A unification of the ADO and colored Jones polynomials of a knot
aa unification of the ado and coloredjones polynomials of a knot
Sonny Willetts
Abstract
In this paper we prove that the family of colored Jones polynomials of a knotin S determines the family of ADO polynomials of this knot. More precisely,we construct a two variables knot invariant unifying both the ADO and thecolored Jones polynomials. On one hand, the first variable q can be evaluatedat r roots of unity with r ∈ N ∗ and we obtain the ADO polynomial over theAlexander polynomial. On the other hand, the second variable A evaluatedat A = q n gives the colored Jones polynomials. From this, we exhibit a mapsending, for any knot, the family of colored Jones polynomials to the family ofADO polynomials. As a direct application of this fact, we will prove that everyADO polynomial is holonomic and is annihilated by the same polynomial as ofthe colored Jones function. The construction of the unified invariant will usecompletions of rings and algebra. We will also show how to recover our invariantfrom Habiro’s quantum sl completion studied in [12]. Main results:
In [1], Akutsu, Deguchi and Ohtsuki gave a generalisation of the Alexander poly-nomial, building a colored link invariant at each root of unity. These ADO invariants,also known as colored Alexander’s polynomials, can be obtained as the action on 1-1tangles of the usual ribbon functor on some representation category of a version ofquantum sl at roots of unity (see [4], [8]). On the other hand, we have the coloredJones polynomials, a family of invariants obtained by taking the usual ribbon functorof quantum sl on finite dimensional representations. It is known ([5]) that given theADO polynomials of a knot, one can recover the colored Jones polynomials of thisknot. One of the results of the present paper is to show the other way around: giventhe Jones polynomials of a knot, one can recover the ADO polynomials of this knot.We denote ADO r ( A, K ) the ADO invariant at r root of unity seen as a polynomialin the variable A , J n ( q, K ) the n -th colored Jones polynomial in the variable q and A K ( A ) the Alexander polynomial in the variable A . Result 1.
There is a well defined map such that for any knot K in S , { J n ( q, K ) } n ∈ N ∗ (cid:55)→ { ADO r ( A, K ) } r ∈ N ∗ , (Detailed version: Theorem 62). The above result is a consequence of the construction of a unified knot invariantcontaining both the ADO polynomials and the colored Jones polynomials of the knot.Briefly put, we get it by looking at the action of the universal invariant (see [13] [14],and also [16]) on some Verma module with coefficients in some ring completion. Forthe sake of simplicity let’s state the result for framed knots.1 a r X i v : . [ m a t h . QA ] M a y unification of the ado and colored jones polynomials of a knot Result 2.
In some ring completion of Z [ q ± , A ± ] equipped with suitable evalua-tion maps, for any -framed knot K in S , there exists a well defined knot invariant F ∞ ( q, A, K ) such that: F ∞ ( ζ r , A, K ) = ADO r ( A, K ) A K ( A r ) , F ∞ ( q, q n , K ) = J n ( q , K ) (Detailed version: Theorem 57 and Corollary 53). A visual representation of the relationship between all these invariants is given atFigure 1.Lets denote J • ( q , K ) = { J n ( q, K ) } n ∈ N ∗ and call it colored Jones function of K .The holonomy of the unified invariant and of the ADO polynomials will follow as asimple application of the two previous results and of the q -holonomy of the coloredJones function as shown in [7]. Mainly, there are two operators Q and E on the set ofdiscrete function over Z [ q ± ] that forms a quantum plane and for any knot K , thereis a two variable polynomial α K such that α K ( Q, E ) J • ( q , K ) = 0 . We say that thecolored Jones function is q -holonomic.This paper gives a proof that the same polynomial α K , in some similar operators as Q and E , annihilates the unified invariant F ∞ ( q, A, K ) and, at roots of unity, annihilates ADO r ( A, K ) . Result 3.
For any -framed knot K and any r ∈ N ∗ : • The unified invariant F ∞ ( q, A, K ) is q -holonomic. • The ADO invariant
ADO r ( A, K ) is ζ r -holonomic.Moreover they are annihilated by the same polynomial as of the colored Jones function.(Detailed version: Theorems 64 and 66).Remark . Keep in mind that these results cover only the case of a knot K in S . Q ( K ) Quantum algebra
Universal invariant F ∞ ( q, A, K ) Unified invariant Z [ q ± , A ± ] completion(Variables: q and A ) J n ( q , K ) Z [ q ± ] completion(Variable: q ) ADO r ( A, K ) A K ( A r ) Z [ ζ r ][ A ± ] completion(Variable: A ) Colored Jones polynomialsADO polynomials q = ζ r A = q n Verma Prop 61 [5] Cor 15
Figure 1: Visual representation of the unified knot invariant.
Summary of the paper:
A way to build a unified element for ADO invariants is to do it by hand. First,one can explicit a formula for the ADO invariant at a r root of unity by decomposingit as a sum of what we will call state diagrams. This explicit formula will allow us tosee what are the obstructions to unify the invariants: first, it will depends on the root2 .willetts of unity ζ r , secondly the range of the sums coming from the action of the truncated R matrix depends on the order r of the root of unity. The first obstruction is easyto overcome since taking a formal variable q instead of each occurrence of ζ r will dothe trick. But for the second one, one could ask that the ranges go to infinity, andthis will bring some convergence issues. A way to make these sums convergent is touse a completion of the ring Z [ q ± , A ± ] denoted ˆ R ˆ I , this will allows us to define agood candidate for the unification.But then, we will have to check that this element contains the ADO invariants. Wewill show that at each root of unity of order r with r ∈ N ∗ , one can define anevaluation map that evaluates q in ζ r , and that the result can be factorized into aproduct of an invertible element of the complete ring and the ADO invariant.So we’ll get an element containing ADO invariants, but the way we built this elementdepends on the chosen diagram of the knot. A way to prove that this element is reallya knot invariant itself is to recover it with a more advance machinery: the universalinvariant of a knot. The universal invariant of a knot was introduced in [13] and[14], the construction can also be found in [16]. It is a knot invariant and an elementof the h -adic version of quantum sl , we will use this fact to construct an integralsubalgebra in which the universal invariant of a -framed knot will lie. The integralityof the subalgebra will allow us to build a Verma module of it whose coefficients willlie in ˆ R ˆ I , and on which the scalar action of the universal invariant gives our unifiedelement. A corollary will be that the unified element is a knot invariant.Completions were studied by Habiro in [12]. For the sole purpose of the factorizationat roots of unity, we had to use a different completion than the ones mentioned in[12]. But as we will see, we can also recover our unified invariant from his algebraicsetup.Once we have this connection between quantum sl and this unified invariant, we canhenceforth relate it also to the colored Jones polynomials, this will allow us to usethe Melvin Morton Rozansky conjecture proved by Bar-Natan and Garoufalidis in [2]in order to get some information on the factorization at roots of unity: briefly put,the unified invariant factorize at root of unity as ADO polynomial over the Alexanderpolynomial.Now we have a unified invariant for both the ADO polynomials and the colored Jonespolynomials, the maps recovering them are also well understood. This will allowus to prove that, given the colored Jones polynomials, one may recover the ADOpolynomials.From the fact that the colored Jones polynomials recovers the unified invariant andfrom the factorisation at roots of unity, we will prove that the unified invariant andADO polynomials follow the same holonomic rule as of the colored Jones function(see [7]). In the same time this paper was made, Brown, Dimofte and Geer got amore general result covering the case of links in their Theorem 4.3 [3].We will also see that the unified invariant is an integral version of the h -adic loopexpansion of the colored Jones function and remark that even if it is not clear ingeneral if it’s a power serie, it has similar properties as the power serie invariantconjectured by Gukov and Manolescu in [9] (Conjectures 1.5, 1.6).Finally we will give some computations of the unified invariant and its factorizationat roots of unity, showing how the inverse of the Alexander polynomial appears. Acknowledgments:
I would like to thank my Ph.D. advisors François Costantinoand Bertrand Patureau-Mirand for their help and guidance.I would also like to thank Christian Blanchet and Anna Beliakova for their usefulremarks. 3 unification of the ado and colored jones polynomials of a knot
In this article, any knots and links are in S and supposed oriented and framed. We will expose in this section how to obtain ADO invariants for links [1], also calledcolored Alexander’s polynomials, from a non semi-simple category of module over anunrolled version of U q ( sl ) . A more detailed and thorough construction can be foundin [4] [8].For any variable q , we denote { n } = q n − q − n , [ n ] = { n }{ } , { n } ! = (cid:81) ni =1 { i } , [ n ]! = (cid:81) ni =1 [ i ] , (cid:2) nk (cid:3) q = [ n ]![ n − k ]![ k ]! .In order to define ADO invariants for knots, and for the sake of simplicity, in thissection q will be an even root of unity. Definition 2.
Let q = e iπr = ζ r root of unity.We work with an "unrolled" version of U ζ r ( sl ) denoted by U Hζ r ( sl ) and defined asfollow:Generators: E, F, K, K − , H Relations: KK − = K − K = 1 KE = ζ r EK KF = ζ − r F K [ E, F ] = K − K − ζ r − ζ − r KH = HK [ H, E ] = 2 E [ H, F ] = − F E r = F r = 0 This algebra has a Hopf algebra structure: ∆( E ) = 1 ⊗ E + E ⊗ K (cid:15) ( E ) = 0 S ( E ) = − EK − ∆( F ) = K − ⊗ F + F ⊗ (cid:15) ( F ) = 0 S ( E ) = − KF ∆( H ) = 1 ⊗ H + H ⊗ (cid:15) ( H ) = 0 S ( H ) = − H ∆( K ) = K ⊗ K (cid:15) ( K ) = 1 S ( K ) = K − ∆( K − ) = K − ⊗ K − (cid:15) ( K − ) = 1 S ( K − ) = K Now, we can look at some category of finite dimensional representation of thisalgebra and endow it with a ribbon structure.
