Minimal Model Program
Minimal Model Program Learning Seminar
Spring 2021
Notes by Patrick Lei
Lectures by Joaquín Moraga
Zoom University
Disclaimer
These notes were taken during the seminar using the vimtex package of the editor neovim. Any errors are mine and not the speakers’. In addition, my notes are picture-free (but will include commutative diagrams) and are a mix of my mathematical style and that of the lecturers. If you find any errors, please contact me at plei@math.columbia.edu.
Also, I stopped attending the seminar after we finished the MMP in dimension . The seminar has continued onto multiplier ideals, and will later cover Birkar-Cascini-Hacon-McKernan and finally even more advanced topics in birational geometry.
Acknowledgements
I would like to thank Joaquín Moraga for organizing the seminar, finding mistakes in the notes, and more. I would also like to thank Nicolás Vilches for providing notes from the time I was unable to attend.
Seminar Website: https://web.math.princeton.edu/~jmoraga/Learning-Seminar-MMP
1 Overview
The goal of the minimal model program is to classify smooth projective complex varieties . Let be the tangent byndle and be the cotangent bundle. Then is called the canonical bundle, and can be written as for some Cartier divisor .
Question 1.0.1.
Can we understand the geometry of using numerical properties of ?
For a curve and a line bundle on , then . Then is ample (resp. antiample) if (resp ) for all curves . Similarly, is numerically trivial if for all curves .
Definition 1.0.2.
We say that is Fano if is antiample, Calabi-Yau if is numerically trivial, and canonically polarized if is ample.
Example 1.0.3.
If is a Fano curve, then . If is a CY curve, then is an elliptic curve. If is a canonically polarized curve, then .
Example 1.0.4.
Let be a smooth hypersurface of degree . Then by the adjunction formula, we have . Therefore, is Fano if , CY if , and canonically polarized if .
Remark 1.0.5.
If is an elliptic curve, then has for some curves and for others.
Now we consider various properties of different varieties:
| Fano | CY | Canonically polarized | |
|---|---|---|---|
| trivial | ? | generally infinite | |
| Automorphisms | linear algebraic groups | ? | finite groups |
| Birational automorphisms | monstrous | ? | finite groups |
| Geometry | simple geometry | ? | complicated, rich |
| Arithmetic | a lot of -points | ? | -points in a proper closed |
We have said nothing about Calabi-Yaus, but of course by the Beauville-Bogomolov decomposition, we can reduce to pure Calabi-Yaus ( for ), hyperkählers, and abelian varieties up to taking a finite cover.
Now let be a closed point on . Then there is a variety that is an isomorphism away from where the fiber above is an exceptional divisor parameterizing tangent directions at .
Example 1.0.6.
Consider points . Then if we blow up these points in a sequence, we obtain a sequence of varieties Over , the morphism is an isomorphism. For , clearly is not isomorphic to (they have different Picard ranks), but they are birational. We say that if they have isomorphic dense open subsets.
Now we can state the goal of the Minimal Model Program. If is projective and has “mild singularities” the goal is to prove that there exists a birational map and a fibration ( and positive dimensional general fiber) such that one of the following holds:
- 1.
is Fano;
- 2.
is Calabi-Yau;
- 3.
and is canonically polarized.
The way we will construct this birational morphism is by studying the geometry of curves on which intersect negatively. If under some hypotheses (extremity on ), we can find contracting precisely the curves which are numerically equivalent to a positive multiple of .
- 1.
If the curves numerically equivalent to a positive multiple of cover , then has positive-dimensional fibers, is a contraction, and the general fiber is Fano. This is called a Mori fiber space.
- 2.
If the curves numerically equivalent to a positive multiple of cover a divisor on , then we say that is a divisorial contraction and thus . still has nice singularities, so we can iterate this process.
- 3.
The last case is called a small contraction. The curves which are numerically equivalent to a positive multiple of cover a set of codimension at least . In this case, may have very bad singularities (by this, we mean that is not -Cartier). We construct a new birational morphism which contracts -positive curves.
Another type of surgery is a flip, which changes a locus of codimension at least . For example, consider on . Then write
Then is (locally) a cone over , so is not -Cartier. Therefore, we can blow up the vertex, and the exceptional divisor is . Now the can be collapsed onto each of the two factors, so we obtain a birational map . Here, if are the resulting curves, we have and .
This gives us the question:
Question 1.0.7.
Do flips always exist?
Now either the algorithm given by iterating this process always terminates or it continues infinitely. We are fine once we reach the Mori fiber space case, and there are only divisorial contractions, so the only possible problem is that there is an infinite sequence of flips.
Conjecture 1.0.8 (Termination of flips).
This algorithm always terminates after finitely many flips with either a Mori fiber space or a variety such that for every curve (in other words, is nef).
Conjecture 1.0.9 (Abundance).
has mild singularities and is nef. Then is basepoint-free for some .
If this is true, and contracts all -trivial curves, then either
- 1.
The general fiber has positive dimension. In this case, .
- 2.
. Then is birational and is canonically polarized.
Therefore, the goal of the MMP is achieved if we can solve the conjectures of existence of flips, termination of flips, and abundance. Existence of flips was proved by Birkar, Cascini, Hacon, and McKernan in 2006 and termination is known in dimension at most and in some cases in dimension . Finally, abundance is known in dimension at most .
2 MMP in Dimension
2.1 Rational Curves
Proposition 2.1.1 (Bend and Break).
Let be proper and be a smooth proper curve. Let and be nonconstant. Next, let be a pointed curve and such that
- 1.
.
- 2.
.
- 3.
is different from for general .
All of these imply that this is a nontrivial deformation of fixing . Then there exists and a union of rational curves such that is algebraically equivalent to and . In particular, there exists a rational curve through .
Proof.
First, compactify and let be the rational map. This map is undefined at by the rigidity lemma, so let be the normalization of the graph of . So we have a map and write .
