PROPERTIES OF SU(2) CENTER VORTEX STRUCTURE IN SMOOTH CONFIGURATIONS

: New analysis regarding the structure of center vortices and their color structure is presented: Using data from gluonic SU(2) lattice simulation with Wilson action a correlation of ﬂuctuations in color space to curvature of vortex ﬂuxes was found. Finite size effects of the S2-homogeneity hint at color homogeneous regions on the vortex surface.


Introduction
The center vortex model of Quantum Chromodynamics [1,2] explains confinement [3] and chiral symmetry breaking [4][5][6] by the assumption that the relevant excitations of the QCD vacuum are Center vortices, closed color magnetic flux lines evolving in time. In four dimensional space-time these closed flux lines form closed surfaces in dual space, see Figure 1. In the low temperature phase they percolate space-time in all dimensions. In Maximal Center gauge the center vortex surface can be detected by the identification of Wilson loops evaluating to non-trivial center elements and by projecting the links to the center degrees of freedom: The piercing of a Wilson loop by a center flux line contributes a non-trivial factor to the numerical value of the loop. The detection of plaquettes pierced by a flux line is schematically shown in Figure 2. The vortex flux has a finite thickness, but is located by thin P-vortices. As long as these P-vortices successfully locate vortices, we speak of a valid vortex finding property. A problem concerning this property was discovered by Bornyakov et al. [7]. We were able to resolve these problems using an improved version of the gauge fixing routines [8][9][10] and could show for relatively small lattices that center vortices reproduce the string tension. We found hints towards a color structure existing on the vortex surface [11]. This color structure might be directly Figure 2. Vortex detection as a best fit procedure of P-Vortices to thick vortices shown in a two dimensional slice through a four dimensional lattice. related to the topological charge of center vortices and could lead to a more detailed explanation of their chiral properties. Their study is especially complicated by the fact that measurements of the topological charge require sufficiently smooth field configurations, whereas the quality of the center vortex detection suffers with increasing smoothness of the lattice, that is, the vortex finding property can get lost during smoothing or cooling the lattice. We want to bring more light into this difficulties by an analysis of the • correlation between color structure and surface curvature, • size of color-homogeneous regions on the surface, • influence of cooling on color-homogeneous regions on the surface. • influence of smoothing the vortex surface on color-homogeneous regions.
As this is the first time these measurements are performed with our algorithms, this work is to be understood as a proof of concept.

Materials and Methods
Our lattice simulation is based on gluonic SU(2) Wilson action with inverse coupling β covering an interval from β = 2.1 to β = 3.6 in steps of 0.05. The corresponding lattice spacing a is determined by a cubic interpolation of literature values given in Table 1 and complemented by an extrapolation according to the asymptotic renormalization group equation We assume a physical string tension of (440 MeV) 2 . Our analysis is performed on lattices of size 8 4 , 10 4 and 12 4 and 16 4 . We use an improved version of maximal center gauge to identify the position of the flux lines building up the vortex surface via center projection. The improvements are described in detail in Refs. [8][9][10] and identify non-trivial center regions by perimeters evaluating to non-trivial center elements. These regions are used as guidance for maximal center gauge. By simulated annealing we look for gauge matrices Ω(x) ∈ SU(2) at each lattice point x, so that the functional is maximized. After fixing the gauge we project to center degrees of freedom and identify the non-trivial plaquettes but save the un-projected lattice for our measurements concerning the color structure.
These measurements are based on the S2-homogeneity which compares the color vector of two plaquettes related to the same lattice point. With the two plaquettes written as W j = σ 0 cos α j + i ∑ 3 k=1 (n j ) k σ k sin α j the S2-homogeneity is defined as h S2 := 0.5 | n 1 + n 2 | ∈ [0, 1] with Pauli-matrices σ k and n j ∈ S 2 . (3) It is a gauge independent quantity proportional to cos α 2 , where α is the angle between the two color vectors n 1 and n 2 .
By calculating the S2-homogeneity of two plaquettes that are pierced by a center flux-line, we restrict the measurement to the vortex surface and can distinguish the two scenarios depicted in Figure  6: • Parallel plaquettes corresponding to a weakly curved flux-line, that is, plaquettes located in the same plane in dual space. • Orthogonal plaquettes, plaquettes that are neighbours in dual space but have different orientation, corresponding to strongly curved flux-lines, Figure 3. Depicted are two plaquettes that are related to the same lattice-point via parallel transport along the single edge depicted as thicker black line. We distinguish two scenarios: left) two parallel pierced plaquettes correspond to a weakly curved flux-line. right) two orthogonal pierced plaquettes correspond to a strongly curved flux-line.
As the curvature of the flux-line directly correlates to the curvature of the vortex surface, this allows an analysis of the correlation between S2-homogeneity and curvature of the vortex surface.
Calculating the S2-homogeneity for different lattice spacings and lattice sizes allows to examine the size of homogeneous regions on the vortex surface by exploiting finite-size effects: we assume that for sufficiently big lattices the S2-homogeneity is independent of the lattice size and at most linear dependent on the lattice spacing. As long as the color structure fits into the lattice we assume a decoupling of this structure from the lattice parameters. Shrinking the lattice by reducing the lattice spacing allows to identify the physical lattice extent below which finite size effects introduce stronger dependencies on the lattice parameters, see Figure 4. Varying the physical lattice size by changing the number of lattice sites and repeating the procedure we can check whether these finite size effects define a physical scale independent of the lattice parameters: at the onset of the finite size effects the lattice spacing times the lattice size should be a constant defining the size of color-homogeneous regions on the vortex surface. With decreasing lattice sizes the onset of the finite size effects occurs at lower values of β, hence we choose small lattices to perform these measurements. As this procedure requires sufficiently good statistics at high values of β, we include in the averages all pairs of plaquettes belonging to the same P-vortex.
For measurements of topological properties sufficiently smooth configurations are required, but smooth configurations complicate the vortex detection. There are many different procedures for generating such smooth configurations or smoothing excitations belonging to specific degrees of freedom. We use Pisa Cooling [17] with a cooling parameter of 0.05. Repeating the aforementioned measurements in cooled configurations we compare the influence of a direct smoothing of the vortex surface, described in detail in [18,19], with the effects of cooling. The smoothing procedure consist of "cutting out" parts of the vortex surface and "sewing together" the surface again, see Figure 5. Additional to the three depicted procedures we have "smoothing 0" which deletes isolated unit-cubes and is part of all other listed smoothing routines. All smoothing procedures remove short range  fluctuations from the vortex surface and are considered to have no influence on the infra-red degrees of freedom, but smoothing 2 has influence on the connectivity of the vortex surface in contrast to the other smoothing methods.