Definition 3.
Let
Rep be the category of finite dimensional U Hζ r ( sl ) -modules suchthat:1. The action of H is diagonalizable.2. The action of K and of ζ H r are the same. Proposition 4.
The irreducible representations of U Hζ r ( sl ) are V α for α ∈ ( C − Z ) ∪ r Z and S i for i ∈ { , . . . , r − } where S i is the highest weight module of weight i anddimension i + 1 , V α is the highest weight module of weight α + r − and dimension r . Definition 5. In V α , we say that x has weight level n if Kx = ζ α + r − − n r x .We can endow Rep with a ribbon structure by giving the action of a R -matrixand a ribbon element.We set F ( n ) = { } n F n [ n ]! for ≤ n < r − .4 .willetts Proposition 6. R = ζ H ⊗ H r r − (cid:80) n =0 ζ n ( n − r E n ⊗ F ( n ) is an R -matrix whose action is welldefined on Rep and it’s inverse is R − = ( r − (cid:80) n =0 ( − n ζ − n ( n − r E n ⊗ F ( n ) ) ζ − H ⊗ H r . Proposition 7. K − r is a pivotal element for U Hζ r ( sl ) compatible with the braiding. We can now take the usual ribbon functor RT in order to obtain a link invariant,but on the V α it will be (since the quantum trace is ). Hence we need to be moresubtle in order to retrieve some information.On irreducible representations, a 1-1 tangle can be seen as a scalar:If L is a link obtain by closure of a 1-1 tangle T , we denote RT ( T ) v = ADO ( T ) v where v is a highest weight vector. Notice that it depends on the 1-1 tangle T chosen.In order to have a link invariant, we must multiply it by a "modified trace".We denote { α } ζ r = ζ α r − ζ α r , { α + k } ζ r = ζ α + k r − ζ − α − k r , { α ; n } ζ r = (cid:81) n − i =0 { α − i } ζ r . Proposition 8. If L a link and T is any 1-1 tangle whose closure is L such that theopen component is colored with V α , set d ( α ) = { α }{ rα } , then ADO (cid:48) r ( L ) := d ( α ) ADO r ( T ) is a link invariant. Although we don’t have to specify α and obtain a polynomial in q α , we cannotdo the same for q . The root of unity q must be fixed in order to define the invariant,hence it is a natural question for one to ask how such invariants behave when the rootof unity changes. From now on we will only work with knots.To see how the ADO polynomials behave when the root of unity changes, we willexplicit a formula for the invariant using the ribbon functor on a diagram D of aknot.Let K a knot colored by V α − r +1 and T a 1-1 tangle whose closure is K , since weare working with knots ADO r ( A, K ) := ADO r ( T ) is well defined. Where A is thefree variable ζ α r .Let’s study this element, by choosing a basis of V α − r +1 and computing the invari-ant with state diagrams. Remark . V α − r +1 is generated by v , v , . . . , v r − where v is a highest weight vector,and v i = F ( i ) v { α ; i } ζ r . Proposition 10. Ev = 0 Ev i = v i − F v i = [ i + 1][ α − i ] v i +1 F ( k ) v i = (cid:2) k + ik (cid:3) ζ r { α − i ; k } ζ r v k + i Kv i = ζ α − i r v i ζ H ⊗ H r v i ⊗ v j = ζ α r ζ − ( i + j ) α r ζ ij r v i ⊗ v j Corollary 11.
We have
ADO r ( A, K ) ∈ ζ fα r Z [ ζ r , ζ ± α r ] where f is the framing ofthe knot. unification of the ado and colored jones polynomials of a knot a k b k b k − i k a k + i k (a) Positive crossing. a k b k b k − i k a k + i k (b) Negative crossing. Figure 2: The two possibilities for the k-th crossing in D .More precisely, in order to calculate a useful form of this invariant one may look at state diagram of a knot. For any knot seen as a (1 , tangle, take a diagram D , labelthe top and bottom strands and starting from the bottom strand, label the strandafter the k -th crossing encountered with the rule described in Figure 2. The resultingdiagram is called a state diagram of D .Let K a knot and D a diagram of the knot seen as a (1 , tangle. Suppose thediagram has N crossings. Now for any state diagram of D we can associate an element: D r ( i , . . . , i N ) =( S (cid:89) j =1 ζ ± ( r − α − (cid:15) j )2 r ) (cid:89) k ∈ pos ζ ik ( ik − r (cid:20) a k + i k i k (cid:21) ζ r { α − a k ; i k } ζ r × ζ − ( a k + b k ) α r ζ a k + i k )( b k − i k )2 r (cid:89) k ∈ neg ( − i k ζ − ik ( ik − r × (cid:20) a k + i k i k (cid:21) ζ r { α − a k ; i k } ζ r ζ ( a k + b k ) α r ζ − a k b k r where neg ∪ pos = [ | , N | ] and k ∈ pos if the k -th crossing of D is positive, else k ∈ neg . a k , b k are the strands labels at the k -th crossing of the state diagram (seeFigure 2), S is the number of + appearing in the diagram, and (cid:15) j the strandlabel at the j-th or , the ± sign is positive if and negative if . Remark . Note that the a k and b k appearing are defined in terms of i j . You canfind some examples of state diagrams in Section 5 Figure 7a, 7b, 8a, 8b. Proposition 13. If D is a diagram of K seen as a a 1-1 tangle we have: ADO r ( A, K ) = ζ fα r r − (cid:88) i =0 D r ( i , . . . , i N )= ζ fα r r − (cid:88) i =0 ( S (cid:89) j =1 ζ ± ( r − α − (cid:15) j )2 r ) (cid:89) k ∈ pos ζ ik ( ik − r (cid:20) a k + i k i k (cid:21) ζ r × { α − a k ; i k } ζ r ζ − ( a k + b k ) α r ζ a k + i k )( b k − i k )2 r (cid:89) k ∈ neg ( − i k × ζ − ik ( ik − r (cid:20) a k + i k i k (cid:21) ζ r { α − a k ; i k } ζ r ζ ( a k + b k ) α r ζ − a k b k r where i = ( i , . . . , i N ) , N is the number of crossings, S the number of + and f is the framing of the knot. .willetts Proof.
Notice that ζ fα r D r ( i , . . . , i N ) is the element obtained by adding to the k -thcrossing a coupon labeled with: q H ⊗ H q ik ( ik − E i k ⊗ F ( i k ) if positive and q − H ⊗ H q − ik ( ik − E i k ⊗ F ( i k ) if negative. Then add a coupon to labeled K r − andlabeled K − r . We get an element of U Hζ r ( sl ) , its action on v ∈ V α + r − thehighest weight vector gives the element ζ fα r D r ( i , . . . , i N ) . Summing them over i k for all k gives the ADO polynomial.Now that we have an explicit formula, can we construct from it a suitable elementthat can be evaluated at roots of unity and recover the ADO invariants?We have two main issues here, first of all, ζ r appears in the formula, we will haveto replace each occurrence with some variable q , in order to see it as a polynomial ora formal series.The second one is more difficult to solve: the action of the R matrices makesappear sums that range to r − , which depends on the order of the root of unity.A solution to this problem, as we will explicit it, is to let the sum range to infinityand define a ring in which such sums converge. Then we will see how to factorize theADO invariant from this new unified form. The approach here will be to unify the invariants: using completions of rings andalgebras, we will explicit an integral invariant in some variable q that can be evaluatedat any root of unity, recovering ADO invariants defined previously.The first subsection will create the right setup to define a unified form inspired bythe useful form of the ADO invariant in Proposition 13. Using a completion of thering of integral Laurent polynomials in two variables q, A , we define a unified form F ∞ ( q, A, D ) by taking the previous form of ADO, replacing the root of unity ζ r by q , ζ α r by A and letting the truncated sums coming from the R -matrices action go toinfinity. Note that at this point, the defined form is not a knot invariant, as it a priori depends on the diagram D of the knot.The second subsection will make the bridge between the first two sections. Byevaluating the unified form at roots of unity ζ r with r ∈ N ∗ , we factor out theADO invariant. We will then explicit a map sending the unified form of a knot to thecorresponding ADO invariants. This will show that the ADO invariants are containedin the unified form and that we can recover them from it. Let’s lay the groundwork for an unified form to exist. It must be a ring in whichinfinite sums previously mentioned converge.Let R = Z [ q ± , A ± ] , we will construct a completion of that ring. For the sake ofsimplicity, we will denote q α := A and use previous notation for quantum numbers.Keep in mind that, here, α is just a notation, not a complex number.We denote { α } q = q α − q α , { α + k } q = q α + k − q − α − k , { α ; n } q = (cid:81) n − i =0 { α − i } q . Definition 14.
Let I n be the ideal of R generated by the following set { { α + l ; n } q , l ∈ Z } . 7 unification of the ado and colored jones polynomials of a knot Lemma 15. I n is generated by elements of the form { n ; i }{ α ; n − i } , i ∈ { , . . . n } .Proof. The proof can be found in Habiro’s article [12]:Replacing K (resp. K − ) by q α (resp. q − α ) in Proposition 5.1, one gets the proof ofthis lemma.We then have a projective system : ˆ I : I ⊃ I ⊃ · · · ⊃ I n ⊃ . . . From which we can define the completion of R , taking the projective limit: Definition 16.
Let ˆ R ˆ I = lim ← n RI n = { ( a n ) n ∈ N ∗ ∈ (cid:81) ∞ i =1 RI n | ρ n ( a n +1 ) = a n } where ρ n : RI n +1 → RI n is the projection map.This completion is a bigger ring containing R : Proposition 17.
The canonical projection maps induce an injective map
R (cid:44) −→ ˆ R ˆ I Proof.