Then we define . Then there exist such that is not an isomorphism over . Then we know that where is a birational transform of and is -exceptional. Then we set to be the restriction of to and .
By a lemma of Abhyankar, we know that is a union of rational curves, and then by the Lüroth theorem, we know that is a union of rational curves and
□
Lemma 2.1.2 (Abhyankar).
Let have mild singularities and be a proper birational morphism. For any , either is a point or is covered by rational curves.
Here is an intuitive image of the bend-and-break process:
Proposition 2.1.3 (Bend and Break II).
Let be a projective variety and be a nonconstant morphism. Let be a smooth pointed curve and such that
- 1.
;
- 2.
;
- 3.
is a surface.
Then is algebraically equivalent either to a reducible curve or a multiple curve.
Proof.
Let be a -bundle containing and consider the rational map . Then we can resolve the basepoints to obtain and induct on .
- Case 1: :
-
Consider the sections at and . Then let is ample on and then we see that and by the projection formula. By the Hodge index theorem, we see that (because are linearly independent). But then we know that , which is a contradiction.
- Inductive step:
-
Consider the diagram
Then is the first blowup in and will be a point such that . Let be the exceptional divisor of and be the strict transform of in . Then intersect at a point . But then is a morphism around . But then , which is reduced and irreducible. If is not defined at , then
which is a contradiction and thus is defined at . Then if is not defined at , then after blowing up , we see that must contain a component of multiplicity at least . Now contracting , we have the desired result by induction on . □
This tells us that to produce rational curves, we simply need to deform them with enough fixed points and use bend and break. But now we need to actually find rational curves.
Theorem 2.1.4.
Let be smooth and projective and be ample. For every , there exists a rational curve through such that
Proof.
Choose some curve through . Then the space of deformations of on fixing has dimension at least
We have several cases:
- 1.
If , then we are done.
- 2.
If , then we can replace with the composition by an endomorphism of large degree , then we see that
whenever is sufficiently large
- 3.
Assume . Then there are no endomorphisms of high degree, so assume are defined over . Then let be the reduction to . Now we apply the Frobenius map , which has degree . By generic flatness, we know that are the same for almost all , so by the same argument as in the genus case, we see have a rational curve on for almost all . By bend and break II, we can find a rational curve of the desired degree.
Then we use the fact that if a statement holds for all large enough, then it holds for the complex numbers, and we obtain a curve. This is analogous to the idea that if , then if the image of contains a Zariski-dense subset, then it contains the generic point. □
Theorem 2.1.5.
Let be a smooth projective variety and let be ample on . Assume there exists such that . Then there exists a rational curve such that and
Theorem 2.1.6 (Cone Theorem).
Let be smooth and projective. Then there exist countably many curves such that and
Proof.
Choose with and let be the closure of . Now choose positive on and negative somewhere on . Let be ample and
This means that is nef. Then let with and . Let be a sequence of curves approximating . Then we see that
is obtained by . Now we will replace with rational curves such that and
Because , we have
Fixing such that is ample, then we see that , so
Taking , we see that
a contradiction. □
Example 2.1.7.
Suppose is not nef. Then there are no rational curves on . For example, there are no rational curves of an abelian variety.
2.2 Singularities of the MMP
Consider pairs such that is normal quasiprojective and is -Cartier. These are called log pairs.
Definition 2.2.1.
Let is a resolution of singularities, is exceptional, and has simple normal crossings. Define the log discrepancy
Definition 2.2.2.
We say that is
- 1.
terminal if for every exceptional over .
- 2.
canonical if for every exceptional over .
- 3.
Kawamata log terminal if for every .
- 4.
log canonical if for every .
Now if is smooth projective and is pseudoeffective, then there exists such that is nef. By abundance, is semiample. Then there is a morphism such that is ample. Then terminal singularities are those that may appear in the terminal model, and canonical singularities are those that may appear on the canonical model.
Recall the adjunction formula: If is log smooth, then . Usually, is log canonical but not klt. But then but for every . Terminal singularities are the smallest category of singularities that we need to understand to run the minimal model program. On the other hand, log canonical is the largest class of singularities in which we can expect the MMP to work.
Example 2.2.3 (Examples of klt singularities).
Both cone singularities and quotient singularities are klt.
Proposition 2.2.4 (Cones).
Let be a log pair and an ample Cartier divisor on . Then define
Then is
- 1.
terminal if and only if with and terminal;
- 2.
canonical if and only if with and canonical;
- 3.
klt if and only if with and is klt;
- 4.
log canonical if and only if with and is log canonical.
In particular, the cone over a Fano is klt, the cone over a Calaby-Yau is log canonical, and cones over canonically polarized varieties are terrible.
Example 2.2.5.
Consider the cone over a rational normal curve of degree . Then resolving , we see that the exceptional and
and therefore .
Consider and elliptic curve. Then , so .
Now consider a finite group. Then has klt singularities.
Example 2.2.6.
In dimension , we have
- 1.
terminal is equivalent to smooth;
- 2.
canonical is equivalent to ADE;
- 3.
klt is equivalent to quotient singularity;
- 4.
log canonical is “equivalent” to quotient and/or elliptic cone.
Example 2.2.7.
In dimension , terminal singularities are classified as quotients of hypersurface singularities, also known as hyperquotient singularities. Then are given by actions of finite groups acting on hypersurface singularities of the form . Even for canonical singularities, we have no idea what they look like.
Example 2.2.8.
In dimension , there are examples of -fold terminal singularities with analytic embedding dimension for every (Kollar, 2010). By contrast, terminal singularities in dimension all have analytic embedding dimension .
Theorem 2.2.9 (Prokhorov, Xu, 2019).
Any klt singularity deforms to a klt cone singularity. For , there exists a flat morphism such that and is a klt cone singularity. This is just a deformation to the normal cone.
