Augmentations and Rulings of Legendrian Links in \#^{k}(S^{1}\times S^{2})

Augmentations and Rulings of Legendrian Links in

C. Leverson Duke University, Durham, NC 27708 cleverso@math.duke.edu
July 19, 2019
Abstract.

Given a Legendrian link in , we extend the definition of a normal ruling from given by Lavrov and Rutherford and show that the existence of an augmentation to any field of the Chekanov-Eliashberg differential graded algebra over is equivalent to the existence of a normal ruling of the front diagram. For Legendrian knots, we also show that any even graded augmentation must send to . We use the correspondence to give nonvanishing results for the symplectic homology of certain Weinstein -manifolds. We show a similar correspondence for the related case of Legendrian links in , the solid torus.

1. Introduction

Augmentations and normal rulings are important tools in the study of Legendrian knot theory, especially in the study of Legendrian knots in . Here, augmentations are augmentations of the Chekanov-Eliashberg differential graded algebra introduced by Chekanov in [Chekanov] and Eliashberg in [EliashbergInvariants]. Chekanov describes the noncommutative differential graded algebra (DGA) over associated to a Lagrangian diagram of a Legendrian link in combinatorially: The DGA is generated by crossings of the link; the differential is determined by a count of immersed polygons whose corners lie at crossings of the link and whose edges lie on the link. This is called the Chekanov-Eliashberg DGA and Chekanov showed that the homology of this DGA is invariant under Legendrian isotopy. Etnyre, Ng, and Sabloff defined a lift of the Chekanov-Eliashberg DGA to a DGA over in [EtnyreInvariants]. Following ideas introduced by Eliashberg in [EliashbergWave], Fuchs [FuchsAug] and Chekanov-Pushkar [ChekanovFronts] gave invariants of Legendrian knots in using generating families, functions whose critical values generate front diagrams of Legendrian knots, by decomposing the generating families. These are generally called “normal rulings.”

These two invariants are very closely related; Fuchs [FuchsAug], Fuchs-Ishkhanov [FuchsIshkhanov], and Sabloff [SabloffAug] showed that the existence of a normal ruling is equivalent to the existence of an augmentation to of the Chekanov-Eliashberg DGA for Legendrian knots in . Here, given a unital ring , an augmentation is a ring map such that and . One of the main results of [Leverson] is that the equivalence remains true when one looks at augmentations to a field of the lift of the Chekanov-Eliashberg DGA from [EtnyreInvariants] to the DGA over for Legendrian knots in . We extend the result to Legendrian links in to prove the main result of this paper.

Theorem 1.1.

Let be an -component Legendrian link in . Given a field , the Chekanov-Eliashberg DGA over has a -graded augmentation if and only if a front diagram of has a -graded normal ruling. Furthermore, if is even, then .

The final statement tells us that for all even graded augmentations , . In particular, if is a knot, then any even graded augmentation sends to .

For , an analogous correspondence can be shown for Legendrian links in . A Legendrian link in with the standard contact structure is an embedding which is everywhere tangent to the contact planes. We will think of them as Gompf does in [GompfHandlebody]. For an example, see Figure 2. In this paper, we extend the definition of normal ruling of a Legendrian link in to a Legendrian link in . We can then define the ruling polynomial for a Legendrian link in and show that the ruling polynomial is invariant under Legendrian isotopy.

Theorem 1.2.

The -graded ruling polynomial with respect to the Maslov potential (which changes under Legendrian isotopy) is a Legendrian isotopy invariant.

In [Ekholms1s2], Ekholm and Ng extend the definition of the Chekanov-Eliashberg DGA over to Legendrian links in . The main result of this paper uses Theorem 1.1 to extend the correspondence between normal rulings and augmentations to a correspondence for Legendrian links in .

Theorem 1.3.

Let be an -component Legendrian link in for some . Given a field , the Chekanov-Eliashberg DGA over has a -graded augmentation if and only if a front diagram of has a -graded normal ruling. Furthermore, if is even, then .

