 : S I 1 0 F =F + e X i=1 t i φi . First Order Deformations (cont.) I Choose φi ∈ Hom(S m , S) i = 1, . . . , e which represent a basis of TX1 . I Set T = C[t1 , . . . , te ] with maximal ideal m = ht1 , . . . , te i. em → S e be the perturbation of F = F 0 defined by Let F 1 : S I 1 0 F =F + e X t i φi . i=1 I The relations R lift with respect to F 1 subject to m2 to some en → S em. R1 : S First Order Deformations (cont.) I Choose φi ∈ Hom(S m , S) i = 1, . . . , e which represent a basis of TX1 . I Set T = C[t1 , . . . , te ] with maximal ideal m = ht1 , . . . , te i. em → S e be the perturbation of F = F 0 defined by Let F 1 : S I 1 0 F =F + e X t i φi . i=1 I The relations R lift with respect to F 1 subject to m2 to some en → S e m . This can be computed using matrix quotients R1 : S in Macaulay2. The Deformation Equation, Part I The Deformation Equation, Part I I Goal: lift this deformation to a “larger” base space. The Deformation Equation, Part I I I Goal: lift this deformation to a “larger” base space. e m , S), e R i−1 ∈ Hom(S en, S e m ), we would Given F i−1 ∈ Hom(S i i like to find F and R such that The Deformation Equation, Part I I I Goal: lift this deformation to a “larger” base space. e m , S), e R i−1 ∈ Hom(S en, S e m ), we would Given F i−1 ∈ Hom(S i i like to find F and R such that 1. F i ≡ F i−1 mod mi , R i ≡ R i−1 mod mi ; The Deformation Equation, Part I I I Goal: lift this deformation to a “larger” base space. e m , S), e R i−1 ∈ Hom(S en, S e m ), we would Given F i−1 ∈ Hom(S i i like to find F and R such that 1. F i ≡ F i−1 mod mi , R i ≡ R i−1 2. F i · R i ≡ 0 mod mi+1 . mod mi ; The Deformation Equation, Part I I I Goal: lift this deformation to a “larger” base space. e m , S), e R i−1 ∈ Hom(S en, S e m ), we would Given F i−1 ∈ Hom(S i i like to find F and R such that 1. F i ≡ F i−1 mod mi , R i ≡ R i−1 2. F i · R i ≡ 0 mod mi+1 . In general, this is not possible!!! mod mi ; The Deformation Equation, Part II The Deformation Equation, Part II I If dim Sing(X ) = 0, there is a finite dimensional C-vector space TX2 containing obstructions to lifting F i , R i . The Deformation Equation, Part II I If dim Sing(X ) = 0, there is a finite dimensional C-vector space TX2 containing obstructions to lifting F i , R i . I Choose V ∈ Hom(S l , S n ) such that its columns represent a basis of TX2 . The Deformation Equation, Part II I If dim Sing(X ) = 0, there is a finite dimensional C-vector space TX2 containing obstructions to lifting F i , R i . I Choose V ∈ Hom(S l , S n ) such that its columns represent a basis of TX2 . e It is possible to inductively construct F i , R i , G i−2 ∈ Hom(S, l i−2 l n e e e S ), C ∈ Hom(S , S ) solving I (F i R i )tr + C i−2 G i−2 ≡ 0 such that: mod mi+1 . The Deformation Equation, Part II I If dim Sing(X ) = 0, there is a finite dimensional C-vector space TX2 containing obstructions to lifting F i , R i . I Choose V ∈ Hom(S l , S n ) such that its columns represent a basis of TX2 . e It is possible to inductively construct F i , R i , G i−2 ∈ Hom(S, l i−2 l n e e e S ), C ∈ Hom(S , S ) solving I (F i R i )tr + C i−2 G i−2 ≡ 0 mod mi+1 . such that: 1. F i , R i , G i−2 , C i−2 reduce to F i−1 , R i−1 , G i−3 , C i−3 modulo mi ; The Deformation Equation, Part II I If dim Sing(X ) = 0, there is a finite dimensional C-vector space TX2 containing obstructions to lifting F i , R i . I Choose V ∈ Hom(S l , S n ) such that its columns represent a basis of TX2 . e It is possible to inductively construct F i , R i , G i−2 ∈ Hom(S, l i−2 l n e e e S ), C ∈ Hom(S , S ) solving I (F i R i )tr + C i−2 G i−2 ≡ 0 mod mi+1 . such that: 1. F i , R i , G i−2 , C i−2 reduce to F i−1 , R i−1 , G i−3 , C i−3 modulo mi ; 2. G i−2 and C i−2 vanish for i < 2; The Deformation Equation, Part II I If dim Sing(X ) = 0, there is a finite dimensional C-vector space TX2 containing obstructions to lifting F i , R i . I Choose V ∈ Hom(S l , S n ) such that its columns represent a basis of TX2 . e It is possible to inductively construct F i , R i , G i−2 ∈ Hom(S, l i−2 l n e e e S ), C ∈ Hom(S , S ) solving I (F i R i )tr + C i−2 G i−2 ≡ 0 mod mi+1 . such that: 1. F i , R i , G i−2 , C i−2 reduce to F i−1 , R i−1 , G i−3 , C i−3 modulo mi ; 2. G i−2 and C i−2 vanish for i < 2; 3. C 0 is of the form V · D, where D ∈ Hom(S d , S d ) is a diagonal matrix. The perturbation limi→∞ F i gives a formally versal deformation over the base space cut out by the rows of limi→∞ G i−2 .