We will apply the following philosophy:
Any theorem for smooth projective varieties should work with klt singularities.
Example 2.2.10.
Let . By Beauville-Bogomolov (1970s), there exists a cover , where is a product of abelian varieties, irreducible Calabi-Yau varieties, and hyperkählers. There is an analogue for klt singularities that was proved by Druel, Campana,…in 2020.
Example 2.2.11.
Let be smooth projective and be nef. Then
where is strict Calabi-Yau, is hyperkahler, and is rationally connected. There is a version for klt singularities in progress.
We will now discuss localization of singularities. Suppose is a variety with terminal singularities and a subvariety of codimension . Then has terminal singularities if and only if is a smooth local ring, and this is equivalent to smoothness at the generic point of . In particular, if is terminal, the singularities must appear in codimension at least .
2.3 Vanishing
Let be a -negative curve. Then if for some effective, we see that , so . Now if is a log smooth pair, we have an exact sequence
If , then . This tells us that vanishing theorems help us find sections of line bundles.
Theorem 2.3.1 (Kodaira Vanishing).
Let be smooth projective and be ample. Then for all .
Sketch of Proof.
Let and be smooth. Then we have . Now we have
Setting with projection , we note that if are smooth, then is smooth. Consider a morphism and its pushforward . Now we consider the surjection
Now we use the result that factors through . By Serre, we see that for arbitrary . □
Theorem 2.3.2 (KV vanishing).
Let be a smooth projective complex and be a line bundle on such that , where is a big and nef -divisor, is a snc divisor, and . Then for .
Proposition 2.3.3.
Let be quasiprojective and normal, Cartier, and a positive integer. Suppose is finite and Cartier such that . If is smooth and is simple normal crossing, then is smooth and has simple normal crossings.
Lemma 2.3.4.
Let be finite. Then splits. If is a coherent sheaf on , then is a direct summand of . Finally, is a direct summand of .
Sketch of KV Vanishing.
Consider and write for some . Then consider such that . Then is a direct summand of . Then is a section of , so we can apply the index cover to obtain with smooth, is smooth, and has simple normal crossings. Then
so
Choosing , we see that is a direct summand of . Therefore,
But now we have reduced the number of components of , so by induction, the cohomology of the pullback vanishes. Now we need to consider . We know is big and nef, so , where is ample and is effective. Then there exists projective and birational such that , where is ample and has simple normal crossings with . Now let be an ample divisor. Then
But then , where is ample. But now we know that
for , so
and therefore for . But now setting , we see that
Theorem 2.3.5.
Let be a proper klt pair with a -Cartier divisor. Suppose that , where is big and nef. Then for . Equivalently, for .
Remark 2.3.6.
This philosophy fails for log canonical singularities.
2.4 Cone Theorems
Note that Kollar-Mori use outdated notation. What they call klt is actually sub-klt and what they call klt with we call klt. This still confuses people today, so let’s blame the person with the Fields medal.
Theorem 2.4.1 (Non-vanishing).
Let be proper with sub-klt1 and be a nef Cartier divisor. Assume is big and nef for some . Then for , we have
Theorem 2.4.2 (Basepoint-freeness).
Let be proper and be klt. Suppose is a nef Cartier divisor. Assume is big and nef for some . Then for , the linear sustem is basepoint-free.
Theorem 2.4.3 (Rationality).
Let be proper and be klt. Suppose that is not nef, is Cartier, and is a nef and big Cartier. Define
Then is rational and its denominator is at most .
Theorem 2.4.4 (Cone Theorem).
Let be a projective klt pair.
- 1.
There are countably many such that and
- 2.
For any ample and , we have
- 3.
If is extremal and -negative, then there exists a contraction morphism such that is mapped to a point if and only if .
- 4.
Let be as above and be a line bundle on such that . Then there exists on such that .
The logical structure is this: We use non-vanishing to find sections, then use various techniques (Kodaira vanishing) to lift enough sections to get basepoint-freeness, then we prove rationality by studying linear systems of the form , and finally we get the cone theorem as a formal consequence of convex geometry. However, we will follow the order of Kollar-Mori.
Remark 2.4.5.
The entire discussion will be carried out using klt singularities. However, these results hold when is log canonical at the cost of replacing nef and big with ample and at the cost of using significantly more machinery that was developed in the last 15 years.
Proof of basepoint-freeness.
By non-vanishing, we know that for . If is the base locus of , it suffices to prove that for . Next, we may consider a log resolution
Now we may perturb such that
is ample. Therefore , so is the fixed part. Therefore and .
We want to prove that there exists with such that for all , is not contained in . Let be an integer, be rational, and . Then we define
But now we see that the first term is is nef, the second term is basepoint-free, and the final two terms form an ample divisor. Therefore is ample. By Kodaira vanishing, we see that and . Now
where is effective and is prime. Therefore
and we have the exact sequence
Therefore we have a surjection for . Now is -exceptional, so
Now by non-vanishing, , so has a section not vanishing on . Because is -exceptional,
by the negativity lemma. Here, if , then , so if , then . However, must be disjoint from , so we have found a section of disjoint from . Therefore is a section of disjoint from . □
Lemma 2.4.6 (Negativity lemma).
Let be birational and proper between normal varieties and be -nef. Then is effective if and only if . If is effective, either or .
Theorem 2.4.7.
Let be a proper klt pair and be big and nef. Then the graded ring
is finitely-generated over .
This result holds even after dropping the big and nef assumption and was proved by Birkar, Cascini, Hacon, and McKernan in 2006.
Conjecture 2.4.8 (Abundance).
Let be projective and klt. If is nef, then it is semiample.
Conjecture 2.4.9 (Effectivity, folklore).
Let be projective klt. If is pseudoeffective (in the closure of the effective cone), then is effective.
2.4.1 Proof of the Cone Theorem
We will now prove the cone theorem. This relies on the following result from convex geometry.
Theorem 2.4.10.
Let and be a closed strictly convex clone. Let such that for some . Assume there exists such that for all with on and that
is rational of the form . Then
with with and such that do not accumulate in .
Let be ample and Cartier. Suppose is nef and define . Then for , we can set
Then is non-decreasing with respect to . Now if , then
Therefore is bounded above, integral, and non-decreasing, so this sequences stabilizes for large enough to . Now define the divisor
so . To prove this, let with . Then we know
For , we have
so . Because is nef, . If , then we stop. If not, we can iterate the above process to decrease again, so the desired result eventually holds. Now holds up to replacing with a large multiple.
Now we will show that for some , . If is a basis for , the linear functions
cannot all vanish, so for some . Now we can reduce to of dimension . This implies that and have the same closure.
Now we need to show that the do not accumulate in . This is a formal argument in linear algebra, so we skip it. Next, we need to prove that
In the limit as , we produce countably many (with a formal argument that is omitted here).
The next step is to prove that if is a -negative face, then there exists a nef Cartier divisor such that . Let be the linear span of and be the set of linear functions vanishing on . Because the generators of are defined over , then is also defined over . Take small enough such that is negative on . Because is extremal, we know . Therefore,
is a closed strictly convex cone intersecting at the origin. We also note that , so we can find a lattice point such that and . Therefore we can find a Cartier divisor which gives a supporting function of .
Now by assumption, is positive on . This means that is strictly positive on for , so is basepoint-free. Now let be the contraction associated by the Stein factorization to the linear system . Because is not an isomorphism, it constracts some curve . Similarly to teh smooth case, we may assume that
Finally, we prove that any line bundle on such that descends to , which means there exists a line bundle on such that . Now let be a Cartier divisor supporting . We know and that is defined by . Therefore, both and are pullbacks of Cartier divisors on . Write , and therefore we see that . This implies that is the pullback of a Cartier divisor on . If , then also supports , so for some Cartier divisor of . We simply set . □
2.4.2 Nonvanishing and Rationality
First we will prove rationality assuming non-vanishing and basepoint-freeness and then we will prove rationality.
Lemma 2.4.11.
Let be a smooth projective variety and be Cartier divisors. Then let be a normal crossing divisor with . Define
Assume that for certain , is nef and is ample. Then is a nonzero polynomial of degree at most .
Proof.
For , the sub is still ample. Then we know for by KV vanishing. By non-vanishing, we know and thus and thus . □
Lemma 2.4.12.
Let be a polynomial of degree at most . Assume vanishes for all sufficiently large integral solutions of for and . Then is rational and in reduced form it has denominator at most .
This is a purely arithmetic fact, so proof is omitted.
Proof of Rationality.
First, we reduce to the case in which is basepoint-free. Define
By basepoint-freeness, we know that is basepoint-free. For , we know and thus rationality of is equivalent to . If the denominator of divides , then the denominator of divides . Replacing with , now is basepoint-free.
Now we need to study the base locus , which is the base locus of . Then we know if and only if . If are large enough to be in the strip between , then stabilizes. To see this, the direction is semiample and then the base locus stabilizes by Noetherian induction to some . Now define to be the set of such that and . Now contains arbitrary large lattice points.
Next, we will define the polynomial and prove that it does not vanish. Define to be a log resolution of . Then if where is -exceptional, then is a polynomial of degree at most . Then note that if , is big and nef, so . Furthermore, we know that
From now on we will assume that is not rational.
Now we show that . If , then
is big and nef. Thus for . For large enough we know by the first lemma, so . But this implies that , so .
We show that for large in the strip. This will lead to a contradiction. Fixing , let be a log resolution satisfying:
- 1.
The divisor is ample;
- 2.
for .
- 3.
, where is the movable part and is the fixed part.
Then we can choose and such that
where is prime and and does not contain in its support. Now maps to a component of . Now define
We can choose with . Then , so
is nef. We conclude that is ample because is nef and the second term in the sum is ample. Therefore,
By the adjunction formula, we know
Applying the lemmas, we conclude that
and thus contains a section not vanishing at . Running the same argument using the negativity lemma implies that actually is disjoint from and thus is a section disjoint from . Therefore and thus is rational.
Now we need to control the denominator of . Assume the denominator is larger than the constant given by the second lemma. Setting , we choose large with . Then we have , so for all . Now choose such that is the maximum, equal to . Then we can show that for . Then there exist large enough in with and . Running the same argument from the previous paragraph gives us the desired conclusion. □
Proof of non-vanishing.
First we will reduce to the case where is smooth and is ample. Choose be a projective resolution and suppose and is a sub-klt pair. Then we know is nef and big, so is ample, so is sub-klt. Writing , we see that and
Replacing with , we have the desired reduction.
Next, we need to rule out the case that is numerically trivial. By KV vanishing, we know
Thus we may assume that is not numerically trivial.
Now we show that there exists such that if is not in the support of , then for we can find with multiplicity greater than . For ample and , we have , so we conclude that
Because is ample, then , where is a curve satisfying . Therefore, as . Now
Thus if , imposing that has multiplicity greater than at imposes at most conditions. As , then we know , so for large enough, some section satisfies the condition. Now we define , so has multiplicity at least at .
Next, consider a log resolution of that dominates . Then
- 1.
, ;
- 2.
is ample for ;
- 3.
where corresponds to the exceptional divisor of the blowup at .
Now we will perturb the coefficients and lift from lower dimension. Set
This is ample as long as and . Because we can choose arbitrarily large, the second condition is always achievable. But now because , we know and , so . Therefore we can write , so
Because is ample, then
so as long as . But by the non-vanishing theorem in lower dimension, this is true. □
2.4.3 The Relative Case
We will state a relative version of the cone theorem. The proof is essentially the same as in the absolute case.
Theorem 2.4.13 (Relative Cone Theorem).