Notice that one can consider Legendrian links in as being Legendrian links in . In this way, this result is a generalization of the correspondence in [Leverson] and Theorem 1.1.

Along with the work of Bourgeois, Ekholm, and Eliashberg in [BEE], Theorem 1.3 gives nonvanishing results for Weinstein (Stein) -manifolds. In particular:

Corollary 1.4.

If is the Weinstein -manifold that results from attaching -handles along a Legendrian link to and has a graded normal ruling, then the full symplectic homology is nonzero.

This follows from Theorem 1.3 as the existence of a normal ruling implies the existence of an augmentation to , which, by [BEE], is necessary for the full symplectic homology to be nonzero.

We show a correspondence for Legendrian links in the -jet space of the circle . In [NgSolidTorus], Ng and Traynor extend the definition of the Chekanov-Eliashberg DGA to Legendrian links in . In [LavrovSolidTorus], Lavrov and Rutherford extend the definition of normal ruling to a “generalized normal ruling” of Legendrian links in and show that the existence of a generalized normal ruling is equivalent to the existence of an augmentation to of the Chekanov-Eliashberg DGA of a Legendrian link in . In §LABEL:sec:solidTorus, we show that this correspondence holds for augmentations to any field of the Chekanov-Eliashberg DGA over .

Theorem 1.5.

Let be a Legendrian link in . Given a field , the Chekanov-Eliashberg DGA over has a -graded augmentation if and only if a front diagram of has a -graded generalized normal ruling.

1.1. Outline of the article

In §2 we recall background on Legendrian links in and . We give definitions of the Chekanov-Eliashberg DGA over , with sign conventions, and augmentations of the DGA in both and . We also define normal rulings for links in and show that the ruling polynomial is invariant under Legendrian isotopy. In §3, we prove Theorem 1.1. In §4, given an augmentation, we construct a normal ruling proving one direction of Theorem 1.3. In §LABEL:sec:rulingToAug, given a normal ruling, we construct an augmentation, finishing the proof of Theorem 1.3. In §LABEL:sec:solidTorus, we prove Theorem 1.5. In the Appendix, we give the nonvanishing symplectic homology result.

1.2. Acknowledgements

The author thanks Lenhard Ng and Dan Rutherford for many helpful discussions. This work was partially supported by NSF grants DMS-0846346 and DMS-1406371.

2. Background Material

2.1. Legendrian Links in

In this section we will briefly discuss necessary concepts of Legendrian links in . We will follow the notation in [Ekholms1s2].

Definition 2.1.

Let . A tangle in is Legendrian if it is everywhere tangent to the standard contact structure . Informally, a Legendrian tangle in is in normal form if

  • meets and in groups of strands, where the groups are of size , from top to bottom in both the and projections,

  • and within the -th group, we label the strands by from top to bottom at in both the and projections and in the projection, and from bottom to top at in the projection.

Every Legendrian tangle in normal form gives a Legendrian link in by attaching -handles which join parts of the projection of the tangle at to the parts at . In particular, the -th -handle joins the -th group at to the -th group at and connects the strands in this group with the same label at and through the -handle. See Figure 2.

Every Legendrian link in has an -diagram of the form given by Gompf in [GompfHandlebody], which we will call Gompf standard form. The left diagram of Figure 2 is an example of a link in Gompf standard form. Any link in Gompf standard form can be isotoped to a link whose -projection is obtained from the -diagram by resolution. The resolution of an -diagram of a link is obtained by the replacements given in Figure 1. For an example, see Figure 2. By [Ekholms1s2], an -diagram obtained by the resolution of an -diagram of a link in Gompf standard form is in normal form. Thus, we can assume that the -diagram of any Legendrian link is in normal form.

