A further generalization of the notion of imbedding space
The hypothesis that Planck constant is quantized having in principle all possible rational values but with some preferred values implying algebraically simple quantum phases has been one of the main ideas of TGD during last years. The mathematical realization of this idea leads to a profound generalization of the notion of imbedding space obtained by gluing together infinite number of copies of imbedding space along common 4dimensional intersection. The hope was that this generalization could explain charge fractionization but this does not seem to be the case. This problem led to a futher generalization of the imbedding space and this is what I want to discussed below.
1. Original view about generalized imbedding space
The original generalization of imbedding space was basically following. Take imbedding space H=M^{4}×CP_{2}. Choose submanifold M^{2}×S^{2}, where S^{2} is homologically nontrivial geodesic submanifold of CP_{2}. The motivation is that for a given choice of Cartan algebra of Poincare algebra (translations in time direction and spin quantization axis plus rotations in plane orthogonal to this plane plus color hypercharge and isospin) this submanifold remains invariant under the transformations leaving the quantization axes invariant.
Form spaces M^{4}= M^{4}\M^{2} and CP_{2} = CP_{2}\S^{2} and their Cartesian product. Both spaces have a hole of codimension 2 so that the first homotopy group is Z. From these spaces one can construct an infinite hierarchy of factor spaces M^{4}/G_{a} and CP2/G_{b} where G_{a} is discrete group of SU(2) leaving quantization axis invariant. In case of Minkowski factor this means that the group in question acts essentially as a combination reflection and to rotations around quantization axies of angular momentum. The generalized imbedding space is obtained by gluing all these spaces together along M^{2}×S^{2}.
The hypothesis is that Planck constant is given by the ratio hbar= n_{a}/n_{b}, where n_{i} is the order of maximal cyclic subgroups of G_{i}. The hypothesis states also that the covariant metric of the Minkowski factor is scaled by the factor (n_{a}/n_{b})^{2}. One must take care of this in the gluing procedure. One can assign to the field bodies describing both self interactions and interactions between physical systems definite sector of generalized imbedding space characterized partially by the Planck constant. The phase transitions changing Planck constant correspond to tunnelling between different sectors of the imbedding space.
2. Fractionization of quantum numbers is not possible if only factor spaces are allowed
The original idea was that the modification of the imbedding space inspired by the hierarchy of Planck constants could explain naturally phenomena like quantum Hall effect involving fractionization of quantum numbers like spin and charge. This does not however seem to be the case. G_{a}× G_{b} implies just the opposite if these quantum numbers are assigned with the symmetries of the imbedding space. For instance, quantization unit for orbital angular momentum becomes n_{a} where Z_{na} is the maximal cyclic subgroup of G_{a}.
One can however imagine obtaining fractionization at the level of imbedding space for spacetime sheets, which are analogous to multisheeted Riemann surfaces (say Riemann surfaces associated with z^{1/n} since the rotation by 2π understood as a homotopy of M^{4} lifted to the spacetime sheet is a nonclosed curve. Continuity requirement indeed allows fractionization of the orbital quantum numbers and color in this kind of situation. Lifting up this idea to the level of imbedding space leads to the generalization of the notion of imbedding space.
3. Both covering spaces and factor spaces are possible
The observation above stimulates the question whether it might be possible in some sense to replace H or its factors by their multiple coverings.
 This is certainly not possible for M^{4}, CP_{2}, or H since their fundamental groups are trivial. On the other hand, the fixing of quantization axes implies a selection of the subspace H_{4}= M^{2}× S^{2}subset M^{4}× CP_{2}, where S^{2} is a geodesic sphere of CP_{2}. M^{4}=M^{4}\M^{2} and CP_{2}=CP_{2}\S^{2} have fundamental group Z since the codimension of the excluded submanifold is equal to two and homotopically the situation is like that for a punctured plane. The exclusion of these submanifolds defined by the choice of quantization axes could naturally give rise to the desired situation.
 H_{4} represents a straight cosmic string. Quantum field theory phase corresponds to Jones inclusions with Jones index M:N<4. Stringy phase would by previous arguments correspond to M:N=4. Also these Jones inclusions are labelled by finite subgroups of SO(3) and thus by Z_{n} identified as a maximal Abelian subgroup.
One can argue that cosmic strings are not allowed in QFT phase. This would encourage the replacement M^{4}×CP_{2} implying that surfaces in M^{4}×S^{2} and M^{2}×CP_{2} are not allowed. In particular, cosmic strings and CP_{2} type extremals with M^{4} projection in M^{2} and thus lightlike geodesic without zitterwebegung essential for massivation are forbidden. This brings in mind instability of Higgs=0 phase.
 The covering spaces in question would correspond to the Cartesian products M^{4}_{na}× CP_{2}_{nb} of the covering spaces of M^{4} and CP_{2} by Z_{na} and Z_{nb} with fundamental group is Z_{na}× Z_{nb}. One can also consider extension by replacing M^{2} and S^{2} with its orbit under G_{a} (say tedrahedral, octahedral, or icosahedral group). The resulting space will be denoted by M^{4}×G_{a} resp. CP_{2}×G_{b}. Product sign does not signify for Caretsian product here.
