NBI-HE-92-52

July 1992

Topography of the hot sphaleron Transitions

arXiv:hep-lat/9207020v1 20 Jul 1992

J. Ambjørn and K. Farakos1

The Niels Bohr Institute

Blegdamsvej 17, DK-2100 Copenhagen Ø, Denmark

Abstract

By numerical simulations in real time we provide evidence in favour of sphaleron like

transitions in the hot, symmetric phase of the electroweak theory. Earlier performed
observations of a change in the Chern-Simons number are supplemented with a
measurement of the lowest eigenvalues of the three-dimensional staggered fermion
Dirac operator and observations of the spatial extension of energy lumps associated
with the transition. The observations corroborate on the interpretation of the change
in Chern-Simons numbers as representing continuum physics, not lattice artifacts.
By combining the various observations it is possible to follow in considerable detail
the time-history of thermal fluctuations of the classical gauge-field configurations
responsible for the change in the Chern-Simons number.

1
Supported by an EEC Fellowship.
Permanent address: National Technical University of Athens, Depart. of Physics, GR 157 73,
Athens, Greece.

1
1 Introduction
It is by now well known that the fermion number is not conserved in the standard
electroweak theory due to the anomaly of the fermionic current and the periodic
vacuum structure in the non-abelian gauge theories [1]. Although the amplitude
for this process is exponentially small at zero temperature a great amplification
can occur at high temperature [2]. The reason is that the fermion-number violating
processes at zero temperature can occur only by tunneling from one classical vacuum
to another, the two being connected by a large gauge transformation with winding
number different from zero. At higher temperature it is possible to move between
the different gauge vacua by classical thermal fluctuations. The energy barrier which
separates the vacua has a height determined by the sphaleron energy Esph , which is
the energy of the static classical solution to the electroweak theory corresponding to
the lowest lying saddle point between two neighbour vacua [3]. If the temperature
T is so “low” that the symmetry is broken we have
2Mw (T )
Esph = B(λ/αw ) (1.1)
αw
where B(λ/aw ) is a factor of order one, which varies only slowly when the ratio
between the Higgs coupling λ and αw = gw2 /4π changes from zero to infinity. In
(1.1) we have included a possible temperature dependence in the W -mass Mw . As
long as x ≡ Esph /T << 1 we can trust a one loop calculation and we get that
the transition probability per volume and time for moving from one vacuum to a
neighbour one is given by [4, 5]

Γ = 0.007(αw T )4 x7 e−x . (1.2)

This calculation is strictly speaking only valid as long as the Boltzmann factor
−x
e dominates the prefactor x7 from zero modes of the sphaleron. It is natural to
expect that (1.2) extrapolates to

Γ = κ(αw T )4 (1.3)

when the symmetry is restored (Mw (T ) = 0) and x is formally zero [4, 6, 7], but
obviously we can get no information about the non-perturbative constant κ from
(1.2) and analytical methods have until now failed in the symmetric phase due
to infrared divergences in the high temperature expansion. The constant κ was
determined by real time computer simulations of the SU(2) gauge-Higgs system
and the result was that κ ∼ O(1) [8].
The cosmological implication of κ ∼ O(1) is that any B + L, the baryon plus
lepton number, generated at GUT temperature, will be washed out before we reach

2
the electroweak transition. In theories where B−L are strictly conserved the problem
of explaining the baryon asymmetry observed in the universe is then pushed to the
electroweak transition temperature.
Since the implications of κ ∼ O(1) are important we have felt a need to corrobo-
rate on the claim in [8] that the configurations observed by computer simulations are
really to be identified with continuum like configurations which change the Chern-
Simon numbers by units of one. The only thing done in [8] was to measure the
“naive” lattice definition of 0t d3 xF F̃ as a function of time when it developed ac-
R R

cording to the classical equations of motion, starting with a hot configuration above
the electroweak transition temperature Tc . Whenever a change in the Chern-Simons
number compatible with unity and followed by some kind of plateau was observed it
was classified as a “sphaleron” like transition2 . The rather rapid change of Chern-
Simons number observed in [8] could potentially be due to lattice dislocations rather
than genuine extended field configurations with a “topological” interpretation. We
have in this paper tried to verify the “continuum nature” of the lattice configurations
in two ways: By measuring the eigenvalues En (t) of the three dimensional Dirac op-
erator at time t and observing that one of them dives to zero within the time period
where the Chern-Simon number changes rapidly by one unit, and further by measur-
ing the energy density as a function of time for the gauge field configurations. The
approach is from this point of view similar to the work analyzing the monopole-like
configurations in lattice QCD [11, 12], since these should also be thought of as three
dimensional configurations.