Results
We start by presenting our findings concerning the correlation between color structure and surface curvature of vortices. The measurements were performed on lattices of size 16 4 and the data show a clear distinction between weakly and strongly curved parts of the vortex surface with respect to S2-homogeneity: the more the vortex surface is curved, the smaller is the value of the color-homogeneity. This can be seen in Figure 6 where we show distinct evaluations of the S2-homogeneity h S2 for parallel plaquette pairs (larger values) and orthogonal plaquette pairs (smaller values) for various smoothing and cooling steps. The difference between parallel and perpendicular plaquettes indicates a clear correlation of fluctuations in color space to fluctuations in space-time. h S2 increases with more intense cooling, especially for parallel plaquettes. Also the mentioned difference and the correlation increases. It is interesting to observe that with cooling this difference between parallel and perpendicular plaquette pairs increases more for pairs off the vortex (red lines) than for pairs on P-vortices (data points). For 0 cooling steps the difference between the red lines is roughly 50% of the difference between data points. This difference increases to ≈ 100% for 5 cooling steps.
Since the thickness of vortices increases with cooling and P-vortices stay thin "off the P-vortex" can still be "on the thick vortex" and those pairs "on the thick vortex" contribute to the "off the vortex" averages. This shows that the correlation between curvature and S2-homogeneity is a property of the thick vortices and not of the surrounding vacuum.
Comparing the different smoothing procedures it can be seen that smoothing 1 and smoothing 3 lead to quantitatively identical results with a homogeneity raised above those of smoothing 0. With smoothing 2 the difference between weak and strong curvature is reduced: The homogeneity of weak curvature is decreased and the homogeneity of strong curvature increased. As smoothing 2 is the only procedure influencing the connectivity of the vortex surface we conclude that this connectivity might be of importance in further considerations of the color structure and that smoothing 2 should be used with care. This is strengthened by the fact, that smoothing 2 results in the vortex becoming overall more homogeneous than the vacuum, see Figure 7. A non-trivial color structure is indicated by color inhomogeneities, hence we suspect that smoothing 2 cuts out parts of the vortex surface that might carry potential color structure. The dependency of the S2-homogeneity on the physical lattice volume allows a more detailed investigation of the color structure on the vortex surface. In Fig. 8 we compare the S2-homogeneity along the vortex surface for different lattice sizes and smoothing procedures. We find a loss of the S2-homogeneity for small lattice volumes and give arguments that the loss of the homogeneity is caused by finite size effects indicating a physical non-vanishing size of color-homogeneous regions. We estimate the size of these homogeneous regions by linear fits to the values of the S2-homogeneity in the large and small volume region as shown in Figure 4. This procedure is applied to different lattice sizes and numbers of cooling steps. Multiplying the lattice spacing at the intersection with the lattice Figure 8. Averaged over 100 configurations, the S2-homogeneity along the vortex surface is compared for different lattice sizes and smoothing procedures. With increasing lattice size the curves shift to smaller lattice spacing indicating finite size effects. extent, we estimate the size of the homogeneous regions. In Figure 9 it can be seen that the values are compatible within errors for smoothing 0, smoothing 1 and smoothing 3 on lattices of size 8 4 and 10 4 . The measurements in lattices of size 12 4 result in higher values with bigger errors. the lattice on homogeneous regions. In Figure 10 the S2-homogeneities are shown after 1 cooling step. Of interest is that with cooling the S2-homogeneities of the different smoothing procedures Figure 10. Averaged over 100 configurations, the S2-homogeneity along the vortex surface is compared for different lattice sizes and smoothing procedures after 1 cooling step. Observe that cooling increases the overall homogeneity on the vortex surface.
become more and more similar to those of smoothing 2. After 2 cooling steps smoothing 1 results in Figure 11. After 1 cooling step the extension of homogeneous regions is not significantly bigger than without cooling. For further analysis the data of smoothing 2 will be ignored.
the S2-homogeneity of the vortex becoming no longer distinguishable from the vacuum, see Figure  12. After 3 cooling steps smoothing 1, smoothing 2 and smoothing 3 result in the S2-homogeneity of the vortex growing above those of the vacuum, while the homogeneity of smoothing 0 is no longer distinguishable from those of the vacuum, see Figure 14. for analysing color structures on the vortex surface. The cause for this loss of the vortex finding property and a possible resolution is of special interest as it hints at possible ways to improve the vortex detection algorithms. This topic is discussed in another article [20]. The S2-homogeneities of the usable part of the data are shown in Figure 18. The line fits used to identify the onset of the finite size effects became problematic, causing a possible underestimation of the error, hence the data resulting from 10 cooling steps should be taken with care.    As the error bars result from intersections of line fits they are not necessarily symmetric. To combine the data, the error bar was interpreted as two separate datapoints. Using the data from the lattize sizes 8 4 and 10 4 with smoothing 0, smoothing 1 and smoothing 3 we estimate the size of the homogeneous regions to the values shown in Figure 20. The estimation starts at 0.85 ± 0.07 fm without cooling, Figure 20. The dependence of the size of homogeneous regions on the number of cooling steps is estimated by combining the data from the lattice sizes 8 4 , 10 4 , respective smoothing 0, smoothing 1 and smoothing 3. reaching 1.1 ± 0.2 fm after 10 cooling steps. This increase seems to have a maximal value, that is, cooling does not let the size of homogeneous regions increase to infinity. The dependency of the S2-homogeneities on the lattice spacing below the onset of the finite size effects showed hints at a stepwise reduction. To distinguish whether these results from random fluctuations of our data or is caused by a quantization of the sizes of the homogeneous regions might be possible with more lattice data.

