1:
2:
3:
4: \section 4. Setting the scale
5:
6: The choice of a physical reference scale
7: is an important step in the analysis of the simulation data.
8: Results obtained on different lattices can then be
9: expressed in units of this scale and thus be compared with
10: one another.
11:
12: \input figure3
13:
14: Following ref.~[\ref{OaImp}], we adopt a mass-independent
15: scheme where the lattice spacing in physical units
16: is the same on all lattices at a given bare coupling.
17: Different choices of the reference scale are possible,
18: none of which appears to be free of some practical or conceptual
19: shortcoming.
20: Here we add a valence strange quark to the theory and
21: determine the scale through
22: the pion mass and the masses of the pseudo-scalar and vector mesons
23: that are made of a strange antiquark and a sea quark (we
24: refer to these as the $K$ and the $\Kstar$).
25: More precisely,
26: we adjust the quark masses so that the ratios $\MK/\MKstar$ and
27: $\Mpi/\MK$ assume some prescribed values
28: and then take $\MK$ as the reference scale.
29:
30: Since we wish to set the scale in a physically sensible way, we
31: require the ratio $\MK/\MKstar$ to be equal to its physical value
32: of $0.554$. This condition fixes the strange-quark mass $\ms$
33: at any given coupling and sea-quark mass $m$ (see fig.~3).
34: Ideally we would like the latter to be such
35: that $\Mpi/\MK$ assumes its physical value too, but this would require
36: a long extrapolation in the sea-quark mass
37: and, moreover, would be a point where the $\Kstar$ is unstable
38: (i.e.~the extrapolation would have to go through a kinematical threshold).
39:
40: We now note, however, that once $\ms$ is fixed,
41: the reference scale in lattice units, $a\MK$,
42: appears to be weakly dependent on $m$, particularly so at small $m$
43: (see fig.~3).
44: The reason for this behaviour
45: (which is seen on all series of lattices) could be that
46: both $\MK$ and $\MKstar$ are functions of
47: $m+\ms$ rather than of $m$ and $\ms$ separately, up to
48: corrections proportional to the squares of the masses.
49: In any case, the observation suggests the reference scale
50: to be defined at
51: the point where, say, $\Mpi/\MK=0.85$, which is
52: within the available data range.
53: This convention, although somewhat unphysical,
54: is entirely satisfactory for the purpose
55: of comparing results from different lattices.
56:
57: The results for the reference scale obtained in this way are
58: summarized in table~2. In order to avoid any confusions, we
59: mark all quantities evaluated at the reference point with
60: a subscript ``ref''. The sea-quark and strange-quark hopping parameters
61: at the reference point, for example, are denoted by $\kappa_{\rm ref}$
62: and $\kappa_{s,{\rm ref}}$.
63: Setting $\MKref=495$ MeV,
64: this leads to the lattice spacings quoted in the last column
65: of the table, while for the pion masses at the smallest sea-quark
66: masses on the
67: $A$, $B$ and $D$ series of lattices we obtain $403$, $381$ and
68: $377$ MeV respectively.
69:
70: \input table2
71:
72: The lattice spacings calculated here are
73: significantly smaller than those previously published by us
74: in a conference report [\ref{Dublin}], where
75: the Sommer radius [\ref{SommerRadius}] was used as reference scale.
76: Larger lattice spacings are also obtained
77: if the scale is set by the $K$ and $\Kstar$ masses, similarly to
78: what was done here, but at larger sea-quark masses (see fig.~3).
79: As a result of the new
80: determination of the lattice spacings, our
81: estimates of the pion masses in MeV are pushed to
82: higher values than those quoted in ref.~[\ref{Dublin}].
83: Moreover, we decided to discard
84: the simulation at the lightest quark mass reported there,
85: because the lattice turned out to be too small for that mass.
86:
87: All this illustrates the fact that at present
88: the assignment of physical units remains somewhat ambiguous.
89: For a definitive solution of the problem,
90: simulations at smaller quark masses will probably be required,
91: and the scale setting may eventually have to be
92: based on the properties of the stable hadrons
93: (the pion and the nucleon in the two-flavour theory).
94: