Topology, symmetry, and phase transitions in lattice gauge ... - tuprints
Topology, symmetry, and phase transitions in lattice gauge ... - tuprints
Topology, symmetry, and phase transitions in lattice gauge ... - tuprints
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
3.4 exploit<strong>in</strong>g universality 33<br />
In this case, K c,∞ was fixed at the known value of 2/ ln(1 + √ 2). The correction to<br />
scal<strong>in</strong>g exponent tends towards 2 from above, ω = 2 + δ with δ → 0 + as we <strong>in</strong>crease<br />
the lower bound N m<strong>in</strong> used <strong>in</strong> the fit (δ starts at around 2 · 10 −4 for the full range of<br />
N shown <strong>in</strong> Fig. 3.9, <strong>and</strong> it falls below 10 −5 at N m<strong>in</strong> around 400). S<strong>in</strong>ce the exponent<br />
of the lead<strong>in</strong>g irrelevant operator that breaks rotational <strong>in</strong>variance is predicted to<br />
be exactly 2 [63], our result is consistent with the conjecture of Ref. [64] that the<br />
only irrelevant operators that appear <strong>in</strong> the 2d nearest neighbor Is<strong>in</strong>g model are<br />
those due to the <strong>lattice</strong> break<strong>in</strong>g of rotational <strong>symmetry</strong>. This may be tested on<br />
a triangular 2d <strong>lattice</strong> where the lead<strong>in</strong>g irrelevant operator to break rotational<br />
<strong>in</strong>variance leads to ω = 4 <strong>in</strong>stead, while the lead<strong>in</strong>g rotationally <strong>in</strong>variant operator<br />
would give an isotropic correction to scal<strong>in</strong>g with ω = 2 <strong>in</strong> either case [64].<br />
On the other h<strong>and</strong>, our correction exponent for SU(2) is clearly at odds with<br />
ω = 2. In this case, it is possible that there exists an irrelevant operator that is not<br />
present <strong>in</strong> the 2d Is<strong>in</strong>g model or the correspond<strong>in</strong>g conformal theory. It is more<br />
likely, however, that our exponent is really an effective exponent. When there are<br />
several nearby compet<strong>in</strong>g exponents it is extremely difficult to extract the smallest<br />
one from simulations. 6<br />
3.4.4 From coupl<strong>in</strong>g to temperature<br />
In 2 + 1 dimensions the coupl<strong>in</strong>g g3 2 has the dimension of mass <strong>and</strong> sets the scale.<br />
The bare <strong>lattice</strong> coupl<strong>in</strong>g is then given by [61],<br />
β = 2N c<br />
ag3B<br />
2 , g3B 2 = g3 2 + c 1 ag3 4 + c 2 a 2 g3 6 + . . . (3.43)<br />
Substitut<strong>in</strong>g the expansion <strong>in</strong>to β <strong>and</strong> us<strong>in</strong>g T = 1/(N t a) gives the <strong>lattice</strong> coupl<strong>in</strong>g<br />
<strong>in</strong> terms of temperature, which at criticality reads,<br />
β c (N t )<br />
2N c<br />
≃ T c<br />
g3<br />
2 1<br />
g3<br />
2 N t − c 1 − c 2 + . . . (3.44)<br />
T c N t<br />
The theory is super-renormalizable with a rapid approach to the cont<strong>in</strong>uum limit,<br />
which is reflected here <strong>in</strong> the lead<strong>in</strong>g order l<strong>in</strong>earity of the critical coupl<strong>in</strong>g <strong>in</strong> N t .<br />
Our high precision estimates for β c (N t ) allow us to determ<strong>in</strong>e the sublead<strong>in</strong>g<br />
corrections to l<strong>in</strong>earity.<br />
It is not clear to the naked eye, but the results <strong>in</strong> Fig. 3.10 are not well fitted by a<br />
straight l<strong>in</strong>e. Indeed, fitt<strong>in</strong>g all data po<strong>in</strong>ts with an acceptable reduced χ 2 requires<br />
three sublead<strong>in</strong>g coefficients that are not all well constra<strong>in</strong>ed. Restrict<strong>in</strong>g to N t ≥ 3<br />
yields a more sensible fit<br />
β c (N t ) = 1.4927(26)N t + 0.868(34) − 1.58(14) 1 N t<br />
+ 1.48(19) 1 N 2 t<br />
+ · · · , (3.45)<br />
6 It was noted <strong>in</strong> [60] that the observed correction to scal<strong>in</strong>g exponent of 2 +1-dimensional SU(2),<br />
ω = 1.64 <strong>in</strong> their case or ω = 1.48(4) <strong>in</strong> ours, agreed well with some predictions for the universality<br />
class of the 2d Is<strong>in</strong>g model that <strong>in</strong>cluded 1.6 [65]. At the time it was discussed whether such non<strong>in</strong>tegral<br />
correction exponents could arise <strong>in</strong> other ferromagnetic models <strong>in</strong> this class, <strong>and</strong> whether<br />
the correspond<strong>in</strong>g correction amplitudes happened to vanish identically <strong>in</strong> the Is<strong>in</strong>g model, see [66].<br />
This appears to be ruled out by a conformal field theory analysis: there is no irrelevant operator with<br />
ω < 2 <strong>in</strong> any unitary model of the 2d Is<strong>in</strong>g class [67, 64].