\labellist\pinlabel

[r] at 91 96 \pinlabel [r] at 91 73 \pinlabel [r] at 91 50 \pinlabel [r] at 91 27

\pinlabel

[B] at 538 83 \pinlabel [B] at 538 63 \pinlabel at 538 43 \pinlabel [t] at 538 26

\pinlabel

[b] at 407 34 \pinlabel [b] at 438 43 \pinlabel [b] at 471 53 \pinlabel [b] at 462 33 \pinlabel [b] at 490 44 \pinlabel [b] at 512 34

\endlabellist
Figure 1. Resolutions of an -diagram in Gompf standard form.
\labellist\pinlabel

at 165 280 \pinlabel at 165 258 \pinlabel at 165 235 \pinlabel at 165 212

\pinlabel

at 165 123 \pinlabel at 165 92

\pinlabel

at 724 280 \pinlabel at 724 258 \pinlabel at 724 235 \pinlabel at 724 212

\pinlabel

at 724 123 \pinlabel at 724 92

\pinlabel

at 1200 383 \pinlabel at 1200 346 \pinlabel at 1200 307 \pinlabel at 1200 270

\pinlabel

at 1200 117 \pinlabel at 1200 80

\pinlabel

at 2015 353 \pinlabel at 2015 325 \pinlabel at 2015 295 \pinlabel at 2015 273

\pinlabel

[b] at 1853 107 \pinlabel [t] at 1853 84 \endlabellist

Figure 2. The left gives a Legendrian -diagram of a link in in Gompf standard form. The right gives the resolution of the Legendrian link to an -diagram of a Legendrian isotopic link.

2.2. Definition of the DGA and augmentations in

This section contains an overview of the differential graded algebra over presented by Ekholm, Ng in [Ekholms1s2]. Let be a Legendrian link in , where the denote the components of and . Let be the number of strands of which go through the -th -handle with the total number of strands at .

2.3. Internal DGA

We will define the internal DGA for a Legendrian link in , but one can easily extend the definition to the internal DGA for a Legendrian link in by defining the internal DGA as follows for each -handle separately.

Let be the -tuple where is the rotation number of the -th component and let be the -tuple of a choice of Maslov potential for each strand passing through the -handle (see §2.5).

Let denote the DGA defined as follows. Let be the tensor algebra over generated by for and for and . Set , , and

for all . Define the differential on the generators by

where , is the Kronecker delta function, and we set for . Extend to by the Leibniz rule

From [Ekholms1s2], we know has degree , , and is infinitely generated as an algebra, but is a filtered DGA, where is a generator of the -th component of the filtration if .

Given a Legendrian link , we can associate a DGA to each of the -handles. We then call the DGA generated by the collection of generators of for with differential induced by , the internal DGA of .

2.4. Algebra

Suppose we have a Legendrian link in normal form with exactly one point labeled within the tangle (away from crossings) on each link component of (corresponding to ). We will discuss the case where there is more than one base point on a given component in §2.11.

Notation 2.2.

Let denote the crossings of the tangle diagram in normal form. Label the -handles in the diagram by from top to bottom. Recall that denotes the number of strands of the tangle going through the -th -handle. For each , label the strands going through the -th -handle on the left side of the diagram from top to bottom and from bottom to top on the right side, as in Figure 2.

Let be the tensor algebra over generated by

  • ;

  • for and ;

  • for , , and .

(In general, we will drop the index when the -handle is clear.)

2.5. Grading

The following are a few preliminary definitions which will allow us to define the grading on the generators of .

Definition 2.3.

A path in is a path that traverses part (or all) of which is connected except for where it enters a -handle, meaning, where it approaches (respectively ) along a labeled strand and exits the -handle along the strand with the same label from (respectively ). Note that the tangent vector in to the path varies continuously as we traverse a path as the strands entering and exiting -handles are horizontal.

The rotation number of a path is the number of counterclockwise revolutions around made by the tangent vector to as we transverse . Generally this will be a real number, but will be an integer if and only if is smooth and closed.