Let be a projective contraction of varieties over an algebraically closed field of characteristic . Let be a klt pair.
- 1.
There are countably many curves such that , , and
- 2.
For any and -ample , we have
- 3.
Let be a -negative extremal face. Then there is a unique contraction commuting with the structure morphisms to such that is mapped to a point if and only if .
- 4.
Let be a line bundle on such that for every curve with . Then there exists a line bundle on such that .
Remark 2.4.14.
So far everything is a “formal” consequence of Kodaira vanishing and resolution of singularities (along with various classical arguments).
Remark 2.4.15.
Bhatt and Lurie proved a version of the Riemann-Hilbert correspondence in positive characteristic. Bhatt proved the Cohen-Macaulayness of the integral closure of an excellent Noetherian domain. Using these techniques and results, the Minimal Model Program has recently been generalized in two different directions:
- 1.
In dimension in mixed characteristic (essentially over ) by Bhatt-Ma-Patakfalvi-Schwede-Tucker-Waldron-Witaszek.
- 2.
In characteristic , most of the MMP works over an excellent -scheme by Murayama-Lyu.
Of course one may wonder why we need to run the MMP relative over a base. Consider a smooth projective family such that is ample over . We want a good compactification, but naively compactifying gives us an arbitrary central fiber that may not even be normal. Using the resolution of singularities, we may consider a log resolution. This has many components and is not ample over in general. Then we can try to run the MMP over the base, and we expect a commutative diagram
such that is nef over the base and the singularities of are slc. Here, slc means that the normalization is log canonical and has nodal singularities at codimension points.
We may also use the relative MMP to study singularities. Here, let and consider a log resolution . Then we write . We may then perturb the coefficients of :
- •
If , we may decrease to ;
- •
If , we can increase to ;
This obtains a new boundary with the same support as . If we run the MMP for over , we obtain a partial resolution of singularities which has the singularities of the minimal model program. Therefore, by studying the exceptional divisors of the previous partial resolution and the singularities of the MMP, we can understand the singularities of .
2.5 Running the Minimal Model Program
2.5.1 Flipping Contractions and Flips
Definition 2.5.1.
A map is a flipping contraction if is and -factorial, , and is a small birational contraction, and is ample over .
Remark 2.5.2.
When we work with log pairs we assume is ample over .
Remark 2.5.3.
is never -factorial. Even worse, is not -Cartier.
Definition 2.5.4.
Let be a flipping contraction. We say that is a flip if it is a small birational map, is -Cartier, and there is a projective morphism such that is ample over .
Lemma 2.5.5.
Let be a small birational map between normal varieties and let be a Weil divisor on . Then .
Lemma 2.5.6.
Let be a flipping contraction and let be a flip. Then and is -factorial.
Proof.
Let be a divisor on and on be the pushforward. We will find such that , where is the extremal ray defining the flipping contraction. We know is -factorial, so is Cartier for . Then we see that
for some Cartier divisor on . Then
is Cartier, so is -Cartier. For equality of , we prove that induces an isomorphism between Weil divisors modulo linear equivalence. □
Lemma 2.5.7.
Let be a projective contraction between normal varieties with . Assume that the exceptional locus of contains a divisor. Then is the contraction of a unique irreducible divisor.
Proof.
Suppose there are two divisors . We can find covering with . Then we can find such that . We will show that is positive. Assume that does not intersect . Then . Choosing a general inside , we may assume . Therefore , so . But is an effective divisor which is contracted, so it must be covered by -negative curves, which gives a contradiction. We conclude that must be the only component. □
Proposition 2.5.8.
Let be a flipping contraction for klt. Then the flip exists if and only if
is a finitely-generated -algebra. If this is the case, then
Proof.
Assume the flip
exists. We know that is small, so
Moreover, is ample over , so
Now assume that is a finitely-generated -algebra and define . Then is an isomorphism in codimension . It could happen that the indeterminancy locus is a divisor. We know is an isomorphism over . Then is just a sum of copies of the structure sheaf on . Therefore, is an isomorphism over and suppose is mapped to a higher codimension cycle by . Then
for some . Because is exceptional over , we have
We have a natural inclusion , so there are no contracted divisors. Thus is small. Then the property that is by Lemma 2.5.6. □
2.5.2 Finite generation of the canonical ring
Conjecture 2.5.9.
Let be a projective morphism and be a klt pair. Then
is a finitely-generated -algebra.
Remark 2.5.10.
If is a smooth projective variety, then is finitely generated over the base field .
2.5.3 Running the MMP
Now we may finally discuss the process of actually running the Minimal Model Program.
- 1.
Let be a klt pair and be a projective morphism. If is nef over , then we stop and call this a minimal model over . If is not nef over , we consider an extremal ray in which is -negative.
- 2.
Let be the contraction define by .
- (a)
We have , is ample over , and the general fiber is klt. Hence the general fiber is klt Fano. In this case, we stop the algorithm and call this a Mori fiber space.
- (b)
We have and contains a divisor in its exceptional locus. This is a divisorial contraction. Then is -factorial and . Write and . Return to step 1.
- (c)
We have and is a small birational map. We find the flip and define and . We know that is -factorial provided that is -factorial and . Return to step 1.
- (a)
Remark 2.5.11.
Using the negativity lemma, we can prove that is klt in step 2b.
The possible outcomes for the MMP are either a minimal model, which assuming abundance maps to the canonical model, or a Mori fiber space.
2.6 More about singularities
2.6.1 Surface singularities
Theorem 2.6.1.
- 1.
A point is a surface klt singularity if and only if is the quotient of be a finite subgroup of .
- 2.
A point is a canonical surface singularity if and only if it is the quotient of by a finite subgroup of .
- 3.
A point is a terminal surface singularity if and only if it is a smooth point.