2 The model and the simulation

The main approximation in the present work is the use of purely classical thermo-
dynamics. It is based on the observation that the energy of the sphaleron is of order
Mw /αw while its extension is of order Mw−1 . For T > Tc we expect similar character-
istica of the sphaleron-like configurations: their energy will be of order T but their
extension of order (αw T )−1 . Due to the smallness of αw the characteristic momenta
of thermal fluctuations which form such sphaleron-like configurations are therefore
much smaller than the generic quantum fluctuations of the hot plasma which are of
order T . Hence the sphaleron-like fluctuations decouple from the quantum fluctua-
tions and one might expect that the thermal fluctuations responsible for the change
2
Of course there are no genuine sphalerons for T > Tc . We use the expression “sphaleron-
like configuration” for any configuration where the gauge field has the same qualitative features
as the sphaleron: The configuration is the one in a sequence of configurations where one of the
eigenvalues of the Dirac operator is zero and it is one where there is a clear lump of energy located
in a considerable region of space.

3
in Chern-Simons number are well described by classical physics for temperatures
above Tc .
The idea is therefore to start out in a classical configuration3

{Aai , φa ; Eia , π a } (2.1)

dictated by the Gibbs distribution exp(−H/T ) where H is the Hamiltonian. Since

we consider the classical theory, H is just the classical Hamiltonian for the gauge-
Higgs system
1 1
Z  
H = d x Eia Eia + Fija Fija + |π|2 + |Di φ|2 + M 2 |φ|2 + λ|φ|4
3
(2.2)
2 4
and in the temporal gauge it should be supplemented by Gauss constraint in the
form
Diab Eib = ig(φ† τ a π − π † τ a φ). (2.3)
After having chosen a configuration according to Gibbs distribution we let the system
evolve according to the classical equations of motion:
dAai δH
= = Eia (2.4)
dt δEia
dEia δH
= − a. (2.5)
dt δAi
If the system (i.e. in practise the lattice) is sufficient large the temperature will be
approximately constant in this micro-canonical simulation.
For details about the discretization of the above classical equations and choice
of (lattice) coupling constants we refer to [8]. Here it is sufficient to say that the
simulations were performed on a 163 lattice and that the bare values of the coupling
constants for the gauge fields and the Higgs fields were choosen such that the tree-
values of the theory corresponded to being in the broken phase with the Higgs mass
equal the W -mass and such that one sphaleron fits onto the lattice. As noticed in
[8] these constraints on the coupling constants force us for the given lattice size to
work at a temperature where the symmetry is actually restored.
In the symmetric phase we can now measure the change in Chern-Simons number
1 t Z Z
a a
Ncs (t) − Ncs (0) = dt d3 x Fµ,ν F̃µ,ν (2.6)
32π 0
2

The results of a typical measurement is shown in fig.1a. One identifies two “sphaleron-
like” transitions, of which the last one seems to correspond to a jump of Chern-
Simons number of two units. Superimposed on these we see a band of short wave-
length thermal fluctuations which carry the main part of the energy. According
3
Throughout this paper we assume that we can ignore the hypercharge sector, which means
that we effectively work with the Weinberg angle θw = 0.

4
to our arguments above the essential features of the sphaleron transitions should
remain unchanged if we strip off the short wave length thermal fluctuations. We
have done this for each of the time-sequence of configurations we get by solving
the classical equations of motion by iterating for a given configuration the simplest
relaxation equation:
∂φ δH ∂A δH
=− , =− (2.7)
∂t δφ ∂t δA
This technique is well known from the study of lattice instantons and monopoles
[9, 10, 11, 12]. The results for the configurations of fig.1a are shown in fig.1b.
Each configuration used in fig. 1a has been subjected to six cooling sweeps of the
kind given in (2.7). By this process the energy stored in the gauge fields drops by
almost a factor 80, in agreement with previously obtained results for instantons and
monopoles. The picture of sphaleron transitions is considerable sharpened and the
fact that the transition survives this cooling shows at least that the assumption of
an effective decoupling of the short and long wave length thermal fluctuations when
we discuss sphaleron-like transitions is internally consistent4 . The Higgs field relaxes
much slower and there seems to be a large degree of decoupling between the Higgs
field and the gauge field in the symmetric phase. In fact we get essentially the same
Chern-Simons picture as in fig. 1b if we ignore the Higgs field and only relax the
gauge field. Insensitivity with respect to the Higgs field has also been observed in
the study of monopole-like configurations [12].

3 Spectral flow
Our aim is to provide further evidence that the sphaleron-like transitions have the
essential characteristica of the continuum, relevant for the anomaly. One important
feature, contained in the Atiyah-Patodi-Singer index theorem [13] and explained in
detail for instance by Christ [14] is that the spectral flow of eigenvalues of the time
dependent Dirac operator HD (A(t) is directly related to the change in Chern-Simons
number if we consider a continuous time sequence of SU(2) gauge potentials A(t)
starting in one gauge vacuum at t = −∞ and ending up in a neighbour one at
t = +∞. In fact, if we consider fermions of a given chirality the change in Chern-
Simons number is equal the number of times eigenvalues of HD (t) cross zero from
below minus the number of times eigenvalues cross zero from above (the value zero
has no special status in this context, but is conveniently chosen since the vacuum
for Ai = 0 is identified with the filled Dirac sea).
4
It does of course not prove that that we can actually replace the short wavelength quantum
thermal fluctuations by short wavelength classical thermal fluctuations.

5
 One interesting point in the present situation is that we have at no time a true
interpolation from one gauge vacuum to a neighbour one, since we are working at a
finite temperature. We expect that the Atiyah-Patodi-Singer index theorem may be
used in this more general context, just moving from one gauge field configuration to a
neighbour one connected by a large gauge transformation, but have not attempted to
provide a rigorous proof of this conjecture. In practise we try, as already mentioned,
to use as smooth (i.e. cold) configurations as possible.
In the following we will be satisfied with verifying that the crossing of eigenvalues
at zero is closely linked to the change of Chern-Simons number as measured in the
most “naive” way, as described above. Since the concept of chirality on the lattice is
non-trivial [15, 12] we will not at the present stage try to unravel the exact counting
and assignment of chirality of the individual modes.
We now turn to the measurement of eigenvalues of the Dirac operator on the
lattice. We have found it most convenient to use the staggered fermion formalism.
The three-dimensional staggered fermion Dirac equation reads:
3
1 h i
iηi (n) Ui χ(n + î) − Ui† (n − î)χ(n − î) = Eχ(n)
X
(3.1)
i=1 2

where Ui (n) are the SU(2) gauge field variables living on the links i located at lat-
tice points n. The formal relation to a continuum gauge connection is Ui (n) =
exp(iAi (n)). The fermion field χ(n) is a one-component spinor and an SU(2)
doublet. η(n) is the Kawamoto-Smit phase (−1)n1 +···+ni−1 . We use antisymmet-
ric boundary conditions, in which case there are no zero modes in the free field case
and it is easier to identify an eigenvalue crossing zero.
In fig. 1c we have shown the time evolution of the lowest positive eigenvalue
for the gauge field configurations which were responsible for the change in Chern-
Simons number shown in fig. 1b. These are the configurations which are cooled and
thereby have lost most of the short range thermal fluctuations. However, even in
the case where the full thermal fluctuations are present we see essentially the same
picture. We have chosen to show in fig. 1c the eigenvalue from fig. 1b only because
it is easier to identify the Chern-Simons number on this figure.
It is seen that the first diving of an eigenvalue to zero coincides with the first
sphaleron transition of fig. 1b. We have not shown the higher eigenvalues of the
Dirac equation, but they do not get below 0.15. They show however a similar
(relative) diving as the lowest mode, and they should, since the choice of zero as the
point to count the crossing of eigenvalues is arbitrary, as mentioned above.
The situation is more complicated for the next change of Chern-Simon number
seen in fig. 1b since it consists of two successive jumps. Accordingly we see indeed

6
in fig. 1c the diving of two eigenvalues. A closer look at fig. 1b reveals that the
Chern-Simons number seems to stop for some time a the value 1/2, precisely the
value of the sphaleron, which we know has a zero mode [16, 17]. Notice that this
plateau at 1/2 is present both before and after cooling and from this point of view
should be taken seriously as a continuum configuration. Its presence is very clearly
reflected in the behaviour of the lowest eigenvalue. As long as the configuration
stays at a value of Ncs (t) ≈ 1/2 the eigenvalue is close to zero as we have shown in
fig. 2. An obvious interpretation of this behaviour could be that the system spends
some time in a sphaleron like configuration without being able to make up its mind
into which valley (gauge vacuum) to fall. If that is the case the word “sphaleron” is
indeed appropriately chosen for this configuration.

4 Energy lumps
In order to understand better the nature of the gauge field configurations responsible
for the change in the Chern-Simons numbers and the spectral flow of eigenvalues
in fig. 1 we have recorded the actual energy distribution of the gauge fields. It
was impossible to see any clear picture when all thermal fluctuations were included.
Only for the cooled configurations did a clear picture emerge. (Again this is con-
sistent with the experience from instantons and monopoles). The average energy
of the gauge fields after cooling is around 0.039 and we have chosen to show the
energy concentrations of the gauge field at two times: before the first sphaleron-like
transition and in the middle of the transition, where the eigenvalue of the Dirac
operator is close to zero (0.002). In both cases we show a sequence of 3D pictures
where regions of space occupied by cubes have an energy density above a certain
threshold (fig. 3). The three thresholds chosen are 0.06, 0.07 and 0.08. Fig. 3a-3c
illustrate a typical situation before the rapid change in Chern-Simons number while
fig. 3d-3f show the spatial distribution in the middle of the transition. We see a
marked difference in the concentration of gauge-field energy for the two situations.
In the case of the zero eigenvalue one can talk about a genuine extended object while
the other situation reflects typical fluctuations in the energy density which will al-
ways be present if we pick a random configuration. We see that the sphaleron-like
configurations in no way can be considered as nice symmetric configuration, but this
is not really to be expected. As a dual representation we show in fig. 4 the energy
density for the extended object present in the case where the eigenvalue is zero (fig.
3d-3f) in a plane where the concentration in the core of the energy lump is high.

7
5 Discussion
We have shown that the configurations responsible for the change in Chern-Simons
number on the lattice indeed seem to share the characteristica of true continuum
configurations with the same properties. They have considerable spatial extension
and eigenvalues of the Dirac equation will dive to zero a some point during the
change of the Chern-Simons number. For the sake of clarity we have in this paper
concentrated on a particular simulation and two sphaleron like transitions, but we
have performed a whole sequence of such simulations and in the process of analyzing
the results we have in fact seen an even more detailed relationship between the three
quantities: the Chern-Simons number, the energy lumps and the eigenvalues. One
type of behaviour is the following: when performing some cooling steps we still find
some fluctuations in Chern-Simons numbers which can not be classified as jumps
of order one and where two eigenvalues will dive towards zero, somewhat displaced.
A possible interpretation is that we have two sphaleron like configurations, the
appearance slightly displaced in time, with values of Ncs (t) of opposite sign. Another
type of behaviour which we have observed is one where an eigenvalue seemingly
crosses zero and comes back again or which moves close to zero and then return. At
the same time the Chern-Simons number will move close to approximately 1/2 and
return to zero again. It is again tempting to view this as a gauge field configuration
which moves to a sphaleron-like configuration and then return back to the same
vacuum-like configuration from which it originated. In this way it seems possible to
map out in detail the movement of gauge field configuration relevant for the change
of Chern-Simons number and to get a detailed knowledge of the mechanism by which
thermal fluctuations are able to co-operate and create sphaleron-like configurations.
Many things could be improved: One could obviously gain a lot if it was possible
to use significantly larger lattices5 . It would then be possible to go to smaller temper-
atures and smoother configurations, and maybe even to address the same questions
in the region where the symmetry is broken, but where the classical transitions from
one vacuum to another are not yet suppressed.
Another important improvement if one wants to dig into a more detailed inves-
tigation of the flow of eigenvalues is the chirality of the eigenmodes. A suitable
definition has been given in [15, 12].
A final very important question which remains to be addressed is the role of the
Higgs field. This role is still obscure to us. As already reported long ago [18] it
seems that the Higgs field in the symmetric phase partly decouples from the gauge
field. Since we effectively are in the symmetric phase the Higgs field will fluctuate
5
Unfortunately we do not have the computer facilities needed for such calculations.

8
and have zeroes, and when we iterate a few times the relaxation equations it still
seems somewhat decoupled from the gauge field. At smaller temperatures one would
expect a strong coupling of the phases of the Higgs field and the gauge field. Since
the real fermions in the electroweak theory couple to the Higgs field it is indeed
important to understand the role of this field. In the continuum the situation is
only clear in the broken phase. If a field configuration in the broken phase (gauge
fields and Higgs fields) interpolates in time between two non-equivalent vacua the
Higgs field has ( by purely topological reasons) to develop a zero in between the
vacuum configurations where the magnitude of the Higgs field is constant. This
zero of the Higgs field is precisely what is present in the sphaleron configuration and
it is this zero which allows the Dirac operator, coupled both to gauge- and Higgs
fields of the sphaleron, to have a normalizable energy eigenmode with eigenvalue
zero, representing precisely the crossing of zero of the energy levels[16, 17]. We do
not know of any detailed investigation of the role of the Higgs field in the symmetric
phase.

Acknowledgement It is a pleasure to thank Morten Laursen for discussions and

him and J. Vink for providing us with a program for the calculation of eigenvalues
of staggered fermion Dirac equation.

References
[1] G. ’t Hooft, Phys.Rev.Lett. 37(1976)8; Phys.Rev. D14(1976)3432
[2] V.A. Kuzmin, V.A. Rubakov and M.E. Shaposhnikov, Phys.Lett. 155B(1985)36
[3] N. Manton, Phys.Rev. D28 (1983) 2019; F.Klinkhamer and N.Manton,
Phys.Rev.D30(1984)2212
[4] P.Arnold and L.McLerran, Phys.Rev. D36(1987)581
[5] A.I. Bochkarev and M.E. Shaposhnikov, Mod.Phys.Lett. A2(1987)417
[6] S.Yu.Khlebnikov and M.E.Shaposhnikov, Nucl.Phys. B308(1988) 885
[7] L. McLerran, E. Mottola and M.E. Shaposhnikov, Phys.Rev. D43 (1991) 2027.
[8] J. Ambjørn, T. Askgaard, H. Porter and M.E. Shaposhnikov, Phys.Lett B244
(1990) 479; Nucl.Phys. B353 (1990) 346.
[9] I.M. Barbour and M. Teper, Phys.Lett B175 (1986) 445.
[10] E. Ilgenfritz, M. Laursen, M. Müller-Preusker, G. Schierholz and H. Schiller,
Nucl.Phys. B268 (1986) 693.

9
[11] M. Laursen and G. Schierholz, Z.Phys. C38 (1988) 501.
[12] S. Hands, Nucl.Phys. B329 (1990) 205.
[13] M.F. Atiyah, V.K. Patodi and I.M. Singer, Math.Proc.Cambridge Philos.Soc.
79 (1976) 71.
[14] N.H. Christ, Phys.Rev. D21 (1980) 1591.
[15] M. Golterman and J. Smit, Nucl.Phys. B245 (1984) 61.
[16] C. Callias, Commun.math.Phys. 62 (1978) 213.
[17] A. Ringwald, Phys.Lett B213 (1988) 61.
[18] J. Ambjørn, M. Laursen and M.E. Shaposhnikov, Phys.Lett B197 (1987) 49;
Nucl.Phys. B316 (1989) 483.

10
 Figure Caption

Fig.1 Measurements of the Chern-Simons as a function of time. Fig. 1a and shows

time evolution with thermal fluctuations included, while fig. 1b shows the
same time evolution with the thermal fluctuations partly stripped off. Fig. 1c
shows the lowest eigenvalue for the configurations which are responsible for
the Chern-Simons numbers shown in fig. 1b.
Fig.2 Fig. 2a shows in more detail the behaviour of the Chern-Simons number in
the neighbourhood of the second sphaleron transition recorded in fig. 1b. Fig.
2b shows details of the lowest eigenvalue corresponding to fig. 2a.
Fig.3 Fig. 3a-3f are 3D pictures of the regions in space where the energy concentra-
tion is larger than 0.06, 0.07 and 0.08 for two configurations before and in the
middle of the first sphaleron transition shown in fig. 1b. The first three figures
refer to a configuration before the transition and for which there are no small
eigenvalues, while the last three figures refer to a configuration in the middle
of the transition. This configuration has an eigenvalue of the Dirac operator
which is approximately zero.
Fig.4 An energy level plot in the x − y-plan of fig. 1c at z-height where the energy
concentration in the core is high.

11