Thus, the rotation number is the rotation number of the path in which begins at the base point on the link component and traverses the link component, following the orientation of the component. In the case where is a link with components , we define

Define

If is the resolution of an -diagram of an -component link in Gompf standard form, then the method assigning gradings follows: Choose a Maslov potential that associates an integer modulo to each strand in the tangle associated to , minus cusps and base points, such that the following conditions hold:

  1. for all and all , the strand labeled going through the -th -handle at and the must have the same Maslov potential;

  2. if a strand is oriented to the right, meaning it enters the -handle at and exits at , then the Maslov potential of the strand must be even. Otherwise the Maslov potential of the strand must be odd;

  3. at a cusp, the upper strand (strand with higher -coordinate) has Maslov potential one more than the lower strand.

The Maslov potential is well-defined up to an overall shift by an even integer for knots. (In [Ekholms1s2], Ekholm and Ng give another method for defining the gradings using the rotation numbers of specified paths.)

Set and , where means the Maslov potential of the strand with label going through the -th -handle. It remains to define the grading on crossings in the tangle, crossings resulting from resolving right cusps, and crossings from the half-twists in the resolution. If is crossing of tangle , then let

where is the strand which crosses over the strand at in the -projection of . If is a right cusp, define (assuming there is not a base point in the loop). If is a crossing in one of the half-twists in the resolution where strand crosses over strand (), then

2.6. Differential

It suffices to define the differential on generators and extend by the Leibniz rule. Define . Set on as in §2.3.

In [Ekholms1s2], the DGA on crossings is defined by looking for immersed disks in the -diagrams of Legendrian links, (see the left diagram in Figure 3). However, Ekholm and Ng note that it is equivalent to look for immersed disks in dip versions of the diagram, (see the right diagram in Figure 3). See Figure 4 for the labeling of the crossings in Figure 3.

\labellist\pinlabel

[B] at 207 382 \pinlabel [B] at 207 345 \pinlabel [B] at 207 307 \pinlabel [B] at 207 270

\pinlabel

[B] at 207 117 \pinlabel [B] at 207 81

\pinlabel

[B] at 1016 361 \pinlabel [B] at 1016 331 \pinlabel [B] at 1016 299 \pinlabel [B] at 1016 273

\pinlabel

[B] at 862 109 \pinlabel [B] at 862 85

\pinlabel

[B] at 1454 382 \pinlabel [B] at 1454 345 \pinlabel [B] at 1454 307 \pinlabel [B] at 1454 270

\pinlabel

[B] at 1454 117 \pinlabel [B] at 1454 80

\pinlabel

[B] at 2000 382 \pinlabel [B] at 2000 346 \pinlabel [B] at 2000 307 \pinlabel [B] at 2000 271

\pinlabel

[B] at 2000 118 \pinlabel [B] at 2000 79 \endlabellist

Figure 3. The left gives a Legendrian -diagram of a link in which has resulted from the resolution of a link in Gompf standard form. The right gives the dipped version of the link where the half of a dip on the left side of the dipped version is identified with the right half of the dip on the right side. See Figure 4 for the labeling of the crossings in the dips.
\labellist\pinlabel

[r] at 0 99 \pinlabel [r] at 0 132 \pinlabel [r] at 0 164 \pinlabel [r] at 0 198

\pinlabel

[tr] at 100 99 \pinlabel [tr] at 100 67 \pinlabel [tr] at 100 35 \pinlabel [tr] at 68 99 \pinlabel [tr] at 68 67 \pinlabel [tr] at 37 99

\pinlabel

[tl] at 147 99 \pinlabel [tl] at 147 67 \pinlabel [tl] at 147 35 \pinlabel [tl] at 179 99 \pinlabel [tl] at 179 67 \pinlabel [tl] at 212 99 \endlabellist

Figure 4. This is the dip at the right of the right figure in Figure 3 with strands and crossings labeled. The labels of the partial dip at the left of the right figure in Figure 3 are the same as the right half of the dip depicted.
Definition 2.4.