It is sufficient to prove that (cid:84) n ∈ N ∗ I n = { } .Since R = Z [ q ± ][ A ± ] , it is a Laurent polynomial ring. Let’s define deg q ( x ) , val q ( x ) the degree and valuation of x in the variable q .Let f k : Z [ q ± , A ± ] → Z [ q ± ] , A (cid:55)→ q k . We have f k ( I n ) ⊂ { n } ! Z [ q ± ] because I n is generated by elements of the form { n ; i }{ α ; n − i } that maps to { n ; i }{ k ; n − i } ,which is divisible by { n } ! .Hence if x ∈ (cid:84) n ∈ N ∗ I n , f k ( x ) ∈ { n } ! Z [ q ± ] for all n , since Z [ q ± ] is factorial, f k ( x ) =0 for all k .Take x ∈ (cid:84) n ∈ N ∗ I n , written x = (cid:80) a n A n with a n ∈ Z [ q ± ] . Take N such that deg q ( x ) < N and val q ( x ) > − N .This implies that deg q ( a n ) < N and val q ( a n ) > − N (since it is the case for x and anyhigher or lower terms could not compensate since the power of A is different beforeeach a n ).Thus since f N ( x ) = 0 , (cid:80) a n q Nn = 0 , we have deg q ( a n q Nn ) < N (1 + 2 n ) and val q ( a n q Nn ) > N (2 n − , then each terms a n q Nn must be . Hence a n = 0 for all n , meaning that x = 0 . Remark . If b ∈ R and b n ∈ I n − for n ≥ , the partial sums N (cid:80) i =0 b n converges in ˆ R ˆ I as N goes to infinity.We denote the limit + ∞ (cid:80) i =0 b n := ( N (cid:80) i =0 b n ) N ∈ N ∗ .Conversely, if a = ( a N ) N ∈ N ∗ ∈ ˆ R ˆ I , let a n ∈ R be any representative of a n in R , then a = + ∞ (cid:80) i =0 b n where b = a and b n = a n +1 − a n for n ∈ N ∗ .8 .willetts We proceed similarly as in the paragraph preceding Proposition 13. Let K be aknot seen as a (1 , tangle and D a diagram of it. For a state diagram of D we define: D ( i , . . . , i N ) =( S (cid:89) j =1 q ∓ ( α − (cid:15) j ) ) (cid:89) k ∈ pos q ik ( ik − (cid:20) a k + i k i k (cid:21) q { α − a k ; i k } q × q − ( a k + b k ) α q a k + i k )( b k − i k ) (cid:89) k ∈ neg ( − i k q − ik ( ik − (cid:20) a k + i k i k (cid:21) q × { α − a k ; i k } q q ( a k + b k ) α q − a k b k where neg ∪ pos = [ | , N | ] and k ∈ pos if the k -th crossing of D is positive, else k ∈ neg . a k , b k are the strands labels at the k -th crossing of the state diagram (seeFigure 2), S is the number of + appearing in the diagram, and (cid:15) j the strandlabel at the j-th or , the ∓ sign is negative if and positive if . Remark . Note that the a k and b k appearing are defined in terms of i j . As men-tioned previously, you can find some examples of state diagrams in Section 5 Figure7a, 7b, 8a, 8b. Definition 20.
Let K a knot and T K . Let D be adiagram of T .We define: F ∞ ( q, A, D ) := q fα + ∞ (cid:88) i =0 D ( i , . . . , i N )= q fα + ∞ (cid:88) i =0 ( S (cid:89) j =1 q ∓ ( α − (cid:15) j ) ) (cid:89) k ∈ pos q ik ( ik − (cid:20) a k + i k i k (cid:21) q { α − a k ; i k } q × q − ( a k + b k ) α q a k + i k )( b k − i k ) (cid:89) k ∈ neg ( − i k q − ik ( ik − (cid:20) a k + i k i k (cid:21) q × { α − a k ; i k } q q ( a k + b k ) α q − a k b k where i = ( i , . . . , i N ) , N is the number of crossings, S the number of + and f is the framing of the knot.We have that F ∞ ( q, A, D ) is a well defined element of q fα ˆ R ˆ I .Note that it is not clear that this element is a knot invariant, it could depend apriori on the diagram D and we will have to prove later that it does not. In this subsection we will see how to evaluate at a root of unity an element of ˆ R ˆ I . Wewill first need some useful lemma.Let r any integer, R r = Z [ ζ r , A ± ] , we use the same previous notations and ζ α r := A . Lemma 21.
For any k , { α − k ; r } ζ r = ( − k ζ − r ( r − r { rα } ζ r .Proof. { α − k ; r } = { α − k } . . . { α − k − r + 1 } = ( − { α − k + 1 } . . . { α − k − r + 2 } = ( − k { α } . . . { α − r + 1 } = ( − k { α ; r } unification of the ado and colored jones polynomials of a knot { α ; r } = (cid:81) r − j =0 ( ζ α − j r − ζ − α + j r )= ζ − r ( r − r ζ − rα r (cid:81) r − j =0 ( ζ α r − ζ j r )= ζ − r ( r − r ζ − rα r ( ζ rα r − ζ − r ( r − r { rα } where the fourth equality is just developing the factorized form of X r − at X = ζ α r .Let I = { rα } ζ r R r and we build the I -adic completion of R r : Definition 22. ˆ R Ir = lim ← n R r I n = { ( a n ) n ∈ N ∗ ∈ (cid:81) ∞ i =1 R r I n | ρ (cid:48) n ( a n +1 ) = a n } where ρ (cid:48) n : R r I n +1 → R r I n is the projection map.This completion is a bigger ring containing R r : Proposition 23.
The canonical projection maps induce an injective map R r (cid:44) −→ ˆ R Ir Proof.
It is sufficient to prove that (cid:84) n ∈ N ∗ I n = { } .Since R r = Z [ ζ r ][ A ± ] , it is a Laurent polynomial ring. Hence, any non zero element x can be uniquely written x = (cid:80) ni = l a n A n where a k ∈ Z [ ζ r ] , ∀ k ∈ { l, l +1 , . . . , n − , n } and a n , a l (cid:54) = 0 .Let’s define len ( x ) = n − l . We have that len ( xy ) = len ( x ) + len ( y ) .Thus if x ∈ (cid:84) n ∈ N ∗ I n is non zero, of length n , ∃ y ∈ R r such that x = { rα } n y hence len ( x ) = 2 rn + len ( y ) , contradiction.Let’s now define the evaluation map from ˆ R ˆ I to ˆ R Ir .At the level of R and R r we have a well defined evaluation map, ev ζ r : R → R r , q (cid:55)→ ζ r . We will extend this map to the completions. Proposition 24. ev ζ r ( I rn ) = I n Proof.
Direct application of Lemma 21.Hence, ev r factorize into maps ψ n : R/I rn → R r /I n , we can then define the mapextension: Proposition 25.
We have a well defined map: ev r : ˆ R ˆ I → ˆ R Ir such that, if ( a n ) n ∈ N ∗ ∈ ˆ R ˆ I , ev r (( a n ) n ∈ N ∗ ) = ( ψ n ( a rn )) n ∈ N ∗ .Proof. If we denote λ n : R/I r ( n +1) → R/I rn the projective maps, it lie on the factthat the following diagram is commutative: R/I r ( n +1) R r /I n +1 R/I rn R r /I n λ n ψ n +1 ψ n ρ (cid:48) n .willetts It is now time to study the element F ∞ ( ζ r , A, D ) := ev r ( F ∞ ( q, A, D )) , we willsee that the ADO invariant ADO r ( A, K ) can be factorized from it.In order to do so, we will need some useful computations: Lemma 26.
We have the following factorizations: • ζ ( i + rl )( i + rl − r = ( − il ζ rl ( rl − r ζ i ( i − r , • { α − a − ru ; i + rl } ζ r = ( − al + rul + ui + li ζ − rl ( r − r ζ − rl ( l − r { rα } lζ r { α − a ; i } ζ r , • (cid:2) a + i + r ( u + l ) i + rl (cid:3) ζ r = ( − al + rul + ui (cid:0) u + ll (cid:1)(cid:2) a + ii (cid:3) ζ r • ζ − rl ( r − r ζ − rl ( l − r = ζ − rl ( rl − r Proof.
1. The first dot is obtained by developing the product.2. The second dot is an application of Lemma 21.First { α − a − ru ; i + rl } = ζ ( i + rl ) ru r { α − a ; i + rl } = ( − iu ( − rul { α − a ; i + rl } .Then, { α − a ; i + rl } = { α − a ; rl }{ α − a − rl ; i } = ( − al ζ − rl ( l − r { α ; rl }{ α − a − rl ; i } = ( − al ζ − rl ( l − r ζ − rl ( r − r { rα } l { α − a − rl ; i } Finally { α − a − rl ; i } = ( − li { α − a ; i } .Put together, we get { α − a − ru ; i + rl } ζ r = ( − al + rul + ui + li ζ − rl ( r − r ζ − rl ( l − r { rα } lζ r × { α − a ; i } ζ r .
3. The third dot follows from the fact that ev ζ r ( { rk } q { r } q ) = ( − − k k . In (cid:2) a + i + r ( u + l ) i + rl (cid:3) ζ r seen as { a + i + r ( u + l ) } ! { a + ru } ! { i + rl } ! , taking only the terms { rk } , we extract ( − ul (cid:0) u + ll (cid:1) . Nowwe only have to deal with non multiples of quantum r . We use the equal-ity { t + r } = ( − { t } in order to have consecutive terms in the denominators(excepted from multiple of r ), indeed { a + ru } ! = { ru } ! { a + ru ; a } and { i + rl } )( − u ( i + rl ) { i + rl + ru ; i + rl } , hence { a + i + r ( u + l ) } ! { a + ru } ! { i + rl } ! = ( − u ( i + rl ) { a + i + r ( u + l ); a }{ a + ru ; a } =( − u ( i + rl ) ( − au ( − a ( u + l ) { a + i ; a }{ a ; a } = ( − ui ( − rul ( − al (cid:2) a + ii (cid:3) ζ r .Putting things together with the quantum r multiple part, we get the desiredresult.4. The last dot is obtained as follow: ζ − rl ( rl − r = rl − (cid:89) k =0 ζ − k r = l (cid:89) j =0 r − (cid:89) k =0 ζ − k − rj r = l (cid:89) j =0 ζ − rj r r − (cid:89) k =0 ζ − k r = ζ − rl ( r − r ζ − rl ( l − r . unification of the ado and colored jones polynomials of a knot We proceed similarly as in the paragraph preceding Definition 20 and define anelement to each state diagram of D that will be used to factorise F ∞ ( q, A, D ) . Let K be a knot seen as a (1 , tangle and D a diagram of it. For a state diagram of D wedefine: D C,r ( l , . . . , l N ) =( S (cid:89) j =1 ζ ∓ rα r ) (cid:89) k ∈ pos (cid:18) u k + l k l k (cid:19) { rα } l k ζ r ζ − ( u k + v k ) rα r × (cid:89) k ∈ neg ( − l k (cid:18) u k + l k l k (cid:19) { rα } l k ζ r ζ ( u k + v k ) rα r where neg ∪ pos = [ | , N | ] and k ∈ pos if the k -th crossing of D is positive, else k ∈ neg , a k , b k ∈ [ | , . . . , r − | ] , a k + ru k , b k + rv k are the strands labels at the k -thcrossing of the state diagram (see Figure 3), S is the number of + appearingin the diagram, and (cid:15) j the strand label at the j-th or , the ∓ sign is negativeif and positive if . Proposition 27.
For a knot K and a diagram of the knot D , r ∈ N ∗ , we have the following factorizationin ˆ R Ir : F ∞ ( ζ r , A, D ) = C ∞ ( r, A, D ) × ADO r ( A, K ) where : C ∞ ( r, A, D ) = + ∞ (cid:88) l =0 D C,r ( l , . . . , l N )= + ∞ (cid:88) l =0 ( S (cid:89) j =1 ζ ∓ rα r ) (cid:89) k ∈ pos (cid:18) u k + l k l k (cid:19) { rα } l k ζ r ζ − ( u k + v k ) rα r × (cid:89) k ∈ neg ( − l k (cid:18) u k + l k l k (cid:19) { rα } l k ζ r ζ ( u k + v k ) rα r where l = ( l , . . . , l N ) , N is the number of crossings, S the number of + and f is the framing of the knot.Proof. For the sake of simplicity, we will only consider positive crossings in the fol-12 .willetts lowing proof. We factorize as follows: F ∞ ( ζ r , A, D ) = ζ fα r + ∞ (cid:88) s =0 ( S (cid:89) j =1 ζ ∓ ( α − (cid:15) j )2 r ) N (cid:89) k =1 ζ sk ( sk − r (cid:20) z k + s k s k (cid:21) ζ r × { α − z k ; s k } ζ r ζ ( − z k − y k ) α r ζ z k + s k )( y k − s k )2 r = ζ fα r + ∞ (cid:88) i + rl =0 ( S (cid:89) j =1 ζ ∓ ( α − (cid:15) j )2 r ) N (cid:89) k =1 ζ ( ik + rlk )( ik + rlk − r × (cid:20) a k + i k + r ( u k + l k ) i k + rl k (cid:21) ζ r { α − ( a k + ru k ); i k + rl k } ζ r × ζ ( − ( a k + ru k ) − ( b k + rv k )) α r ζ a k + ru k )+( i k + rl k ))( b k + rv k − ( i k + rl k ))2 r = ζ fα r r − (cid:88) i =0 ( S (cid:89) j =1 ζ ± ( r − α − (cid:15) j )2 r ) N (cid:89) k =1 ζ ik ( ik − r (cid:20) a k + i k i k (cid:21) ζ r × { α − a k ; i k } ζ r ζ ( − a k − b k ) α r ζ a k + i k )( b k − i k )2 r × + ∞ (cid:88) l =0 ( S (cid:89) j =1 ζ ∓ rα r ) N (cid:89) k =1 (cid:18) u k + l k l k (cid:19) { rα } l k ζ r ζ ( − u k − v k ) rα r The second equality is obtained by changing variables s k = i k + rl k ≤ i k ≤ r − and writing the strands labels at crossings z k as z k = a k + ru k ≤ a k ≤ r − and y k as y k = b k + rv k ≤ b k ≤ r − .Note that a k , b k solely depends on i k and u k , b k on l k . This relies on the fact that (cid:2) n + mn (cid:3) q = 0 at q = ζ r if n, m ≤ r − and n + m ≥ r .The third one is obtained by replacing each term with its factorization given by Lemma26, the crossed terms between i k and l k are just signs, that eventually compensate.Hence, we have the factorization. a k + ru k b k + rv k b k + rv k − i k − rl k a k + ru k + i k + rl k (a) Positive crossing. a k + ru k b k + rv k b k + rv k − i k − rl k a k + ru k + i k + rl k (b) Negative crossing. Figure 3: The two possibilities for the k-th crossing in D when factorizing.In order to get back ADO r ( A, K ) from F ∞ ( ζ r , A, D ) , we need to prove that C ∞ ( r, A, D ) is a unit in ˆ R Ir . Proposition 28. If a = ( a n ) n ∈ N ∗ ∈ ˆ R Ir and a ∈ R r /I is a unit, then a is a unit in ˆ R Ir .Proof. Let a = ( a n ) n ∈ N ∗ ∈ ˆ R Ir such that a is a unit of R r /I .Let’s prove that a n is also a unit in R r /I n . Indeed, if y is an element of R r /I n such13 unification of the ado and colored jones polynomials of a knot that a n y = a y = 1 mod I then ∃ z ∈ I.R r /I n such that a n y = 1 + z , z = a n y − thus z n = ( a n y − n , which proves that a n is invertible.Hence, a − = ( a − n ) n ∈ N ∗ is the inverse of a in ˆ R Ir .Since C ∞ ( r, A, D ) = ( (cid:81) Sj =1 ζ ∓ rα r ) mod { rα } ζ r is an invertible element of R r /I ,then C ∞ ( r, A, D ) is a unit of ˆ R Ir . Corollary 29.
ADO r ( A, K ) = F ∞ ( ζ r , A, D ) C ∞ ( r, A, D ) − Finally, one can recover C ∞ ( r, A, D ) with F ∞ ( q, A, D ) , this will prove that notonly that ADO is contained in F ∞ ( q, A, D ) but that it’s possible to extract them withthe sole datum of F ∞ ( q, A, D ) .For r = 1 , one gets ev ( F ∞ ( q, A, D )) = F ∞ ( ζ , A, D )= C ∞ (1 , A, D ) × ADO ( A, K )= q fα C ∞ (1 , A, D ) Remark . Note that
ADO ( A, K ) is only defined as the case r = 1 in Prop 13,which is well defined. Nevertheless, the algebraic setup at Section 2 fails at r = 1 since [ E, F ] is not well defined.But then C ∞ (1 , A, D ) ∈ (cid:92)Z [ A ± ] { α } := lim ← n Z [ A ± ] { α } n , for each r we have a well definedmap: g r : (cid:92)Z [ A ± ] { α } → (cid:92)Z [ A ± ] { rα } , q α (cid:55)→ q rα such that g r ( C ∞ (1 , D )) = C ∞ ( r, A, D ) .This proves the following proposition: Proposition 31.
For all r , we have a well defined map F C r = g r ◦ ev : ˆ R ˆ I → (cid:92)Z [ A ± ] { rα } and for any knot K and any diagram D of the knot, F ∞ ( q, A, D ) (cid:55)→ C ∞ ( r, A, D ) . Corollary 32.
For all r , we have a well defined map ev r × F C r : ( ˆ R ˆ I ) × → ( ˆ R Ir ) × and for any knot K and any diagram D of the knot, F ∞ ( q, A, D ) (cid:55)→ ADO r ( A, K ) .Proof. Let x ∈ ( ˆ R ˆ I ) × an invertible element, since F C r is a ring morphism, F C r ( x ) is invertible. Then Id × F C r ( F ∞ ( q, A, D )) = F ∞ ( ζ r , A, D ) × C ∞ ( r, A, D ) − = ADO r ( A, K ) . We have built by hand an element F ∞ ( q, A, D ) in some completion of a ring, fromwhich we have evaluation maps that recovers the ADO invariants. This element isbuilt from the diagram of a knot, thus it depends a priori on it. In order to provethat this element is indeed a knot invariant, we will see how to obtain it using Hopfalgebra machinery.The first subsection will be dedicated to create an integral subalgebra of the h -adicversion of quantum sl containing the universal invariant of a framed knot.14 .willetts This will allow us to define, in the second subsection, a Verma module on it.Since the algebra previously defined is integral, this will also be the case for theVerma module, whose coefficients will lie in ˆ R ˆ I . The unified form F ∞ ( q, A, D ) willbe seen as the scalar action of the universal invariant on this Verma module. Sincethe universal invariant is a knot invariant, so will be F ∞ ( q, A, D ) .This algebraic setup is made to get back the unified form and prove its invariance,and it is a completion which is very close to that of Habiro’s in [12]. But theyare not the same, and we will see in the third subsection how to connect this workto Habiro’s setup in the article. We will interpret our ring completion ˆ R ˆ I as somesubalgebra completion found in [12], allowing to prove some nice properties on thering structure (integral domain, subring of some h adic ring). Moreover we will showthat the unified invariant can also be recovered from Habiro’s algebraic setup, usingthe same process as in the second subsection, but with his completions.Afterwards, we will see that we can also recover the colored Jones polynomials fromthe unified invariant. First this will allow us to study the factorisation in Proposition27, and find that C ∞ ( r, A, D ) is just the inverse of the Alexander polynomial. Lastly,using the unified invariant as a bridge between the family of colored Jones polynomialsand the family of ADO polynomials, we will show that they are equivalent, meaningthat we can recover one family with the other.As a direct application of this facts, we will show that the unified invariant andevery ADO polynomials follow the same holonomic rule as the colored Jones function(see [7]). In order to build F ∞ ( q, A, D ) from Hopf algebra, we will need some "big enough"integral version quantum sl , but not too big in order to have a ˆ R ˆ I Verma module onit. First let’s define the biggest integral quantum sl , U h . Definition 33.
We set U h := U h ( sl ) the Q [[ h ]] algebra topologically generated by H, E, F and relations [ H, E ] = 2 E, [ H, F ] = − F, [ E, F ] = K − K − q − q − where q = e h and K = q K = e hH .It is endowed with an Hopf algebra structure: ∆( E ) = 1 ⊗ E + E ⊗ K (cid:15) ( E ) = 0 S ( E ) = − EK − ∆( F ) = K − ⊗ F + F ⊗ (cid:15) ( F ) = 0 S ( E ) = − KF ∆( H ) = 1 ⊗ H + H ⊗ (cid:15) ( H ) = 0 S ( H ) = − H And an R -matrix: R = q H ⊗ H ∞ (cid:88) i =0 { } n q n ( n − [ n ]! E n ⊗ F n R − = ∞ (cid:88) i =0 ( − n { } n q − n ( n − [ n ]! E n ⊗ F n q − H ⊗ H Altogether with a ribbon element: K − u where u = (cid:80) S ( β ) α if R = (cid:80) α ⊗ β .15 unification of the ado and colored jones polynomials of a knot Hence, if K is a knot and T a 1-1 tangle whose closure is K .We set Q U h ( K ) ∈ U h the universal invariant associated to T in U h . The definition ofthis element is given in Ohtsuki’s book [16] subsection 4.2.It is a knot invariant.Let us now build a suitable subalgebra of U h and a ˆ R ˆ I Verma module on it. Wewill then see that the universal invariant is in some extent inside the subalgebra andits scalar action on the Verma module will give us F ∞ ( q, A, D ) .The subalgebra considered is an integral version of U q ( sl ) defined by: Definition 34.
Let U := U Dq ( sl ) the Z [ q ± ] subalgebra of U h generated by E, F ( n ) , K where F ( n ) = { } n F n [ n ]! .It inherits the Hopf Algebra structure from U h . Remark . The R -matrix is not an element of U but we have: R = q H ⊗ H ∞ (cid:88) i =0 { } n q n ( n − [ n ]! E n ⊗ F n = q H ⊗ H ∞ (cid:88) i =0 q n ( n − E n ⊗ F ( n ) Hence, aside from q H ⊗ H (that we can control in the universal invariant as wewill see further on), we need the convergence of ∞ (cid:80) i =0 q n ( n − E n ⊗ F ( n ) in some tensorproduct of the algebra with itself.Thus we need to complete the algebra U .We denote { H + m } q = Kq m − K − m q − m , { H + m ; n } q = (cid:81) n − i =0 { H + m − i } q . Definition 36.
Let L n be the Z [ q ± ] ideal generated by { n } ! .Let J n be the U two sided ideal generated by the following elements: F ( i ) { H + m ; n − i } q where m ∈ Z and i ∈ { , . . . , n } . Lemma 37. J n is finitely generated by elements of the form F ( i ) { n − i ; j }{ H ; n − i − j } , j ∈ { , . . . , n − i } , i ∈ { , . . . , n } .Proof. The proof can be found in Habiro’s article [12], Proposition 5.1.Following the completion described by Habiro in is article [12] Section 4. We have:1. L n ⊂ J n (see Prop 5.1 in Habiro’s article [12] ),2. ∆( J n ) ⊂ (cid:80) i + j = n J i ⊗ J j ,3. (cid:15) ( J n ) ⊂ L n ,4. S ( J n ) ⊂ J n .Thus we can define the completion ˆ U := lim ← n U J n as a (cid:92)Z [ q ± ] := lim ← n Z [ q ± ] L n algebra( (cid:92)Z [ q ± ] is Habiro’s ring).And it is endowed with a complete Hopf algebra structure: ˆ∆ : ˆ U → ˆ U ˆ ⊗ ˆ U , ˆ (cid:15) : ˆ U → (cid:92)Z [ q ± ] , ˆ S : ˆ U → ˆ U .willetts where : ˆ U ˆ ⊗ ˆ U = lim ← k,l ˆ U ⊗ (cid:92)Z [ q ± ] ˆ U ˆ U ⊗ (cid:92)Z [ q ± ] J k + J l ⊗ (cid:92)Z [ q ± ] ˆ U and J n is the closure of J n in ˆ U .This completion is a bigger algebra than U : Proposition 38.
The canonical projection maps induce an injective map U (cid:44) −→ ˆ U .Proof. Take the projective maps j n : U → U /J n , they induce a map j : U → ˆ U . Thismap in injective because if j ( x ) = 0 then x ∈ (cid:84) n ∈ N ∗ J n . But since J n ⊂ h n U h then (cid:84) n ∈ N ∗ J n ⊂ (cid:84) n ∈ N ∗ h n U h . It is a well known fact that (cid:84) n ∈ N ∗ h n U h = { } .Moreover, since J n ⊂ h n U h we have a map i : ˆ U → U h . Since we do not knowif this map is injective, we consider ˜ U := i ( ˆ U ) the image in U h . It is also an Hopfalgebra. Remark . ∞ (cid:80) i =0 q n ( n − E n ⊗ F ( n ) ∈ ˜ U ˆ ⊗ ˜ U We will need a lemma to compute some commutation rules.
Lemma 40. ( E ⊗ × q H ⊗ H = q H ⊗ H × ( E ⊗ × (1 ⊗ K )( F ( n ) ⊗ × q H ⊗ H = q H ⊗ H × ( F ( n ) ⊗ × (1 ⊗ K − n ) Proof.
Notice that since EH n = ( H + 2) n E and q H ⊗ H = (cid:80) ( h n n n ! ) H n ⊗ H n , then ( E ⊗ × q H ⊗ H = q ( H +2) ⊗ H × ( E ⊗
1) = q H ⊗ H × (1 ⊗ K ) × ( E ⊗ . The same can be done for F ( n ) .Let us now construct the universal invariant Q U h ( K ) by hand, seeing it as the (1 , -tangle with coupons.We can picture it as a 1-1 tangle with (2,2) coupons for R -matrix i.e. = R , = R -1 . And also with (1,1) coupons for the pivotal element: = K , = K − . The rest of the tangle remains unchanged.By replacing the R matrix with its formula, we get sums of diagrams with (1,1)-coupons E n , F ( n ) , K and (2,2)-coupons q H ⊗ H , q − H ⊗ H (that we can decompose intosum of (1,1)-coupons if seen as exponentials).To compute the universal invariant, start from the top of the tangle and multiply (tothe right) every coupons encountered.Now let’s see that we can separate the universal invariant into two pieces. Theexample of the trefoil knot will illustrate the process all along.The first step is to represent the knot with R matrices (2,2)-coupons and K ± (1,1)-coupons for the pivotal elements, as illustrated in Figure 4b .Now we write R as a sum, so the (2,2)-coupons labeled by R become the compositionof (2,2)-coupons labeled by q H ⊗ H and (1,1)-coupons labeled by E n or F ( n ) (see Figure4c ). 17 unification of the ado and colored jones polynomials of a knot (a) The trefoil knot. RRR K (b) The universal invariant. q H ⊗ H q H ⊗ H q H ⊗ H K (cid:80) n ,n ,n q n n − q n n − q n n − × E n E n E n F ( n ) F ( n ) F ( n ) (c) R-matrices seen has sums. Figure 4: The example of the trefoil knot.We now slide down - following the orientation - the (1,1)-coupons E n , F ( n ) and K n , taking first (at any step) the closest to the bottom (see Figure 5a). During thisprocess, the only non trivial commutations that appear are between E n ⊗ or ⊗ E n or F ( n ) ⊗ or ⊗ F ( n ) and q H ⊗ H or q − H ⊗ H . By Lemma 40, this only add some (1 , coupons labeled by K ± n (see Figure 5b).We are only left with coupons labeled by q H ⊗ H and q − H ⊗ H , you can see Figure6a for the example of the trefoil knot. These are exponentials and we can see themas the sum q ± H ⊗ H = (cid:80) ( ( ± h ) n n n ! ) H n ⊗ H n , the (2,2)-coupons now become a sum of(1,1)-coupons as shown for the trefoil knot in Figure 6b.Now take one of the (1,1)-coupon labeled by H n and slide it towards the second (1,1)-coupon labeled by H n (see Figure 6b). Since one is only left with coupons labeled bypowers of H , everything commutes and we get coupons H n (as shown in Figure 6c),summing them over n gives us (1,1)-coupons labeled by q ± H (see Figure 6d).We thus have the following proposition:18 .willetts q H ⊗ H q H ⊗ H q H ⊗ H K (cid:80) n ,n ,n q n n − q n n − q n n − × E n E n E n F ( n ) F ( n ) F ( n ) (a) The first two steps, sliding coupons. q H ⊗ H q H ⊗ H q H ⊗ H K (cid:80) n ,n ,n q n n − q n n − q n n − × E n E n F ( n ) F ( n ) F ( n ) K − n × E n (b) The third step, passing through the (2,2) coupons. Figure 5: Sliding coupons example with the trefoil knot.
Proposition 41. If K a knot and D a diagram of a 1-1 tangle T whose closure is K ,then: Q U h ( K ) = q f H Q ˜ U ( D ) where Q ˜ U ( D ) ∈ ˜ U and f is the writhe of the diagram. Now it is time to construct the Verma module on which the universal invariant willact as F ∞ ( q, A, D ) .Let V α be a ˆ R ˆ I -module freely generated by vectors { v , v , . . . } , and we endow itwith a action of ˜ U : Ev = 0 , Ev i +1 = v i , Kv i = q α − i v i , F ( n ) v i = (cid:20) n + ii (cid:21) q { α − i ; n } q v n + i Proposition 42.
The endomorphism of V α are scalars i.e. End ˜ U ( V α ) = ˆ R ˆ I Id V α . unification of the ado and colored jones polynomials of a knot q H ⊗ H q H ⊗ H q H ⊗ H (a) Quadratic part of trefoil universal in-variant. q H ⊗ H q H ⊗ H H n H n (cid:80) n h n n [ n ]! × (b) Illustration of quadratic simplifica-tion: first step. q H ⊗ H q H ⊗ H H n (cid:80) n h n n [ n ]! × (c) Illustration of quadratic simplifica-tion: second step. q H ⊗ H q H ⊗ H q H (d) Illustration of quadratic simplifica-tion: third step. Figure 6: Quadratic factorization and simplification for the trefoil knot.
Proof.
Let f ∈ End ˜ U ( V α ) . Kf ( v i ) = f ( Kv i ) = q α − i f ( v i ) , thus ∃ λ i ∈ ˆ R ˆ I such that f ( v i ) = λ i v i .Now since Ef ( v i +1 ) = f ( Ev i +1 ) = f ( v i ) , then λ i +1 v i = λ i v i hence we define λ := λ i and we have f = λId V α .We set q ± H v i = q ± ( α − i )22 v i ∈ q ± α V α . Proposition 43. Q U h ( K ) is in the center of U h .Proof. See [11] Proposition 8.2.
Proposition 44. If K is a knot and D is a diagram of a 1-1 tangle T whose closureis K , then: Q U h ( K ) v = q f H Q ˜ U ( D ) v = F ∞ ( q, A, D ) v . Hence, in particular, F ∞ ( q, A, D ) is independent of the choice of the diagram and wedenote it F ∞ ( q, A, K ) . .willetts Proof.
Using analogously the useful form of ADO invariant defined in Proposition13 to create a similar "useful form" for the action of Q U h ( K ) , one gets exactly thedefinition of F ∞ ( q, A, D ) in Definition 20. This subsection will be dedicated to connect our setup (the ring setup ˆ R ˆ I and thequantum algebra setup ˜ U ) to Habiro’s algebra setup in [12]. As we will see, we willget that our ring ˆ R ˆ I is contained in some h -adic ring, and hence that it is an integraldomain. Moreover we will see how to get back our unified invariant F ∞ ( q, A, K ) fromHabiro’s quantum algebra completions.Let us define U h the Q [[ h ]] subalgebra of U h topologically generated by H . And U the Z [ q ± ] subalgebra of U generated by H . We can complete the algebra into ˆ U = lim ← n U < { H + m ; n } , m ∈ Z > . We then have the following proposition.
Proposition 45. U h ∼ = Q [ α ][[ h ]] the h -adic completion of the polynomial ring Q [ α ] , ˆ U ∼ = ˆ R ˆ I .Proof. The first statement is the definition of U h replacing H with formal variable α .The second statement comes from the fact that, replacing K by A , U ∼ = Z [ q ± , A ± ] and { H + m ; n } ∼ = I n .Now, one can use Proposition 6.8 and 6.9 in Habiro’s article [12] and we have: Proposition 46.
We have that ˆ R ˆ I ⊂ Q [ α ][[ h ]] , and thus ˆ R ˆ I is an integral domain. Moreover, elements in ˆ R ˆ I can be uniquely expressed. This fact comes from Corol-lary 5.5 in [12]. Recall that q α := A and let { α ; n } (cid:48) = (cid:81) n − i =0 ( q α − q i ) we have thefollowing proposition: Proposition 47.
We have the following isomorphism: ˆ R ˆ I ∼ = lim ← n (cid:100) Z [ q ][ A ]( { α ; n } (cid:48) ) . Moreover, any element t ∈ ˆ R ˆ I can be uniquely written (cid:80) ∞ n =0 t n { α ; n } (cid:48) where t n ∈ (cid:100) Z [ q ] + (cid:100) Z [ q ] A .Proof. See Corollary 5.5 in [12].
Remark . This means that the unified invariant F ∞ ( q, A, K ) can be uniquely writtenas a serie (cid:80) ∞ n =0 t n { α ; n } (cid:48) where t n ∈ (cid:100) Z [ q ] + (cid:100) Z [ q ] A .Now let us present the quantum algebra setup used by Habiro. Our algebraand completion was done for the sole purpose of getting a nice form for our unifiedinvariant, allowing us to factorize it at each roots of unity. Habiro’s quantum algebrasetup has been studied more in details, and thus have more proprieties.21 unification of the ado and colored jones polynomials of a knot Let U Hab be the Z [ q ± ] subalgebra of U h generated by elements K ± , e, F [ n ] where e = { } E and F [ n ] = F n { n } ! . Let ˜ J n be the ideal generated by elements e i { H + m ; n − i } for all m ∈ Z .We denote ˆ U Hab := lim ← n U Hab ˜ J n .Now that his algebra setup is stated, let us make the connection with our unifiedinvariant F ∞ ( q, A, K ) . To do so, we will build a unified invariant with Habiro’s setupand prove that it is in fact F ∞ ( q, A, K ) .First remark that: Remark . If K a knot and D a diagram of a 1-1 tangle T whose closure is K , then: Q U h ( K ) = q f H Q ˜ U ( D ) where Q ˜ U ( D ) ∈ ˆ U Hab and f is the writhe of the diagram.We can then define the corresponding Verma module on ˆ R ˆ I .Let V αHab be the ˆ R ˆ I -module freely generated by vectors { v , v , . . . } , endowed withan action of ˆ U Hab : ev = 0 , ev i +1 = { α − i } q v i , Kv i = q α − i v i , F [ n ] v i = (cid:20) n + ii (cid:21) q v n + i . Then in a similar fashion we have that:
Remark . If K is a knot and D is a diagram of a 1-1 tangle T whose closure is K ,then there exists an element F Hab ∞ ( q, A, K ) ∈ q f α × ˆ R ˆ I such that: Q U h ( K ) v = q f H Q ˜ U ( D ) v = F Hab ∞ ( q, A, K ) v . Now, since V αHab ˆ ⊗ Q [[ H ]] ∼ = V α ˆ ⊗ Q [[ H ]] as U h module with coefficient in Q [ α ][[ h ]] .Then F Hab ∞ ( q, A, K ) = F ∞ ( q, A, K ) as elements of Q [ α ][[ h ]] . And since ˆ R ˆ I ⊂ Q [ α ][[ h ]] , F Hab ∞ ( q, A, K ) = F ∞ ( q, A, K ) as elements of ˆ R ˆ I . The colored Jones polynomials and the study of C ∞ (1 , A, K ) Knowing that the unified invariant comes from the universal invariant, we can usethis fact to recover the colored Jones polynomials. When we evaluate A = q α at q n in F ∞ ( q, A, K ) , we obtain the n -colored Jones polynomial denoted J n ( q, K ) (with nor-malization J n ( q, unknot ) = 1 ). Moreover we can use this fact to compute C ∞ (1 , A, K ) as the inverse of the Alexander polynomial, simplifying the factorization in Proposi-tion 27. Lemma 51.