 One expects the discrete subgroups of SU(2) emerge naturally in this framework if one allows the action of these groups on the singular submanifolds M^{2} or S^{2}. This would replace the singular manifold with a set of its rotated copies in the case that the subgroups have genuinely 3dimensional action (the subgroups which corresponds to exceptional groups in the ADE correspondence). For instance, in the case of M^{2} the quantization axes for angular momentum would be replaced by the set of quantization axes going through the vertices of tedrahedron, octahedron, or icosahedron. This would bring noncommutative homotopy groups into the picture in a natural manner.
Also the orbifolds M^{4}/G_{a}× CP_{2}/G_{b} can be allowed as also the spaces M^{4}/G_{a}× (CP_{2}×G_{b}) and (M^{4}×G_{a})× CP_{2}/G_{b}. Hence the previous framework would generalize considerably by the allowance of both coset spaces and covering spaces.
4. Do factor spaces and coverings correspond to the two kinds of Jones inclusions?
What could be the interpretation of these two kinds of spaces?
 Jones inclusions appear in two varieties corresponding to M:N<4 and M:N=4 and one can assign a hierarchy of subgroups of SU(2) with both of them. In particular, their maximal Abelian subgroups Z_{n} label these inclusions. The interpretation of Z_{n} as invariance group is natural for M: N< 4 and it naturally corresponds to the coset spaces. For M:N=4 the interpretation of Z_{n} has remained open. Obviously the interpretation of Z_{n} as the homology group defining covering would be natural.
 M:N=4 should correspond to the allowance of cosmic strings and other analogous objects. Does the introduction of the covering spaces bring in cosmic strings in some controlled manner? Formally the subgroup of SU(2) defining the inclusion is SU(2) would mean that states are SU(2) singlets which is something nonphysical. For covering spaces one would however obtain the degrees of freedom associated with the discrete fiber and the degrees of freedom in question would not disappear completely and would be characterized by the discrete subgroup of SU(2).
For anyons the nontrivial homotopy of plane brings in nontrivial connection with a flat curvature and the nontrivial dynamics of topological QFTs. Also now one might expect similar nontrivial contribution to appear in the spinor connection of M^{2}×G_{a} and CP_{2}×G_{b}. In conformal field theory models nontrivial monodromy would correspond to the presence of punctures in plane.
 For factor spaces the unit for quantum numbers like orbital angular momentum is multiplied by n_{a} resp. n_{b} and for coverings it is divided by this number. These two kind of spaces are in a well defined sense obtained by multiplying and dividing the factors of H by G_{a} resp. G_{b} and multiplication and division are expected to relate to Jones inclusions with M:N< 4 and M:N=4, which both are labelled by a subset of discrete subgroups of SU(2).
 The discrete subgroups of SU(2) with fixed quantization axes possess a well defined multiplication with product defined as the group generated by forming all possible products of group elements as elements of SU(2). This product is commutative and all elements are idempotent and thus analogous to projectors. Trivial group G_{1}, twoelement group G_{2} consisting of reflection and identity, the cyclic groups Z_{p}, p prime, and tedrahedral, octahedral, and icosahedral groups are the generators of this algebra.
By commutativity one can regard this algebra as an 11dimensional module having natural numbers as coefficients ("rig"). The trivial group G_{1}, twoelement group G_{2} generated by reflection, and tedrahedral, octahedral, and icosahedral groups define 5 generating elements for this algebra. The products of groups other than trivial group define 10 units for this algebra so that there are 11 units altogether. The groups Z_{p} generate a structure analogous to natural numbers acting as analog of coefficients of this structure. Clearly, one has effectively 11dimensional commutative algebra in 11 correspondence with the 11dimensional "halflattice" N^{11} (N denotes natural numbers). Leaving away reflections, one obtains N^{7}. The projector representation suggests a connection with Jones inclusions. An interesting question concerns the possible Jones inclusions assignable to the subgroups containing infinitely manner elements. Reader has of course already asked whether dimensions 11, 7 and their difference 4 might relate somehow to the mathematical structures of Mtheory with 7 compactified dimensions.
 How do the Planck constants associated with factors and coverings relate? One might argue that Planck constant defines a homomorphism respecting the multiplication and division (when possible) by G_{i}. If so, then Planck constant in units of hbar_{0} would be equal to n_{a}/n_{b} for H/G_{a}× G_{b} option and n_{b}/n_{a} for H×(G_{a}× G_{b}) with obvious formulas for hybrid cases. This option would put M^{4} and CP_{2} in a very symmetric role and allow much more flexibility in the identification of symmetries associated with large Planck constant phases.
For more details see the chapter Does TGD Predict the Spectrum of Planck Constants?.