Let be generators. Define to be the set of orientation-preserving immersions

(up to smooth reparametrization) that map to the dip version of such that

  1. is a smooth immersion except at ,

  2. are encountered as one traverses counterclockwise,

  3. near , covers exactly one quadrant, specifically, a quadrant with positive Reed sign near and a quadrant with negative Reeb sign near , where the Reeb sign of a quadrant near a crossing is defined as in Figure 5.

To each immersed disk, we can assign a word in by starting with the first corner where the quadrant covered has negative Reeb sign, , and listing the crossing labels of all negative corners as encountered while following the boundary of the immersed polygon counterclockwise, . We associate an orientation sign to each quadrant in the neighborhood of a crossing , defined in Figure 5, and use these to define the sign of a disk to be the product of the orientation signs over all the corners of the disk. We denote this sign by . In many cases there is a unique disk with positive corner at (with respect to Reeb sign) and negative corners at and in these we define to be the sign of the unique disk. (In exceptional cases there may be more than one disk with positive corner at and negative corners at .)

\labellist\pinlabel

[b] at 56 57 \pinlabel [t] at 56 57 \pinlabel [l] at 56 57 \pinlabel [r] at 56 57

\pinlabel

[br] at 248 55 \pinlabel [tl] at 248 55 \pinlabel [bl] at 248 55 \pinlabel [tr] at 248 55 \endlabellist

Figure 5. The signs in the figure give the Reeb signs of the quadrants around the crossings. The orientation signs are for all quadrants of crossings of odd degree. For crossings of even degree, we use the convention indicated in the left figure if the crossing comes from the -projection and the convention in the right figure if the crossing is in a dip, which will be discussed in §2.10, where the shaded quadrants have orientation sign and the other quadrants have orientation sign .

Define or to be the signed count of the number of times one encounters the base point while following counterclockwise, where the sign is positive if we encounter the base point while following the orientation of the link component and negative if we encounter the base point while going against the orientation.

We define

and extend to by the Leibniz rule.

In [Ekholms1s2], Ekholm and Ng prove the map has degree and is a differential, .

\labellist\pinlabel

[B] at 174 284 \pinlabel [B] at 174 260 \pinlabel [B] at 174 235 \pinlabel [B] at 174 211

\pinlabel

[B] at 174 115 \pinlabel [B] at 174 81

\pinlabel

[B] at 779 284 \pinlabel [B] at 779 260 \pinlabel [B] at 779 235 \pinlabel [B] at 779 211

\pinlabel

[B] at 779 115 \pinlabel [B] at 779 81

\pinlabel

[B] at 1189 383 \pinlabel [B] at 1189 346 \pinlabel [B] at 1189 307 \pinlabel [B] at 1189 270

\pinlabel

[B] at 1189 117 \pinlabel [B] at 1189 81

\pinlabel

[B] at 1727 383 \pinlabel [B] at 1727 345 \pinlabel [B] at 1727 307 \pinlabel [B] at 1727 270

\pinlabel

[B] at 1727 118 \pinlabel [B] at 1727 79

\pinlabel

[b] at 1252 399 \pinlabel [b] at 1250 302 \pinlabel [t] at 1249 252

\pinlabel

[b] at 1400 388 \pinlabel [b] at 1509 388 \pinlabel [b] at 1441 281 \pinlabel [b] at 1570 364 \pinlabel [b] at 1600 401 \pinlabel [b] at 1408 187 \pinlabel [b] at 1448 205 \pinlabel [t] at 1439 169 \pinlabel [b] at 1481 190

\endlabellist
Figure 6. The left gives a Legendrian -diagram in in Gompf standard form. The right gives the dip form of the normal form. Recall the labels on the crossings in the dips from Figure 4 for the top -handle and label the left crossing and the right in the dip of the bottom -handle.
Example 2.5.

The following is the definition of the DGA for the Legendrian link in Figure 6. Here is generated by over . We set for . Define a Maslov potential on the strands near by

Then we have the following gradings: , , ,

where is the crossing of the strands in the bottom -handle. Since , we know for .

For ease of notation, we will use to denote . We then have the following differentials: