Linking covariant and canonical LQG: new solutions to the Euclidean Scalar Constraint by Alesci, Thiemann, and Zipfel

This week I have been continuing my work on the Hamiltonian constraint in Loop Quantum Gravity,  The main paper I’ve been studying this week is ‘Linking covariant and canonical LQG: new solutions to the Euclidean Scalar Constraint’. Fortunately enough linking   covariant and canonical LQG was also the topic of a recent seminar by Zipfel in the ilqgs spring program.

The authors of this paper emphasize that spin-foam models could realize a projection on the physical Hilbert space of canonical Loop Quantum Gravity (LQG). As a test  the authors analyze the one-vertex expansion of a simple Euclidean spin-foam. They find that for fixed Barbero-Immirzi parameter γ= 1 the one vertex-amplitude in the KKL prescription annihilates the Euclidean Hamiltonian constraint of LQG. Since for γ = 1 the Lorentzian part of the Hamiltonian constraint does not contribute this gives rise to new solutions of the Euclidean theory. Furthermore, they fi nd that the new states only depend on the diagonal matrix elements of the volume. This seems to be a generic property when applying the spin-foam projector.

To circumvent the problems of the canonical theory, a
covariant formulation of Quantum Gravity, the so-called spin-foam model was introduced. This model is mainly based on the observation that the Holst action for GR  de fines  a constrained BF-theory. The strategy is first to quantize discrete BF-theory and then to implement the so called simplicity constraints. The main building block of the model is a linear two-complex  κ embedded into 4-dimensional space-time M whose boundary is given by an initial and final gauge invariant spin-network, Ψi respectively Ψf , living on the initial respectively final spatial hyper surface of a
foliation of M. The physical information is encoded in the spin-foam amplitude.


where Af , Ae and Av are the amplitudes associated to the internal faces, edges and vertices of  κ and B contains the boundary amplitudes.

Each spin-foam can be thought of as generalized
Feynman diagram contributing to the transition amplitude from an ingoing spin-network to an outgoing spin-network. By summing over all possible two-complexes one obtains the complete transition amplitude between ψi and ψf .

The main idea in this paper is that if f spin-foams provide a rigging map  the physical inner product would be given by


and the rigging map would correspond  to


Since all constraints are satis ed in Hphys the physical scalar product must obey


for all ψout, ψn ∈ Hkin.

As a test  the authors consider an easy spin-foam amplitude and show that


where  κ is a two-complex with only one internal vertex such that  Φ is a spin-network induced on the boundary of  κ and Hn is the Hamiltonian constraint acting on the node n.

 Hamiltonian constraint

The classical Hamiltonian constraint is





The constraint can be split into its Euclidean part H = Tr[F∧e] and Lorentzian part HL = C- H.



where V is the volume of an arbitrary region  ∑ containing the point x. Smearing the constraints with lapse function N(x) gives


This expression requires a regularization in order to obtain a well-de fined operator on Hkin. Using a triangulation T of the manifold  into elementary tetrahedra with analytic links adapted to the graph Γ of an arbitrary spin-network.


Three non-planar links de fine a tetrahedron . Now  decompose H[N] into a sum of one term per each tetrahedron of the triangulation,


To define  the classical regularized Hamiltonian constraint as,


The connection A  and the curvature are regularized  by the holonomy h in SU (2), where in the fundamental representation m = ½. This gives,


which converges to the Hamiltonian constraint if the triangulation is sufficiently fine.

As seen in the post

This can be generalized with a trace in an arbitrary irreducible representation m leading to


this converges to H[N] as well.


The important properties of the Euclidean Hamiltonian constraint are;

when acting on a spin-network state, the operator reduces to a sum over terms each acting on individual nodes. Acting on nodes of valence n the operator gives


The Hamiltonian constraint on di ffeomorphism invariant states is independent from the refi nement of the triangulation.


Since the Ashtekar-Lewandowsk volume operator annihilates coplanar nodes and gauge invariant nodes of valence three H does not act on the new nodes – the so called extraordinary nodes.

Action on a trivalent node

To ompute the action of the operator Hmdeltaon a trivalent node where all links are outgoing, denote a trivalent node by |n(ji,jj , jk)> ≡ |n3>, whereas ji, jj ,jk are the spins of the adjacent links ei, ej , ek:



To quantize  Hmdelta[N] the holonomies and the volume are replaced by their corresponding operators and the Poisson bracket is replaced by a commutator. Since the volume operator vanishes on a gauge invariant trivalent node  only need to compute;


so, h(m) creates a free index in the m-representation located at the node , making it non-gauge invariant and a new node on the link ek:


so we get,


where the range of the sums over a, b is determined by the Clebsch-Gordan conditions and


The complete action of the operator on a trivalent state |n(ji, jj , jk)> can be obtained by contracting the trace part with εijk. So, H projects on a linear combination of three spin networks which differ by exactly one new link labelled by m between each couple of the oldlinks at the node.

Action on a 4-valent node

The computation for a 4-valent node |n4> is


where i labels the intertwiner – inner link.

The holonomy h(m) changes the valency of the node and the Volume therefore acts on the 5-valent non-gauge invariant node. Graphically this corresponds to


and finally we get:


This can be simpli ed to;



Using  the de finition of an Euclidean spin-foam models as suggested by Kaminski, Kisielowski and Lewandowski (KKL) and since we,re   only interested in the evaluation of a spin-foam amplitude we choose a combinatorial de finition of the model:

Consider an oriented two-complex  de fined as the union of the set of faces (2-cells) F, edges (1-cells) E and vertices (0-cells) V such that every edge e is a 1-face of at least one face f (e ∈ ∂f) and every vertex V is a 0-face of at least one edge e (v ∈∂e).

We call edges which are contained in more than one face f internal and denote the set of all internal edges by Eint.  All vertices adjacent to more than one internal edge are also called internal and denote the set of these vertices by Vint. The boundary ∂κ is the union of all external vertices (-nodes) n ∉ Vint and external edges (links) l ∉Eint.

A spin-foam is a triple (κ, ρ; I) consisting of a proper foam whose faces are labelled by irreducible representations of a Lie-group G, in this case here SO(4) and whose internal edges are labeled by intertwiners I. This induces a spin-network structure ∂(κ, ρ; I) on the boundary of κ.

The BF partition function can be rewritten as


Av de fines an SO (4) invariant function on the graph Γ induced on the boundary of the vertex v


In the EPRL model  the simplicity constraint is imposed weakly so,


It follows immediately that


defi nes the EPRL vertex amplitude with


Expanding the delta function in terms of spin-network function and integrating over the group elements gives


In order to evaluate the fusion coefficients by graphical calculus it is convenient to work with 3j-symbols instead of Clebsch-Gordan coefficients. When replacing the Clebsch-Gordan coefficients we have to multiply by an overall factor cc.

Spin-foam projector

Given any couple of ingoing and outgoing kinematical states ψout, ψin, the Physical scalar product can be
formally de fined by


where η is a projector – Rigging map onto the Kernel of the Hamiltonian constraint.

Suppose that the transition amplitude Z


can be expressed in terms of a sum of spin-foams, then


this can be interpreted as a function on the boundary graph ∂κ;


in the EPRL sector.

Restricting the boundary elements h ∈ SU (2)  ⊂ SO (4) then;


where |S>N is a normalized spin-network function on SU(2). This fi nally implies;





Compute new solutions to the Euclidean Hamiltonian constraint by employing spin-foam methods. Show thatlinkingequ3.15


in the Euclidean sector with γ= 1 and s = 1, where κ is a 2-complex with only one internal vertex.

Trivalent nodes

Consider the simplest possible case given by an initial and final state  |Θ>, characterized by two trivalent nodes joined by three links:



the only states produced by the HamiltonianHm acting on a node, are given by a linear combination of spin-networks that diff er from the original one by the presence of an extraordinary (new) link. In particular the term sHwill be non vanishing only if |s> is of the kind:


The simplest two-complex κ(Θ,s) with only one internal vertex de fining a cobordism between   |Θ> and |s> is a tube  Θ x[0,1] with an additional face between the internal vertex and the new link m,

Since the space of three-valent intertwiners is one-dimensional and all labelings jf are fixed by the states   |Θ> , |s>  the fi rst sum  is trivial.


Γv= s and therefore we have,


where the sign factor is due to the orientation of s, The fusion coefficients contribute four 9j symbols since,


The full amplitude is




This yields,


The last two terms are equivalent to the first term when exchanging jk↔ jj. The EPRL spin-foam reduces just to the SU(2) BF amplitude that is just the single 6j  in the first line.

Now using the defi nition of a 9j in terms of three 6j’s,


The 9j’s involved in this expression can be reordered using the permutation symmetries  giving,


the statessphysare solutions of the Euclidean Hamiltonian constraint

The spin-foam amplitude selects only those terms which depend on the diagonal elements on the volume. This  simplifies the calculation since we do not have to evaluate the volume explicitly.

Four valent nodes

The case with ψ in = ψout = |n4>where


The matrix element <s| Hm|n4>  is non-vanishing if |s>is of the form


Choose again a complex  of the form


with one additional face jl.

The vertex trace  can be evaluated by graphical calculus


The fusion coefficients give two 9j symbols for the two trivalent edges and two 15j- symbols for the two four-valent edges.The fusion coefficients reduce to 1 when γ=1.Taking the scalar product  we obtain,


Taking the scalar product with the Hamiltonian gives,


Summing over a and using the orthogonality relation gives,


The three 6j’s in the two terms defi ne a 9j ,summing over the indexes and b respectively gives:


the final result is,


As for the trivalent vertex the spin-foam amplitude just takes those elements into account which depend on the diagonal Volume elements.


LQG is grounded on two parallel constructions; the canonical and the covariant ones. In this paper the authors  construct a simple spin-foam amplitude which annihilates the Hamiltonian constraint .They found that in the euclidean sector with signature s = 1 and Barbero-Immirzi parameter γ = 1 the Euclidean Hamiltonian constraint is annihilated by a spin-foam amplitude Z for a simple two-complex with only one internal vertex. The one vertex amplitudes of BF theory are explicit analytic solutions of the Hamiltonian theory.

Also the 6j symbol associated to every face is annihilated by the Euclidean scalar constraint. This is a generalization of the work by Bonzom-Freidel in the context of 3d gravity where they found that the 6jis annihilated by a suitable quantization of the 3d scalar constraint F = 0 . The spin-foam amplitude diagonalizes the Volume.


Related articles

Enhanced by Zemanta

2 thoughts on “Linking covariant and canonical LQG: new solutions to the Euclidean Scalar Constraint by Alesci, Thiemann, and Zipfel”

  1. From the introduction at this site:
    Loop quantum gravity is a canonical approach to the quantization of general relativity. One of the issues in this approach is the identification of the physically correct.” :Unquote
    From wikipedia: “In the canonical approach, theological concerns take precedent”

    From me: Why not use your mathematical skills to work out if the model suggested at might be a bit closer to reality? It could eventually earn you a nobel prize…

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s