The idea of the proof is as follows: If is -Cartier, we can take its index one cover. This is a finite Galois morphism that is quasi-étale (étale in codimension ) such that is a Cartier divisor. Then we can write . Because is klt and is Cartier, its log discrepancies are positive integers, so it is canonical. Therefore it suffices to study canonical singularities.
For canonical singularities, we obtain the following:
Theorem 2.6.2.
Let be a canonical surface singularity. Then has embedding dimension . Moreover, up to analytic change of coordinates, the following is a complete list of possible singularities:
- Type A:
-
For , the singularity has equation and dual graph
- Type D:
-
For , the singularity has equation and dual graph
- Type E:
-
The singularity has equation and dual graph
The singularity has equation and dual graph
The singularity has equation and dual graph
The idea of the proof is to study the dual graph of the resolution and use the Weierstrass preparation theorem to write down the equations.
2.6.2 Geography of singularities
The geography of singularity types with implications looks like this:
We will now define some of these singularity classes and prove some of the implications.
Definition 2.6.3.
Let be a Noetherian local ring and be a finite -module. Write . Then is called Cohen-Macaulay of one of the following equivalent conditions holds:
- 1.
There exists , where such that is not a zero divisor in for all . Here, is called a -regular sequence.
- 2.
If for and , then is an -regular sequence.
A coherent sheaf on a scheme is Cohen-Macaulay if is Cohen-Macaulay over for every . Finally, a scheme is Cohen-Macaulay if its structure sheaf is Cohen-Macaulay.
There is a condition of Serre, which is slightly weaker than being Cohen-Macaulay. We say that a ring satisfies the properties
- :
-
For all such that , then is regular.
- :
-
For all , .
Examples 2.6.4.
Normality is equivalent to and , so all normal surfaces are Cohen-Macaulay. If is Cohen-Macaulay and a group acts on , then is Cohen-Macaulay by Hochster-Roberts. This means that quotient singularities are Cohen-Macaulay. The rings and are Cohen-Macaulay. However, the scheme is not Cohen-Macaulay at the origin.
Definition 2.6.5.
Let be a variety over a field of characteristic and be a resolution of singularities. We say that is rational if
- 1.
(here, is normal).
- 2.
The higher derived pushforwards are given by for .
We say that has rational singularities if every resolution is rational.
Equivalently, the morphism is a quasi-isomorphism in the derived category.
Examples 2.6.6.
All singularities of the form are rational. If , then , so is rational. However, the cone over an elliptic curve is not rational.
Definition 2.6.7.
A point is symplectic if is normal at and there exists a symplectic form on which extends to the exceptional divisor over on any resolution.
Definition 2.6.8.
Let be an embedding of a scheme into a regular scheme and be a log resolution of which is an isomorphism outside . Let be the reduced preimage of in . Then we say has Du Bois singularities if is a quasi-isomorphism.
Remark 2.6.9.
Du Bois singularities commonly appear in Hodge theory.
Remarks 2.6.10.
- •
Serre duality for Cohen-Macaulay sheaves
- •
If is Cartier and is Cohen-Macaulay, then so is .
Proposition 2.6.11.
Let be a variety over a field of characteristic and be a resolution. The following are equivalent:
- 1.
is rational.
- 2.
is Cohen-Macaulay and .
Proof.
Let be projective and be an ample Cartier divisor on . Then for . By Serre duality, we have for and . Then the Leray spectral sequence
Assuming that is rational, we then obtain
Now we will prove that for and only if is Cohen-Macaulay.2 Let be a general element. We obtain the exact sequence
By vanishing, we see that for and . Therefore by induction, is Cohen-Macaulay and thus is Cohen-Macaulay. Taking the case of , we obtain
for all . This implies that by a result in Hartshorne.
Now assume that is Cohen-Macaulay and . We will induct on the dimension, and CM curves are smooth, so we proceed to the inductive step. We will prove that is supported in some -dimensional locus. Let be general and , so is a resolution. Then
By induction, we see that is trivial outside a zero dimensional set. Therefore for or if and . By the spectral sequence, we obtain for and
This implies that for . When , is the dual to the morphism
because . Thus is an isomorphism and thus . □
Lemma 2.6.12.
Let be a klt pair, be basepoint free, and be a general element. Then is klt.
Lemma 2.6.13.
Let be finite with . Then if both are log pairs, then is klt (resp. log canonical) if and only if is klt (resp. log canonical).
The idea of the proof is to use the Riemann-Hurwitz formula on a log resolution to compare the discrepancies and observe that for some positive integer (a ramification index).
Theorem 2.6.14 (Elkik 1981).
If is dlt, then has rational singularities.
Proof.
Write , where is a log resolution, , and . By KV vanishing, for . Now let be an ample Cartier divisor on . WE have a commutative diagram
Therefore we have
and for and by KV vanishing. We want to conclude that for . Using vanishing and the spectral sequence, we see that is an isomorphism. But then by vanishing again, the top left corner is , and thus everything in the diagram must vanish. Therefore is Cohen-Macaulay.
Now if we take , we obtain an injection . By Serre duality, the map is a surjection, and surjectivity for implies that , and they are isomorphic because they are both rank reflexive sheaves. Therefore has rational singularities. □
Proposition 2.6.15.
If is Gorenstein (this means is Cartier), then rational is equivalent to canonical.
Proof.
We know canonical implies dlt implies rational. Now suppose that is rational and Gorenstein and let be a resolution. Then , where are integral and effective with no common support. To prove that is canonical, we need to prove that . If we push-forward and , then we get associated primes on the image of , and this contradicts . Therefore is trivial, so is canonical. □
Proposition 2.6.16.
Symplectic singularities are rational and Gorenstein.
Proof.
Let be a resolution. If be a -form, then generates the line bundle . The fact that extends to a regular holomorphic form implies that extends. Thus . In addition, we see that . □
We can characterize the condition of being Gorenstein as follows: A local ring is Gorenstein if and only if there exists a regular sequence such that is Gorenstein and -dimensional. By Nakayama’s lemma, this implies that is Gorenstein for every .
Lemma 2.6.17.
Let be an index canonical threefold singularity and be a general hyperplane section. Then either is a Du Val singularity or an elliptic singularity.
The idea of the proof is that if is a resolution, then . If , we have a Du Val singularity, and if , then we have an elliptic singularity.
Theorem 2.6.18.
All terminal threefold singularities of index are cDV (compound Du Val), which means a one-parameter deformation of a Du Val singularity.
Example 2.6.19.
Let be a finite group. We know is a quotient singularity and thus klt. If , then is symplectic.
Example 2.6.20.
If is a cone and is symplectic, then it is isomorphic to a Lie group quotient by the smallest nonzero nilpotent orbit.
2.7 Duality Theory
Given a projective variety , we want a sheaf such that for every Cohen-Macaulay sheaf. Recall that Serre duality establishes this duality for for any quasicoherent sheaf. Moreover, if is locally free, then
Definition 2.7.1.
A dualizing sheaf is a coherent sheaf with a surjection
such that for all coherent , induces an isomorphism
The pair is unique if it exists.
Proposition 2.7.2.
Let be a finite morphism, . Let ,
- 1.
There exists an isomorphism . In particular, there is a natural isomorphism .
Proof.
Assume and . Then we have reduced to the identity
but this is obvious. Simply write for some . □
Proof.
Let be a finite morphism of proper schemes of pure dimension . If exists, then exists and . □
Proof.
We simply compute
Here, finiteness gives us for and thus the Leray spectral sequence applies. □
Corollary 2.7.3.
The sheaf exists and is an -sheaf for projective varieties over .
Proof.
We always have a finite surjective morphism (embed in large projective space, then project down). Then exists, so does also. Also is if and only if is, but . □
Corollary 2.7.4.
If is a projective scheme of pure dimension over , a coherent sheaf such that has pure dimension , then
- 1.
If is Cohen-Macaulat, so is . The converse holds if is .
- 2.
If is , then is Cohen-Macaulay if and only if is.
Proof.
Let be surjective and finite. Then is Cohen-Macaulay if and only if is locally free, so is Cohen-Macaulay if and only if
and is locally free if and only if is . □
Theorem 2.7.5.
Let be a projective scheme of pure dimension , a Cohen-Macaulay sheaf such that has pure dimension. Then
Proof.
As before, let be finite and surjective. We know that . Then is locally free, so by Serre duality, we have
and therefore we obtain
Proposition 2.7.6.
Let be a projective Cohen-Macaulay scheme of pure dimension , a Cartier divisor. Then .
Proposition 2.7.7.
Let be a normal projective variety of dimension . Then .
These two facts tive the classical adjunction formula.
Proposition 2.7.8.
Let be smooth and be a smooth Cartier divisor. Then .
Next, we want to compare singularities of with .
Remark 2.7.9.
If is Cartier in codimension , then . If there are quotient singularities in codimension , then , where is a codimension point of , orbifold of order .
The idea of adjunction is that if has nice singularities, then so does . The convsers is called the inverse of adjunction.
Proposition 2.7.10.
Let be normal and be a normal Weil divisor that is Cartier in codimension . Let be closed and . Assume that is -Cartier. Then
Proof.
Let be a log-resolution of and write and is smooth. By doing more blowups, we may assume that if ie exceptional such that , then the center of is contained in . We have
and , and therefore
Now is distinct from , so if , then is -exceptional and the center of is contained in . this shows that every log discrepancy that happens in also happens in in such a way that the center lies inside . □
Therefore, if is plt, then is klt. The inverse of adjunction is more difficult. Suppose that has nice singularities, but has bad singularities. In particular, there exist exceptional over such that . This contradicts the philosophy that in a Fano object, the locus where it’s more singular is contracted. The theorem that realizes this is called Kollàr-Shokurov connectedness.
Definition 2.7.11.
Let be a log-canonical. We say that is a log-canonical center if there exists over with and . We denote by the union of all log-canonical centers. This is a closed subset.
Theorem 2.7.12 (Kollaàr-Shokurov connectedness).
Let be projective, be log-canonical, and be ample over . Then is connected over .
Example 2.7.13.
Let be the structure morphism and note that . The coefficients of the dropped points must have sum stricly less than .
Remark 2.7.14.
Big and nef suffices in place of ample.
Theorem 2.7.15.
Let be a proper birational mapping with smooth and normal. Let be a smooth normal crossings -divisor on such that and is -nef. Write
Then is connected in any neighborhood of any fiber of .
Remark 2.7.16.
A field is if every homogeneous of degree at most has a nontrivial zero. A field is pseudo-algebraically closed if every geometrically integral -variety has a -point.
Conjecture 2.7.17 (Ax).
Every pseudo-algebraically closed field of characteristic is .
This is true and proved by Kollàr.
Proof of Theorem.
Note that . By KV vanishing, we have . Then we have the exact sequence
so . Now is -exceptional, so . Writing is a neighborhood of , we obtain
This will give a surjection , which is not true by connectedness. □
Theorem 2.7.18.
Let be normal and be a normal Weil divisor Cartier in codimension . Let be a -divisor such that is -Cartier. Then
- 1.
is plt near if and only if is klt.
- 2.
Assume is -Cartier and it klt. Then is log canonical near if and only if is log canonical.
Proof.
Take a log resolution of and . Write and . Then contains a component of codimensino at least . By adjunction, we have
so is plt near if and only if and is klt if and only if . By Kollàr-Shokurov connectedness, given , there exists such that is connected. Here . Moving around finishes the argument. □
We will now consider adjunction for higher codimension centers. We know is snc if and only if is. For , let be a log-canonical center on . If has codimension in , do adjunction to and repeat. If has codimension at least , define such that the singularities of around can be compared to those of .
This is indeed possible. First blow up enough so that maps to . Write . Then we have
and this is trivial over . We want to write , where measures the singularities of the fiber , the variation of moduli.
2.8 Terminal threefold MMP
Let be a -factorial terminal projective threefold and be a flipping contraction. Suppose , is ample over , and has terminal singularities and is smooth in codimension . First, we want to construct a rational map which is an isomorphism in codimension such that , is ample over , and has terminal singularities. Also, is unique and coincides with
provided that this ring is a finitely generated -algebra. Denote this ring by .
Proposition 2.8.1.
The ring is finitely generated as an -algebra if and only if it is finitely generated locally over (and this is even true locally analytically).
For a geometric picture, let be a terminal threefold and suppose contracts several curves. Then is not -factorial in general. In 1988, Mori proved that these curves can be contracted one-by-one in the analytic category (in the algebraic category, this is clearly impossible by the previous discussion). Now our terminal threefold flipping contractions lead us to studying extremal neighborhoods. Here, we consider an analytic neighborhood of the fiber above . We may assume that is a rational singularity. Now what happens if our extremal neighborhood is smooth?
Proposition 2.8.2.
Let be an extremal neighborhood. Then , , and has a smooth member.
Proof.
We know that because and . Now we have an exact sequence
and thus we have an isomorphism . Taking the degree, we conclude that . Now if we tensor with , we obtain the exact sequence
Because , we have .
Now let and be a smooth divisor on the germ . Then extends naturally to a divisor of . We need to show that . Here, we know that and the isomorphism is induced by . Thus . □
Corollary 2.8.3.
In the above case, is the blowup of a smooth threefold along a smooth curve .
Corollary 2.8.4.
Let be a terminal threefold flipping contraction. For every , contains a singularity.
By the classification of terminal threefold singularities, for any extremal neighborhood, there are at most three singular points and there is a nice divisor passing through some of them.
Theorem 2.8.5 (Mori 1988).
Let be an extremal neighborhood. Then one of the following holds for ():
- 1.
Either has a member with Du Val singularities, or
- 2.
has a member such that the double cover of branched over with only Du Val singularities.
Because is Du Val, we know is purely log terminal. Now the finite generation of the ring is implied by the finite generation of the ring
This is now a problem about projective surfaces, and apparently this is doable by hand. Here, is a log-Calabi Yau surface with at most three singular points. Now we conclude the existence of flips for terminal threefolds.
Now we need to prove termination of flips.
Theorem 2.8.6.
An arbitrary sequence of three-dimensional extremal canonical -flips is finite.
Lemma 2.8.7.
Let be a -flip of a three=dimensional canonical pair , where . Let be a flipped curve and be the exceptional divisor obtained by blowing up . Then is smooth along and
where the multiplicities are of along .
Proof.
Because is a flipped curve, is smooth along the generic point of . Indeed, is terminal along , so it is smooth. If there is a nonterminal valuation with center on , then there is a noncanonical valuation on . □
We will now define the difficulty function. Let be a canonical pair with pairwise distinct prime divisors. Set and . Also write
For example, with compound Du Val singularities, we have .
Remark 2.8.8.
The difficulty function measures the number of nonterminal valuations.
We are now ready to prove termination of canonical threefold flips.
Proof of termination of flips.
Write , where . Then in the sequence of flips, we have
If , then . Therefore, after finitely many flips, and thus there are no more flips.
Now assume that . Then we know is nondecreasing. If is the flipped curve for , assume it is contained in . Then and . Therefore for , contains no flipped curves.
Now denote by the normalization of . Then is a birational morphism. The exceptional curves of for are linearly independent. At some point, we have for . This means that both the flipping and flipped curves are disjoint from . Thus , so now flips for induce flips on , and by induction, these flips stop. □
Conjecture 2.8.9 (Abundance).
If is klt and is nef, then is semiample.
This is proved for terminal threefolds by Kawamata. Now we have settled the minimal model program for terminal threefolds. Existence and termination of flips were proved in 1988 by Mori and later in a more general case by Kollár-Shokurov, and abundance was proved by Kawamata in the 1990s. A historical sketch of more recent developments is as follows:
| Existence of Flips | Termination of Flips |
|---|---|
| In the 2000s Kollár-Shokurov proved existence for log-canonical threefolds | In 2004, Alexeev-Hacon-Kawamata proved termination for fourfolds with effective. |
| In 2005, Hacon and McKernan proved flips exist in dimension provided the MMP works in dimension . | In 2000, Fujino proved termination of flips for a terminal fourfold. |
| In 2006, Birkar-Cascini-Hacon-McKernan proved existence of flips for klt. | They also proved termination of flips for klt and big. |
| In 2010, Hacon-Xu and Birkar proved exietence of flips for log canonical. | In 2018, Joaquín proved termination of flips for a log canonical fourfold with pseudoeffective. |
Termination is unknown for klt uniruled fourfolds and in dimension at least . Abundance is known for log canonical pairs in dimension and log canonical pairs in dimension with uniruled. It is unknown for log canonical pairs with pseudoeffective. Some people believe that abundance should be tackled using different techniques from the minimal model program.
2.9 Applications to Singularities
Conjecturally, the MMP contracts or flips the locus . This is known in dimension and follows from termination and abundance. Recall that if and is an ample divisor on , then
This is a countable union of algebraic varieties. Then (someone) showed the existence of a divisor on a certain blowup of whose is a countable union of curves. Here, the dynamics of what happens lead to something non-algebraic.
The first application we will give is terminalization. Let be a klt pair. Then there exists a projective birational morphism such that is terminal and extracts exactly the divisors with .
To prove this, let be a log resolution of . Then we know . This is not necessarily a boundary, so write for the result of killing all negative coefficients. By negativity, we see that contains all divisors with . Now we run the minimal model program for , and this contracts all of these divisors. Because is terminal, it remains terminal after we run the minimal model program.