Quantum criticality is the TGD counterpart of the inflation and the flatness of 3space follows from the condition that no local dimensional quantities are present in 3geometry. Also the imbeddability fo M^{4} is an important piece of story and restricts the set the parameters of imbeddable cosmologies dramatically.
One can try to understand the situation microscopically in terms of the cosmic strings which gradually develop higher than 2D M^{4} projection during cosmic evolution and become magnetic flux tubes carrying magnetic monopole fluxes explaining the presence of magnetic fields in cosmology.
At microscopic level magnetic flux tubes are the key structural elements. The phase transitions increasing Planck constant for the matter associated with flux tubes and thus also the lengths of magnetic flux tubes should be important as also the phase transitions increasing padic prime and reducing Planck constant originally emerged in the modelling of TGD inspired quantum biology are highly suggestive. First transitions would mean adiabatic expansion with no heat generation and latter transitions would liberate magnetic field energy since flux conservation forces field strength to be reduced and leads to liberation of magnetic energy producing ordinary matter and dark matter. Dark energy in turn is identifiable as magnetic energy.
The key question concerns the mechanism causing the isotropy and homogeny of the cosmology. There are two possible identifications.
 According to two decades old TGD proposal primordial cosmology before the emergence of spacetime sheets could be regarded as string gas in M^{4}_{+}× CP_{2} at Hagedorn temperature determined by CP_{2} radius: T_{H}∼ hbar/R(CP_{2}). This phase could be present also after the transition to radiation dominated cosmology and consist of strings, whose thickness is gradually increasing and which contain carry dark energy and dark matter. The horizon radius is infinite for this cosmology thus providing at least partial explanation for the homogeny and isotropy and visible matter would represent deviations from it.
 The accelerating expansion period towards the end of the critical period could smooth out inhomogenities and thus provide an additional mechanism leading to homogenous and isotropic Big Bang. This for given spacetime sheet representing RW cosmology: in manysheeted cosmology one can imagine distribution of parameters for the cosmology. The rapid expansion period could however also develop large fluctuations! Indeed, the time a_{F}<a_{1} (density would be infinite for a_{1}) for its end  and therefore local mass density  must have a distribution after the rapid expansion ends. This expansion would generate separate smoothed out radiation dominated spacetime sheets with slightly different mass densities and cosmic temperatures. A splitting to smooth radiation dominated subcosmologies would take place.
Therefore TGD scenario could be very different from inflationary scenario. The problem is to decide which option is the most feasible
one.
The formulas used to make back of the envelope (see this) calculations in inflation theory discussed in a guest posting in Lubos's blog given some idea about TGD counterpart for the generation of gravitons. Inflationary period is replaced with essentially unique critical cosmology containing only its duration as a free parameter. The fluctuations in the duration of this parameter explain scalar temperature fluctuations assoiated with CMB.
How the local polarization of CMB is generated?
There is a nice discussion about the mechanism leading to the generation of CMB polarization
(see this). The polarization is generated after the decoupling of CMB photons from thermal equilibrium and is due to the scattering of photons on free electrons during decoupling. This scattering is known as Thomson scattering. The page in question contains schematic illustrations for how the polarization is generated. The scattering from electrons polarizes the photons in direction orthogonal to the scattering plane. In thermal equilibrium the net polarization of scattered radiation vanishes. If however the scattered photons from two perpendicular directions have different intensities a net polarization develops.
Polarized photons could be produced only during a short period during recombination scattering from free electrons was still possible and photons could diffuse between regions with different temperature. Polarized photons were generated when electrons from hot and cold regions where scattering on same electrons. CMB polarization indeed varies over sky but not in long length scales since photons could not diffuse for long lengths.
So called quadrupole anisotropy of CMB temperature contains information about the polarization. There are three contributions: scalar, vector, and tensor.
 Scalar contributions is due to density fluctuations reflecting themselves as temperature fluctuations and does not distinguish between polarizations: this is what has been studied mostly hitherto. A natural TGD mechanism for their generation would be different time for the end of the critical period leading to splitting of critical cosmology to radiation dominated cosmologies with slightly different temperatures.