Discussion
Our results show that curvature of flux lines correlates to fluctuations in color space. This correlation becomes stronger when cooling procedures are applied, but weakens when the connectivity of the vortex surface is harmed. The correlation seems to be a special feature along the center vortex surface as it weakens when restricting the analysis to plaquettes that are not pierced. As intersections and twistings of vortices can produce topological charge and the twisting of the vortex is correlated to color fluctuations, see [21], our data strengthens the assumption that the color structure of vortices directly relates to topological charge, enabling center vortices to explain chiral symmetry breaking.
Complications concerning measurements of the topological charge and center vortices raise the question, whether the topological properties of center vortices get lost during cooling or if the properties are kept but the vortex detection fails. Our data indicate the latter because a possible loss of the color structure due to expanding color homogeneous regions on the vortex surface is not probable as the size of these regions seems to be bound by an upper limit. Cooling increases the S2-homogeneity of the Vortex stronger than it increases the homogeneity of the vacuum.
The finite size of these color homogeneous regions is of further interest concerning the question whether the vortex surface covers the S 2 defining the color space or not. It indicates that finite areas on this color S 2 can not correspond to arbitrarily big areas on the vortex surface.
To stay sufficiently far away from finite size effects and guaranty a sufficiently high resolution concerning homogeneous regions we advice to keep the physical lattice volume well above and the lattice spacing well below 0.78 fm. As our analysis at a higher number of cooling steps suffered from a loss of the vortex finding property, further improvements at detecting center vortices in lattice simulations are needed. Further analysis of this loss of the vortex finding property is to be published in [20].