What is new with Koide sum rules?

arivero

Ok I can find two kinds of works where the square root of yukawa coupling has some fundamental role. One is "flavons", as in http://arxiv.org/abs/1203.1489v3 Other is Composite Higgs, and particularly Contino (eg http://arxiv.org/abs/1005.4269v1) works like to make explicit this. My memory fails me, so perhaps Mitchell has already catalogued other cases.

Edit:oh, also the higgs in NCG seems to be a product of two more elementary numbers, from bialgebras or elsewhere

mitchell porter

We had two NCG higgs papers this month, so perhaps we can discuss this.

In Estrada and Marcolli, we have an action with parameters "a ... e" that are functions of the Yukawa matrices, and some other "f" parameters that are more fundamental, and then we have various relations among a ... f that exist at unification energy.

This action is an expansion of a "spectral action functional" which includes variation of a Dirac operator. The Higgs field arises from those "inner fluctuations" of the Dirac operator that are associated with the finite noncommutative part F of the "almost commutative" space M x F that defines the model. (Inner fluctuations associated with M produce the gauge fields.) These fluctuations have the form u[D,u*], where [,] is a commutator and u,u* are I think unitary elements of the algebra A in the spectral triple <A,H,D>. Alejandro, are u and u* what you were talking about?

The new work this month has been about obtaining a 125 GeV Higgs in the noncommutative SM, either by imposing asymptotic safety and reproducing the Shaposhnikov-Wetterich argument (this is what Estrada and Marcolli did), or by including the scalar that gives Majorana mass to the RH neutrino, in the RG equations (this is what Chamseddine and Connes did). The AS argument gives the right value within a GeV; the other model just shifts the range of possible Higgs masses so that it includes the observed value.

Turning to the world of Koide relations, there have been a few studies of how the values of the expression in the Koide formula flow, for different triplets; and we also have the work of Sumino, which imposes boundary conditions on RG flow at the EWSB scale, in order to explain the exactness of the Koide relation for the pole masses.

So the obvious way to explain the Koide formula in the NCG context, would be to use a high-scale (unification-energy) version of the Sumino mechanism, that employs the "a...f constraints" to engineer the necessary low-energy relations. I'm not sure if this is possible, but if it will jumpstart discussion again, I'm willing to think about it...

S.Daedalus

This whole discussion is way above my level of expertise in high-energy physics, but I have a side question if I may: if I understood correctly, one of the main problems with the Koide approach is that it's a connection between the low-energy masses of the theory, which should have no fundamental significance due to renormalization group flow. But couldn't there be something like supersymmetry nonrenormalization at work, that is, the parameters m that turn up in the low energy theory are actually identical to the high energy masses?

mitchell porter

Actually those nonrenormalization properties disable the Sumino mechanism for preserving the Koide relation, because it relies on vertex corrections that no longer exist under susy! Koide and Yamashita developed an alternative but it doesn't work as well. Still, perhaps one can hope that susy will simplify the RG equations in some other way.

Some resources.

mitchell porter

Another useful NCG paper is Kolodrubetz & Marcolli. Also see this lecture, especially slide 10. It seems that one wants to construct a cascade of effective field theories, with a Sumino model at the final stage.

Returning to comment #65... The original Koide triplet relates yukawas from a single mass matrix, but the new triplets for quarks all combine up-type yukawas with down-type yukawas, so the transformation looks unnatural. It's as if we need an extended Higgs mechanism that includes "up-down yukawas". We could suppose they are there and set them to zero... but what would they be? The Standard Model mass matrices tabulate coefficients of Yukawa terms in the Lagrangian. These new "up-down Yukawa terms" would require something new.

Nonetheless:

[tex]\left( \begin{array}{ccc}
y^u_{11} & 0 & y^u_{12} & 0 & y^u_{13} & 0 \\
0 & y^d_{11} & 0 & y^d_{12} & 0 & y^d_{13} \\
y^u_{21} & 0 & y^u_{22} & 0 & y^u_{23} & 0 \\
0 & y^d_{21} & 0 & y^d_{22} & 0 & y^d_{23} \\
y^u_{31} & 0 & y^u_{32} & 0 & y^u_{33} & 0 \\
0 & y^d_{31} & 0 & y^d_{32} & 0 & y^d_{33} \\ \end{array} \right)[/tex]

... if I may be permitted to introduce this interleaving of up and down Yukawa matrices, without exactly saying what it is; and if we suppose that the "up" and "down" parts are each diagonalized as much as possible, with diagonal entries ordered by size; then the Koide waterfall amounts to saying that there is a "Brannen symmetry" for each 3x3 block on the main diagonal.

edit: Whoops, I missed a stage. The Brannen symmetry relates the square roots of the masses. So we would be looking at blocks on the diagonal of a 6x6 matrix whose square is the matrix above.

edit #2: The Brannen transformation for a particular block could look like this:
[tex]\left( \begin{array}{ccc}
1 & 0 & 0 & 0 & 0 & 0 \\
0 & \sqrt{2} & e^{i\delta} & e^{-i\delta} & 0 & 0 \\
0 & e^{-i\delta} & \sqrt{2} & e^{i\delta} & 0 & 0 \\
0 & e^{i\delta} & e^{-i\delta} & \sqrt{2} & 0 & 0 \\
0 & 0 & 0 & 0 & 1 & 0 \\
0 & 0 & 0 & 0 & 0 & 1 \\ \end{array} \right)[/tex]

mitchell porter

There's a new paper on Koide triplets today, in which the author experiments with a Brannen parametrization of up-type masses and down-type masses, and comes out with phases of 2/27 and 4/27. The usual Brannen phase for the charged leptons is 2/9, i.e. 6/27. These are numbers which I first saw on Marni Sheppeard's blog, and which I thought she discovered through discussion with Dave Look, so I'll be writing to the author to let him know - as well as to mention the tbcsud "waterfall" of triplets discussed in this thread. I think of the waterfall as real, and tend to dismiss those quark family triplets as spurious. Given the idea that there are unknown "Koide symmetries" responsible for the "authentic triplets", I suppose it's possible that the same symmetries could be present in uct and dsb too, but with a lot more noise.

arivero

Of course I am of the same opinion, but still I am sligthly amused that the waterfall uses delta_scb = 3 delta_L, and he gets delta_L= 3 delta_U.

mitchell porter

Or to put it another way, the scb angle is 2/3, the eμτ angle is 2/9, and the uct angle is 2/27.

We also have that the eμτ mass scale is 313 MeV (one-third the proton mass, i.e. constituent mass of a first-generation quark), and the scb mass scale is three times that.

From Sheppeard's blog (1 2), I get that the mass scale for a uct triplet would be about 20 GeV.

edit: A few months ago I was thinking about what sort of model would produce just these "family phases" - what Zenczykowski calls δ_L, δ_D, δ_U - simply because that's easier to think about. I was interested in an Adler-type 3HDM (three-Higgs-doublet model) with circulant mass matrices. But you could take any model of the charged-lepton sector, that produces a Koide relation, and try to apply it separately to the up-type and down-type quarks - for example, Ernest Ma's supersymmetric model.

arivero

Of course, it is very problematic to have angles which are not a submultiple of the circumference. Up to now, the main motivation for the factor of three was to consider the case where one of the masses is 0, fixing thus the angle, and then the orthogonality between the triples with 15 degrees and 45 degrees, this is pi/12 and pi/4. The angle of 2/3 I though of it as a perturbation from pi/4, the angle of 2/9 as a perturbation from pi/12.

mitchell porter

Since we don't know where these quantities come from, I don't think we can say that their form is problematic. Would their origin be easier to understand if they were simple fractions of π? Also, it's hard to think of e.g. 2/9 (the actual phase, for e-μ-τ) as a perturbation of π/12 (the phase for e-μ-τ, in the "modified waterfall" that lands on Harari-Haut-Weyers values for d-u-s masses), because normally a perturbation of a quantity x just gives you "x plus a small mess", it doesn't give you a simple rational number! I have noticed that 2/3 (possible phase for s-c-b) is obtained by the first two terms in the Leibniz formula for π/4, as if it were a truncation. One could start thinking about formulas with Grassmann variables, so all the higher terms vanish...

Another line of investigation would be to look for the significance of the "Brannen angle" in the other frameworks that manage to produce Koide triplets. Sumino, in his paper which tries to explain the exactness of the original Koide formula despite RG running, also presents an original derivation of the triplet itself (from the interactions in the scalar sector of his model). Then there's Ma, mentioned above; then there's Koide's original preon theory. Carl Brannen's formula plays no apparent role in any of these, but I wonder if they still look simple when expressed using his variables?

edit: Some comments on whether m_u=0 is still a live option. (For the general reader of this thread: Alejandro found a "waterfall" of interlocking Koide triplets which works well for the four heaviest quarks and which can be extended to the remaining quarks. The modified waterfall is a version adjusted so that the up quark has exactly zero mass. The heavy quark masses become less accurate but the Brannen angles assume interesting values, and the idea is that the real waterfall is a perturbed version of this modified waterfall, see his paper for details.)

Michael Dine gave a talk as recently as 2009 implying that it was still being considered by theorists like Seiberg and Kaplan. Dine's 1993 review "Topics in string phenomenology" points out two ways to get m_u=0 from string theory, one from anomalous discrete symmetry, the other from a horizontal symmetry as described in a series of papers (1 2 3).

From the other side, 1103.3304 gives in a few sentences (page 83) the reason why workers in lattice QCD might dismiss the m_u=0 option as an explanation for no strong CP violation. This argument needs to be confronted with the ideas in reference 3, listed above.

arivero

I am curious about how sensible the prediction of the top mass is to the factor 3 in the jump from leptons to quarks. So here is the "bc -l program"

Newcomers can see that the same formula is used for mtop and mtau. And of course it is also the same formula, in angular form, for the rest of the calculation involving delta and m. Koide everywhere.

so with the factor 3 argued in my paper, we get
Which is in the center of the combination of Tevatron (173.18) and LHC (173.34). In fact, the weighed average of Tevatron and CMS (september) should be 173.265 ± .679 GeV, so the prediction is pretty in the center.

Which are the 1 sigma limits for the mass and angle factors, with this average? Well, pretty narrow, but still some place for perturbative corrections:

arivero

Probably it is a red herring, but some comments from mitchell have indirectly driven me to look at the mass formula for an stack of D-branes. I am not sure in how they are in the superstring case, but already in the bosonic string they look a lot as a generalisation of Koide mixing:

[tex]
M^2 = \big((n + {\theta_i - \theta_j \over 2 \pi}) {R' \over \alpha'}\big)^2 + {N-1 \over \alpha'}
[/tex]

THis is f. 174 in arXiv:hep-th/0007170v3

For n=1, N=1, and i,j from 1 to 3 with i different of j, the stack of three D-branes looks Koide's formula. I am not sure of which is the mass formula in this case (nor in the fermionic/superstring case...)
It should be, if M^2 where instead a seesawed product of two masses, [itex]M^2=m_{ij} M_0[/itex]


[tex]
m_{ij} = {R'^2 \over M_0 \alpha'^2} (n + {\theta_i - \theta_j \over 2 \pi})^2
[/tex]

Note that the basic fact is that the sum of the three differences [itex]\theta_i - \theta_j [/itex] is zero, as in the case of the sum of three cosines in Koide.

EDIT: Michael Rios suggested, last year, to use three coincident branes to emulate Koide.
EDIT2: Today is the birthday of Lubos Motl, this is my birthday gift: string theory becomes predictive

arivero

Unrelated to the previous comment, except for the fact that strings dof come in groups of 8, it could be worthwhile to rethink again the 12x8 ideas in the light of Koide. The "Koide waterfall" in #58 above, with mass of up quark exactly zero in three Koide steps, provides, if we also use the orthogonality condition, some intriguing pairing of leptons and quarks:

t:174.10 GeV
b:3.64 GeV
tau, c:1.698 GeV
mu, s:121.95 MeV
e, u:0
d:8.75 MeV

On other hand, the most naive way of building a multiplet with 8 degrees of freedom is to use an electroweak pair of Dirac fermions: neutrino, electron for instance, or any up, down combination. This is still possible here, and even it could be convenient if we consider that we are going to broke this pairing of leptons and quarks. But looking at this table, we could take it serioustly and consider that one lepton and the three colours of a quark should be the components of a multiplet. Then the unpaired quarks would correspond to see-sawed neutrinos and the whole table is

Of course we have sixteen degrees of freedom in each line and it is still to see how they should be managed in groups of 8, either by chirality or by particle/antiparticle.

What is intriguing in any case is the mu,s pairing: a charged lepton with a down type quark. It could point to the need of using a SU(2)xSU(2) L-R symmetry.

mitchell porter

Pati-Salam as we know it, doesn't allow such a scheme. The orthodox way to embed the waterfall in Pati-Salam would be to use the conventional generation structure (three sets of two "four-color quarks", one of which divides into an up-type quark and a neutrino, the other of which divides into a down-type quark and a charged lepton), a selection of Higgses (there must at least be one to break SU(4)_c to SU(3)_c and another to break U(1)_B-L x SU(2)_R to U(1)_Y), then work out the 3x3 Yukawa matrices for the "four-color quarks", and finally the effective Yukawas for the SM quarks and leptons. And since the waterfall has that intricate structure, probably the best way is via flavons: the Yukawas are VEVs of "flavon" fields. (Koide himself uses flavons in his yukawaon models of recent years.) We can then try to obtain a Pati-Salam waterfall from flavon symmetries.

This doesn't have the simplicity of just directly associating e-mu-tau with u-s-c (or with s-c-b, as might have been suggested by Georgi-Jarlskog), but at least it is a type of theory which it is known can be constructed. If you do it this way, the orthodox way, you do get to preserve the direct association of muon with strange quark. So you might suppose that the second generation is a sort of pivot, where there is approximate equality of masses, connecting waterfall Koide triplets on the quark side, and the usual family Koide triplets on the lepton side.

Intriguingly, if you imagine interleaving the Yukawas for charged and uncharged leptons in the fashion of #73, then the Brannen transformation matrix for family Koide triplets (rather than sequential, waterfall triplets) looks like this:
[tex]\left( \begin{array}{ccc}
1 & 0 & 0 & 0 & 0 & 0 \\
0 & \sqrt{2} & 0 & e^{i\delta} & 0 & e^{-i\delta} \\
0 & 0 & 1 & 0 & 0 & 0 \\
0 & e^{-i\delta} & 0 & \sqrt{2} & 0 & e^{i\delta} \\
0 & 0 & 0 & 0 & 1 & 0 \\
0 & e^{i\delta} & 0 & e^{-i\delta} & 0 & \sqrt{2} \\ \end{array} \right)[/tex]
(or the obvious counterpart where the two interleaved blocks change places).

As I wrote in #73, an "interleaving of Yukawa matrices" has no physical meaning that I can identify. But what I like about this perspective is that the "family Brannen transformation" and the "sequential Brannen transformation" could both plausibly be part of some larger algebraic structure. In both cases they're based on a 3x3 block within the 6x6 matrix, it's just that the spacing is different. So the idea is that a Pati-Salam embedding of waterfall + original Koide could result from a flavor symmetry containing that "larger algebraic structure", with family symmetries dominating on the lepton side and sequential symmetries dominating on the quark side, and with the second generation providing the bridge.

arivero

I have scanned a couple of collections which show the history of Koide before Koide. The first set pivotes on Harari-Haut-Weyers and its refutations, the second set is some extra articles of the same age, found while I explored the first selection.

While they will be useful mostly to Carl Brannen, perhaps Mitchell and other crowd can enjoy them too.

https://docs.google.com/open?id=1Ufl...qkMlReePUCCGDj

https://docs.google.com/open?id=1vRf...PnIwW634ZOhATP

And yes, I use a monitor which can pivote 90 degrees. But nowadays you can always cancel the gravitational sensor of your iPad, can you?

arivero

Wow, Zenczykowski paper was accepted for PhysRev D last Thursday (Dec 13, 2012).

If it is still v2, I am a bit sorry that he has not changed the references (for instance, to include the waterfall) but still a good thing.

arivero

Just for the record, it is interesting to look to the solutions in the lepton side ascending from the e-mu-tau triple. Remember we are conjecturing a descent where some leptonic object partners with every quark. It could be reasonable to think of Dirac mass terms for neutrinos, for instance.

-, t:174.10 GeV
-, b:3.64 GeV
tau, c:1.698 GeV
mu, s:121.95 MeV
e, u:0
-, d:8.75 MeV

Now, once we have broken the pairing, we can use Koide separately in each sector... just to see if it has some sense.

To ascent from mu, tau to the next two levels, the equation with the above values has discriminant cero, we should look with some care with branch of the answer is it really taken, but anyway here you have both branches. For both of them, the second step is unique, due to negative roots forbidding other solutions.

I like this second branch: the (tau, nu1, nu2) triplet equal to (1.777, 1.813, 121.95). It could be saying that the lepton sequence moves to increase the gap between the two final states. And while we left the 174.1 GeV endpoint, we still are in a nice mass range.