 There is also so called vorticity distribution due to the flow which has vorticity and would due to defects/string like objects present also in TGD. The simplified situation corresponds to are region in which one has two flows in opposite direction locally. Depending on whether the scattering photons are upstream or down stream they are blueshifted or redshifter so that the temperatures are slightly different in upstream and down.The flows in opposite direction give rise to a situation in which photons with different temperatures scatter and produce polarization. The effects of vorticity are expected to disappear during the fast expansion period. Probably because the gradients of velocity giving rise to vorticity are smoothed out.
 The third contribution is tensor contribution and due to gravitons generating stretching and squeezing of space in two orthogonal directions defining polarization tensor. Stretching increases wavelengths and decreases temperature. Squeezing does the opposite. Therefore temperature differences distinguishing between the two directions are generated and the outcome is polarization of the CMB background much later. This corresponds to the so called E and B modes.
One can decompose polarization as vector field to two parts: the first one  the Emode  is gradient and thus irrotational and second is curl and thus rotational and with vanishing divergence (incompressible liquid flow is a good concrete example).
How the polarization anisotropies could be generated in TGD Universe?
One can try to understand microscopically how the polarization anisotropies are generated in TGD framework using poor man's arguments.
 One can introduce a vision vision about fractal 3D network of cosmic strings forming a kinds of grids with nodes in various scales. These grids would be associated with different levels of the hierarchy of spacetime sheets associated with manysheeted spacetime. Coordinate grid is of course an idealization since three coordinate lines would meet in single node. A weaker form of grid would involve meeting of two coordinate lines at given node. There is data about our own galactic nucleus understood if it correspond to the node at which two magnetic flux tubes meet. Ordinary visible matter would be generated in nodes.
One might say that galaxies are due to traffic accidents in which dark matter arriving along two cosmic strings collides in the crossing of the roads. Flux tubes would be attracted together by gravitational attraction so from the crossing.
 Amusingly, the notion grid emerged also TGD inspired quantum biology as a proposal for how living system codes morphogenetic position information. Flux tubes carry dark matter and ordinary matter is associated with the nodes at which coordinate lines meet each other. This web can give rise to a generalization of topological quantum computation using 2braids. Coordinate lines define strings which can be knotted in 3dimensions and define braids making possible topological quantum computation using macroscopic quantum phases defined by the dark matter. The time evolutions of coordinate lines defines string world sheets and in 4D spacetime the string world sheets can be knotted and braided so that also higher level TQC becomes possible with string reconnection and going above or below the other define two bits in each node.
 The presence of grid could also explain the honeycomb like structure of Universe with the recent typical size of honeycomb about 10^{8} ly.
 In this framework the illustrations for how the gravitational waves induce the polarization of CMB. The radiation beams entering from opposite directions can be assigned with two magnetic flux tubes meeting at the node and in slightly different temperatures due to the interaction with gravitons much earlier. The gravitons can be regarded as larger spacetime sheets at which the two flux tubes had contacts so that space associated with the flux tubes was forced to stretch or squeeze. This in turn increased of reduced photon wavelength so that photon temperature at flux tubes was different and the difference were preserved during subsequent evolution.
Back on the envelope calculations in TGD framework
One can modify the back on the envelope calculations of John Preskill
(see this) in Lubos's blog to see what could happen in TGD framework. Now one however starts from the critical cosmology fixed apart from its duration and looks what it gives rather than starting from Higgs potential for inflaton field. The obvious counterpart for inflaton scalar field would be magnetic field intensity having same dimension but one should avoid too concrete correspondences.
The key question is whether the critical period generates the rapid expansion smoothing out inhomogenities or whether it generates them. The original guess that it smooths them out turns out be wrong in closer examination.
 The basic equation in inflationary model is given by
(da/dt)^{2}= V/m_{p}^{2}
If V is small this has as solution a(t)= a(0)exp(Ht) if H= V^{1/2}/m_{p} is constant. De Sitter cosmology allows partial imbedding in TGD but the imbedding is naturally static and has interpretation as blackhole interior with constant mass density. One can find coordinates in which the solution looks like expanding cosmology without Big Bang but these coordinates are not natural from the view of imbedding space.
 In TGD the expression for da/dt for critical cosmology is
da/dt= [a_{0}^{2}a^{2}]^{1/2}/[a_{0}^{2}R^{2}a^{2}]^{1/2} .
a_{0} is roughly the duration of cosmology and R is CP_{2} radius of order 10^{3.5} Planck lengths. The almost uniqueness follows from the condition that the imbedding is such that the induced metric at the 3surfaces defined by intersections with hyperboloids of M^{4}_{+} is flat rather than hyperbolic. This cosmology differs from deSitter cosmology.
 For a→ 0 one has
da/dt ≈ a_{0}^{2}/[a_{0}^{2}R^{2}]≈ 1 .
so that one has da/dt ≈ 1 and a≈ t for small values of a in accordance with the replacement of Big Bang with a "silent whisper amplified to a Big Bang" (density of matter goes as 1/a^{2}) Hubble constant goes like H∝ 1/a so that Hubble radius divergence. This does not guarantee that horizon radius becomes infinite. Rather, the horizon is finite and given in good accuracy by the duration a_{1}=[a_{0}^{2} R^{2}]^{1/2} of the period. One can however explain the isotropy and homogenity of the string gas in lightcone M^{4}_{+} carrying flux tubes carrying dark matter and energy in terms of the infinite horizon of M^{4}.
There is no exponential time evolution at this period since one has a≈ t in good approximation for a/a_{0}<<1. The TGD counterpart of V would behave like 1/a^{2}, which conforms with the idea that V corresponds to energy density.
 As the limit a→ a_{1}=[ta_{0}^{2}R^{2}]^{1/2} is approached, the expansion rate approaches infinite and for a>a_{1} at the latest one expects radiation dominated cosmology: otherwise a region of Euclidian signature of the induced metric results. The expectation is that a transition to radiation dominated cosmology takes place before a=a_{1} at which also energy density would diverge. The question is whether this period means smoothing out of inhomogenities or generation of them or both.
Consider now what could happen near the end of the Minkowskian period of critical cosmology.
 Although it is not clear whether rapidly accelerating expansion is needed to to smooth out homogenities, one can just find what conditions this would give on the parameters. For a_{i}= kR at which phase transition began the condition that a was increased at least by factor e^{50}∼ 5× 10^{21} (50 efolds) this would give a_{1}≈ a_{0}>e^{50}kR. For k∼ 1 this gives something like 10^{18} seconds, which happens to correspond atomic length scale. Below it will be found that this period more naturally corresponds to the period during which large fluctuations in density distribution and metric are generated.
 The earlier estimate for the emergence of radiation dominated cosmology assumed that the transition to radiation dominated cosmology takes place at CP_{2} temperature defining Hagedorn temperature at which temperature of the string gas cannot be raised anymore since all the energy goes to the generation of string excitations rather than to kinetic energy, gives a_{F}∼ 10^{10} seconds, which is by factor 10^{8} larger. If this were true, the fast expansion period a_{F} would increase the scale factor to about 68 efolds equivalent to 98 2folds. pAdic prime p≈ 2^{196} would correspond to padic length scale about L(196) ∼ .1 meters. The crucial assumption would be that the the time a_{F} at which the expansion ends is same everywhere. There is no reason to assume this and this would mean that the period in question generates inhomogenities and isotropies of mass distribution and temperature distribution.
Note that if the distribution of the time a_{F}<a_{1} at which the critical period ends is responsible for the CMB fluctuations then the number of foldings characterizes the smoothness of given local radiation dominated cosmology and could be rather large.
 The rapid accelerating expansion occurs as g_{aa} approaches zero. Indeed, for
a→ a_{1}= [a_{0}^{2}R^{2}]^{1/2}
a very rapid expansion occurs and da/dt approaches infinite value. Near to a_{1} one can write a/a_{1}=1δ and solve δ approximately as function of t as
δ =[3R^{2}/4a_{1}^{2}]^{2/3} [tt_{1}/a_{1}]^{2/3} ,
t_{1}= ∫_{0}^{a1} [1a^{2}/a_{1}^{2}1/2/[1a^{2}/a_{1}^{2}]^{1/2} .
Hubble constant behaves as
H= (da/dt)/a = (R^{2}/2a_{1}^{3})δ^{1/2} .
 What is interesting is that applying the naive dimensional estimate for the amplitude of gravitational fluctuations to be δ h_{T}^{2}∼ H^{2}/m_{P}^{4}. This would mean that at the limit a→ a_{F}< a_{1} gravitational fluctuations become very strong and generate the strong graviton background. Same applies to fluctuations in mass density.
Summary
The possibility of very rapid expansion near a=a_{F}<a_{1} leading to radiation dominated cosmology should have some deep meaning. The following tries to catch this meaning.
 The explosive period could lead to a radiation dominated cosmologies from string dominated cosmology with Hagedorn temperature. It could involve h_{eff} increasing phase transitions for string gas during the initial period and liberation of magnetic energy during the end period as massless particles: this would explain why the mass density of the spacetime sheet increases dramatically. The critical cosmology could correspond to a phase transition from a phase with Hagedorn temperature identified as T_{H}∝ hbar/R_{H} to radiation dominated cosmology.
 The cooling of string gas would lead to the generation of hierarchy of Planck constants and liberation of the magnetic energy of strings as massless particles during the end of critical period topologically condensing to spacetime sheets such as massless extremals. This process could correspond to the rapid increase of energy density towards the end of the critical period.
 Isotropy and homogenity appear both at the level of imbedding space and spacetime sheets. The infinite horizon of M^{4}_{+} would explain the isotropy and homogenity of string gas in H both before and after the emergence of spacetime sheets at Hagedorn temperature around a∼ R(CP_{2}). In particular, the smoothness of the cosmology of dark matter and dark energy would find explanation. The rapid expansion would in turn smooth out inhomogenities of individual spacetime sheets.
 The Hubble scale 1/H approaches to zero as a=a_{F}<a_{1} is approached. The rapid expansion destroys anisotropies and inhomogenities of radiation dominated spacetime sheet corresponding to particular value of a_{F}. The distribution for values of a_{F} in turn explains CMB scalar fluctuations since the energy density in final state is highly sensitive to the precise value of a_{F}. This distribution would be Gaussian in the first approximation. One can say that the fluctuation spectrum for inflaton field is replaced with that for a_{F}.
 Also the generation of gravitational radiation and its decoupling from matter could take place during the same end period. After this gravitational fields would be essentially classical and assignable to spacetime sheets. Essentially formation of gravitationally bound states would be in question analogous to what happens photons decouple from matter much later. The reduction of the temperature of string gas below Hagedorn temperature could generate also the massless graviton phase decoupling from matter and inducing the temperature fluctuations and polarization during decoupling.
Gravitons and also other particles would topological condense at "massless extremals" (MEs,topological light rays) and particles  in particular photons  would interact with gravitons by generating wormhole contacts to gravitonic MEs. The interaction between MEs assignable to gravitational radiation and photons would have caused the fluctuations of CMB temperature.
To sum up, if the TGD inspired picture is correct then Penrose would have been correct in the identification of string theory as fashion and inflationary cosmology as fantasy ( Lubos has reacted strongly to this). Also the fact that inflationary cosmology is at the verge of internal contradiction due the fact that the assumption of field theoretic description is in conflict with the large graviton background suggests that inflationary cosmology is not for long with us anymore.
For details see the chapter TGD and Cosmology or the article BICEP2 might have detected gravitational waves.