If we denote V n the ( n + 1) dimensional highest weight module of ˜ U and V n the Verma module of highest weight q n and highest weight vector v , then V n ∼ = ˜ U v ⊂ V n .Proof. It comes down to F ( n +1) v = { n ; n + 1 } v n +1 = 0 × v n +1 = 0 . Proposition 52. If V n is the Verma of highest weight q n , Q U h ( K ) v = q fn q fn J n ( q , K ) v . .willetts Moreover, if we denote (cid:92)Z [ q ± ] Habiro’s ring completion of Z [ q ± ] by ideals ( { n } !) ,we have some well defined evaluation maps j n : ˆ R ˆ I → (cid:92)Z [ q ± ] q α (cid:55)→ q n .This allows us to state the following corollary. Corollary 53. F ∞ ( q, q n , K ) = q fn q fn J n ( q , K ) Hence, F ∞ ( q, A, K ) plays a double role in this dance, evaluating its first variable q at a root of unity ζ r , gives us the r -th ADO polynomial multiplied by this C ∞ ( r, A, K ) element. But if one evaluates the second variable A at q n , one gets the n -th coloredJones polynomial.We will use this double role to study the factorization of ADO polynomials. In-deed, the Melvin-Morton-Rozansky conjecture (MMR) proved by Bar-Natan andGaroufalidis in [2] makes the junction between the inverse of the Alexander poly-nomial and the colored Jones polynomials. We will use the h -adic version that westate below in Theorem 54 (see [6] Theorem 2).We denote A K ( t ) the Alexander polynomial of the knot K , with normalisation A unknot ( t ) = 1 and A K (1) = 1 . Theorem 54. (Bar-Natan, Garoufalidis)For K a knot, we have the following equality in Q [[ h ]] : lim n →∞ J n ( e h/n ) = 1 A K ( e h ) For the sake of simplicity, let’s assume that the knot K is framed so f = 0 .Now note that since F ∞ ( q, q n , K ) = J n ( q , K ) ∈ Z [ q ± ] and Z [ q ± ] ⊂ Q [[ h ]] wehave a map Q [[ h ]] → Q [[ h ]] , h (cid:55)→ hn that sends F ∞ ( q, q n , K ) (cid:55)→ F ∞ ( q /n , q, K ) aselements of Q [[ h ]] . But now F ∞ ( q /n , q, K ) converges to F ∞ (1 , q, K ) in the sensestated in [6] below Theorem 2, namely: lim n →∞ F ∞ ( q /n , q, K ) = F ∞ (1 , q, K ) ⇐⇒ lim n →∞ coeff ( F ∞ ( q /n , q, K ) , h m ) = coeff ( F ∞ (1 , q, K ) , h m ) , ∀ m ∈ N , where, for any analytic function f , coeff ( f ( h ) , h m ) = m ! d m dh m f ( h ) | h =0 . By Theorem 54, F ∞ (1 , q, K ) = A K ( q ) in Q [[ h ]] .On the other hand, if we denote (cid:92)Z [ A ± ] { } A the ring completion of Z [ A ± ] byideals (( A − A − ) n ) , then F ∞ (1 , A, K ) = C ∞ (1 , A, K ) in (cid:92)Z [ A ± ] { } A . Indeed setting q = 1 in Definition 20 and looking at the definition of C ∞ (1 , A, K ) in Proposition 27,one gets F ∞ (1 , A, K ) = C ∞ (1 , A, K ) .Since (cid:92)Z [ A ± ] { } A (cid:44) −→ Q [[ h ]] , A → e h (see [12] Proposition 6.1 and [10] Corollary4.1), then we have the following proposition: Proposition 55. If K is framed, C ∞ (1 , A, K ) = A K ( A ) By the discussion in the paragraph preceding Proposition 31, we have:
Corollary 56. If K is framed, C ∞ ( r, A, K ) = A K ( A r ) This allows us to state a factorization theorem:23 unification of the ado and colored jones polynomials of a knot
Theorem 57. (Factorization)For a knot K and an integer r ∈ N ∗ , we have the following factorization in ˆ R Ir : F ∞ ( ζ r , A, K ) = A rf × ADO r ( A, K ) A K ( A r ) where f is the framing of the knot. The colored Jones polynomials determine the ADO polynomials
One may ask what is the relationship between ADO invariants and the colored Jonespolynomials.Does one family of polynomial determines completely the other? For the sake ofsimplicity the knot K is supposed framed in this paragraph.This is the case for { ADO r ( A, K ) } r ∈ N ∗ → { J n ( q , K ) } n ∈ N ∗ , knowing the ADOpolynomials allows to find the colored Jones polynomials. This result was stated in[5] Corollary 15.With our setup, we can get back this result as follows. Remark . Notice that F ∞ ( ζ r , ζ N r , K ) = J N ( ζ r , K ) = ADO r ( ζ N r , K ) A K (1) = ADO r ( ζ N r , K ) .Since J N is a polynomial knowing an infinite number of value of it determines it.Given the family of polynomials { ADO r ( A, K ) } r ∈ N ∗ we then know each value of J N at any root of unity hence we know J N entirely.But we can also have the other way around: { J n ( q , K ) } n ∈ N ∗ → { ADO r ( A, K ) } r ∈ N ∗ . Knowing only the colored Jones polynomials recover the ADO polynomials. We willprove it by seeing that the colored Jones polynomials determines the unified invariant F ∞ ( q, A, K ) .Let ∀ k ∈ N , f k : Q [ α ][[ h ]] → Q [[ h ]] , α (cid:55)→ k the evaluation map. Proposition 59. ∩ k ∈ N ker ( f k ) = { } Proof.
Let x ∈ ∩ k ∈ N ker ( f k ) , we write x = (cid:80) n g n ( α ) h n where g n ( α ) ∈ Q [ α ] .Then, for each k ∈ N , we have that g n ( k ) = 0 ∀ n . Since g n are polynomials thatvanish at an infinite number of point, they are .Hence ∩ k ∈ N ker ( f k ) = { } .Let f : Q [ α ][[ h ]] → (cid:81) k ∈ N Q [[ h ]] , x (cid:55)→ ( f k ( x )) k ∈ N . ker ( f ) = ∩ k ∈ N ker ( f k ) = { } , hence f is injective. Remark . For any knot K , f ( F ∞ ( q, A, K )) = { J n ( q , K ) } n ∈ N ∗ . Proposition 61.
For any knot K , F ∞ ( q, A, K ) = f − ( { J n ( q , K ) } n ∈ N ∗ ) . Setting g : ˆ R ˆ I → (cid:81) r ∈ N ∗ ˆ R Ir , x (cid:55)→ ( ev r × F C r ( x )) r ∈ N ∗ , we get the following theorem: Theorem 62.
The map h = g ◦ f − : Im ( f | ˆ R ˆ I ) → (cid:81) r ∈ N ∗ ˆ R Ir is such that for every knot K , { ADO r ( A, K ) } r ∈ N ∗ = h ( { J n ( q , K ) } n ∈ N ∗ ) . .willetts Application 1: the unified invariant and the ADO invariants are q -holonomic The fact that the colored Jones polynomials are q -holonomic was proved in [7]. Letus state what it means and then let’s prove that the unified invariant and the ADOpolynomials verify the same holonomic rule. For the sake of simplicity we will workwith -framed knot.Let Q : Z [ q ± ] N ∗ → Z [ q ± ] N ∗ and E : Z [ q ± ] N ∗ → Z [ q ± ] N ∗ such that: ( Qf )( n ) = q n f ( n ) , ( Ef )( n ) = f ( n + 1) . Note that these operators can be extended to operators on Q [[ h ]] N ∗ .Let us denote J • ( q , K ) = { J n ( q , K ) } n ∈ N ∗ the colored Jones function. Now, fromTheorem 1 in [7], for any knot K there exists a polynomial α K ( Q, E, q ) such that α K ( Q, E, q ) J • ( q , K ) = 0 . We say that J • ( q , K ) is q -holonomic.We define similar operators on Q [ α ][[ h ]] and show that the same polynomial α K ,taken in terms of those new operators, annihilates F ∞ ( q, q α , K ) .Let ˜ Q : Q [ α ][[ h ]] → Q [ α ][[ h ]] and ˜ E : Q [ α ][[ h ]] → Q [ α ][[ h ]] such that if we take x ( α ) := (cid:80) + ∞ k =0 x k ( α ) h k ∈ Q [ α ][[ h ]] with x k ( α ) ∈ Q [ α ] : ˜ Q ( x ( α )) = q α x ( α ) , ˜ E ( x ( α )) = x ( α + 1) . where x ( α + 1) := (cid:80) + ∞ k =0 x k ( α + 1) h k Remark . Here, ˜ Q is just the multiplication of any element with q α .Notice that if you take the injective map f : Q [ α ][[ h ]] → Q [[ h ]] N ∗ , x ( α ) (cid:55)→ ( x ( k )) k ∈ N ∗ previously defined, you have: f ◦ ˜ Q = Q ◦ f, f ◦ ˜ E = E ◦ f. Hence, f ◦ α K ( ˜ Q, ˜ E, q ) = α K ( Q, E, q ) ◦ f . Since α K ( Q, E, q ) J • ( q , K ) = 0 and f ( F ∞ ( q, q α , K )) = J • ( q , K ) we obtain f ◦ α K ( ˜ Q, ˜ E, q )( F ∞ ( q, q α , K )) = 0 . The injec-tivity of f gives the following theorem. Theorem 64.
For any framed knot K , α K ( ˜ Q, ˜ E, q )( F ∞ ( q, q α , K )) = 0 . Now let us look at what happens at roots of unity. To do so we must restrictourselves to a ring allowing evaluation at roots of unity such as ˆ R ˆ I .Since ˜ Q ( I n ) ⊂ I n and ˜ E ( I n ) ⊂ I n , we can restrict the operators ˜ Q and ˜ E to ˆ R ˆ I , forthe sake of simplicity we will still write them ˜ Q : ˆ R ˆ I → ˆ R ˆ I and ˜ E : ˆ R ˆ I → ˆ R ˆ I .Now let r ∈ N ∗ , let Q : ˆ R Ir → ˆ R Ir and E : ˆ R Ir → ˆ R Ir such that if we take x ( α ) = (cid:80) ∞ k =0 x k ( α ) { rα } k ∈ ˆ R Ir with x k ( α ) ∈ Z [ ζ r , A ] (recall that we denote ζ α r := A ): Q ( x ( α )) = ζ α r x ( α ) , E ( x ( α )) = x ( α + 1) where x ( α + 1) = (cid:80) ∞ k =0 x k ( α + 1)( − k { rα } k .Since ev r ◦ ˜ Q = Q ◦ ev r and ev r ◦ ˜ E = E ◦ ev r , the same formula holds: α K ( Q, E, ζ r )( F ∞ ( ζ r , ζ α r , K )) = 0 . By Theorem 57, α K ( Q, E, ζ r )( ADO r ( ζ α r , K ) A K ( ζ rα r ) ) = 0 .25 unification of the ado and colored jones polynomials of a knot Remark . We have the following identities: Q ( ADO r ( ζ α r , K ) A K ( ζ rα r ) ) = q α ADO r ( ζ α r , K ) A K ( ζ rα r ) = Q ( ADO r ( ζ α r , K )) A K ( ζ rα r ) E ( ADO r ( ζ α r , K ) A K ( ζ rα r ) ) = E ( ADO r ( ζ α r , K )) E ( A K ( ζ rα r )) = E ( ADO r ( ζ α r , K )) A K ( ζ rα r ) (because ζ r ( α +1)2 r = ζ rα r )Hence α K ( Q,E,ζ r )( ADO r ( ζ α r , K )) A K ( ζ rα r ) = 0 , which proves the following theorem: Theorem 66.
For any framed knot K , α K ( Q, E, ζ r )( ADO r ( ζ α r , K )) = 0 .Remark . In the upcoming article [3], Brown, Dimofte and Geer proved that theADO invariant of links are q-holonomic (Theorem 4.3), which generalise Theorem 66.Their Theorem 4.4 actually gives a converse statement of Theorem 66: any polynomialannihilating the ADO family will also annihilate the colored Jones. This proves thatthe ADO and colored Jones family are annihilated by the same polynomials.
Application 2: the unified invariant is the loop expansion of the coloredJones function
Let’s first introduce the loop expansion of the colored (see section 2 in [6]). We canwrite the colored Jones polynomials as an expansion (see [18] for more details): J n ( e h , K ) = + ∞ (cid:88) k =0 P k ( e nh ) A K ( e nh ) k +1 h k where P k ( X ) ∈ Q [ X, X − ] .Hence we get an element: J α ( q , K ) = + ∞ (cid:88) k =0 P k ( e αh ) A K ( e αh ) k +1 h k ∈ Q [ α ][[ h ]] that is such that f ( J α ( q , K )) = J • ( q , K ) .This means that it evaluates into the colored Jones at α = n , we call it loop expansionof the colored Jones function . Proposition 68.
For any knot K , we have the following identity in Q [ α ][[ h ]] : J α ( q , K ) = F ∞ ( q, q α , K ) . Proof.
The fact that f is injective proves the proposition. Remark . Putting everything together, this subsection implies that the unifiedinvariant F ∞ ( q, A, K ) is an integral version of the colored Jones function, built in a ringallowing evaluations at roots of unity. The integrality and the existence of evaluationmaps allow us to recover the ADO polynomials, the fact that the completion ring isa subring of an h -adic ring allows us to connect it to other notions of colored Jonesfunction/ invariants.Another approach, described by Gukov and Manolescu in [9], would be to see theunified invariant as a power serie in q, A (as opposed to a quantum factorial expansionas we have here). This would be another integral version of it.Indeed, because it verifies Proposition 68 and Theorem 64, the unified invariant F ∞ ( q, A, K ) except being a power serie, also verifies conjecture 1.5 and 1.6 in [9].Thus, if F ∞ ( q, A, K ) could be written as a power serie, it would fully verify the con-jectures.This is the case for positive braid knots, as show by Park in [17]. This means thatfor a positive braid knot, the unified invariant and the GM power serie coincide.26 .willetts This section will be dedicated to compute the unified invariant F ∞ ( q, A, K ) on someexamples. We will also explicitly compute C ∞ (1 , A, K ) and see that it is equal to theinverse of the Alexander polynomial.To do so we will use state diagrams and compute the unified invariant from it. Youcan also use them to compute the ADO polynomials (see [15] section 4). Recall that q α := A . ii i (a) The trefoil knot.
000 0 j ji + ji i (b) The figure eight knot. Figure 7: Examples of state diagrams to compute the invariants.
The Trefoil Knot:
We will denote the trefoil knot : F ∞ ( q, A, ) = q α (cid:88) i q α − i q i ( i − { α ; i } q q − iα C ∞ (1 , A, ) = q α (cid:88) i q − iα { α } iq = q α − q − α { α } q = q α A ( q α ) The Figure Eight Knot:
We will denote the figure eight knot : F ∞ ( q, A, ) = (cid:88) i,j q i − j ) q − iα q jα ( − i q − i ( i − (cid:20) i + jj (cid:21) q { α − j ; i } q q ( i + j ) α × q − ij q − j ( j − { α ; j } q q − ( i + j ) α q ij = (cid:88) i,j q i − j ) q ( j − i ) α ( − i (cid:20) i + jj (cid:21) q q − i ( i − q − j ( j − { α ; i + j } q unification of the ado and colored jones polynomials of a knot C ∞ (1 , A, ) = (cid:88) i,j q ( j − i ) α ( − i (cid:18) i + jj (cid:19) { α } i + jq = (cid:88) N N (cid:88) i =0 q Nα q − iα ( − i (cid:18) i + jj (cid:19) { α } Nq = (cid:88) N q Nα { α } Nq N (cid:88) i =0 ( − q − α ) i (cid:18) i + jj (cid:19) = (cid:88) N q Nα { α } Nq (1 − q − α ) N = (cid:88) N { α } Nq = 11 − { α } q = 1 A ( q α ) The Cinquefoil Knot:
We will denote it by : F ∞ ( q, A, ) = q α (cid:88) i,j,k q α − i − j + k ) q − i − j + k ) α q i ( k − j ) q k ( i − j ) q i ( i − q j ( j − × q k ( k − { α ; i } q { α − k + j ; j } q { α − i + j ; k } q × (cid:20) kk − j (cid:21) q (cid:20) i − j + kk (cid:21) q C ∞ (1 , A, ) = q α (cid:88) i,j,k q − i − j + k ) α { α } i + j + kq (cid:18) kk − j (cid:19)(cid:18) i − j + kk (cid:19) = q α (cid:88) i,j,k q − i − j + k ) α { α } i + j + kq (cid:18) i − j + kj, k − j, i − j (cid:19) = q α (cid:88) N (cid:88) j,k q − Nα { α } N +2 jq (cid:18) Nj, k − j, N − k (cid:19) = q α (cid:88) N q − Nα { α } Nq (cid:88) j,k { α } jq (cid:18) Nj, k − j, N − k (cid:19) = q α (cid:88) N q − Nα { α } Nq (2 + { α } q ) N = q α − q − α { α } q (2 + { α } q )= q α A ( q α ) The three twist Knot: .willetts We will denote it by : F ∞ ( q, A, ) = q − α (cid:88) i,j,k q i − j + k ) − α q (5 i +5 j − k ) α q ij q j − k )( i + j ) q i ( i − × q j ( j − q k ( k − ( − i + j + k { α ; i } q { α − i ; j } q { α − j + k ; k } q × (cid:20) jj − k (cid:21) q (cid:20) i + jj (cid:21) q C ∞ (1 , A, ) = q − α (cid:88) i,j,k q (5 i +5 j − k ) α ( − i + j + k { α } i + j + kq (cid:18) jj − k (cid:19)(cid:18) i + jj (cid:19) = q α (cid:88) i,j,k q (5 i +5 j − k ) α ( − i + j + k { α } i + j + kq (cid:18) i + jk, j − k, i (cid:19) = q − α (cid:88) N (cid:88) j,k q (5 N − k ) α ( − N + k { α } N + kq (cid:18) Nk, j − k, N − i (cid:19) = q − α (cid:88) N q Nα ( − N { α } Nq (cid:88) j,k q − kα ( − k { α } kq × (cid:18) Nk, j − k, N − i (cid:19) = q − α (cid:88) N q Nα ( − N { α } Nq (2 − q − α { α } q ) N = q − α
11 + q α { α } q (2 − q − α { α } q )= q − α A ( q α )00 00 i i − ji − j + k i − j + ki − j + k k − jk (a) The Cinquefoil Knot.
00 00 ii + j − ki + j − k i + jj − kji (b) The three twist Knot. Figure 8: Examples of state diagrams to compute the invariants.
References [1] Y. Akutsu, T. Deguchi, and T. Ohtsuki. Invariants of colored links.
J. KnotTheory Ramifications , 1(2):161–184, 1992.29 unification of the ado and colored jones polynomials of a knot [2] D. Bar-Natan and S. Garoufalidis. On the melvin–morton–rozansky conjecture.
Inventiones mathematicae , 125(1):103–133, 1996.[3] J. Brown, T. Dimofte, and N. Geer. The ado invariants are a q-holonomic family. arXiv preprint , 2020.[4] F. Costantino, N. Geer, and B. Patureau-Mirand. Quantum invariants of 3-manifolds via link surgery presentations and non-semi-simple categories.
Journalof Topology , 7(4):1005–1053, 2014.[5] F. Costantino, N. Geer, and B. Patureau-Mirand. Relations between witten–reshetikhin–turaev and nonsemisimple sl (2) 3–manifold invariants.
Algebraic &Geometric Topology , 15(3):1363–1386, 2015.[6] S. Garoufalidis and T. T. Le. An analytic version of the melvin-morton-rozanskyconjecture. arXiv preprint math/0503641 , 2005.[7] S. Garoufalidis and T. T. Lê. The colored jones function is q-holonomic.
Geom-etry & Topology , 9(3):1253–1293, 2005.[8] N. Geer, B. Patureau-Mirand, and V. Turaev. Modified quantum dimensions andre-normalized link invariants.
Compositio Mathematica , 145(1):196–212, 2009.[9] S. Gukov and C. Manolescu. A two-variable series for knot complements. arXivpreprint arXiv:1904.06057 , 2019.[10] K. Habiro. Cyclotomic completions of polynomial rings.
Publications of theResearch Institute for Mathematical Sciences , 40(4):1127–1146, 2004.[11] K. Habiro. Bottom tangles and universal invariants.
Algebraic & GeometricTopology , 6(3):1113–1214, 2006.[12] K. Habiro. An integral form of the quantized enveloping algebra of sl2 and itscompletions.
Journal of Pure and Applied Algebra , 211(1):265–292, 2007.[13] R. Lawrence. A universal link invariant using quantum groups. Technical report,PRE-31446, 1988.[14] R. Lawrence. A universal link invariant. In
The interface of mathematics andparticle physics . 1990.[15] J. Murakami. Colored alexander invariants and cone-manifolds.
Osaka Journalof Mathematics , 45(2):541–564, 2008.[16] T. Ohtsuki.
Quantum Invariants: A Study of Knots, 3-manifolds, and TheirSets . K & E series on knots and everything. World Scientific, 2002.[17] S. Park. Large color r -matrix for knot complements and strange identities, 2020.[18] L. Rozansky. The universal r-matrix, burau representation, and the melvin–morton expansion of the colored jones polynomial. Advances in Mathematics ,134(1):1–31, 1998.
Institut de Mathématiques de Toulouse, UMR5219,UPS, F-31062 Toulouse Cedex 9, France
E-mail address : [email protected]@math.univ-toulouse.fr