Filename
stringlengths 22
64
| Paragraph
stringlengths 8
5.57k
|
|---|---|
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 6: Distribution of φ offsets for underlying hydrological processes, and power spectral density of the governing physical process. One can see that the predominant number of offsets are distributed ∼ 0, and the governing physical process behaves similarly to most underlying power spectra. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Section 5 demonstrated a framework to determine a governing streamflow process. In doing so, we highlighted several key points. Critically, the under- lying hydrological time series closely resemble the governing physical process’ behaviours. Most notably the temporal dynamics share a periodic spike in streamflow, which corresponds to high power at low frequencies, and decreasing power at higher frequency components in the frequency domain. However, it is possible that temporal or spectral dynamics may change over time if the underlying process is nonstationary. Furthermore, the underlying time series and the governing process may display differences in their behaviour prior to and after our alignment algorithm. In this section, we test the existence of nonstationarity among our collection of underlying streamflow processes, and in the governing physical process before and after the procedure introduced in Section 5. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 7: Evolution of normalized first eigenvalue for raw and aligned collections of streamflow. Both the initial and aligned collection display broad similarity in the behaviour of the normalized first eigenvalue. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Next, we explore the evolutionary behaviour of both collections’ first eigen- vector. The evolutionary eigenvectors are shown in figure 7, with figure 8a and figure 8b showing the initial collection and aligned collection’s first eigenvector, respectively. The first notable observation is the general uniformity in magnitude along all coefficients. That is, each hydrological station comprises a similar level of attribution to the first eigenvector’s total variance. Furthermore, this trend is consistent over time in both the initial and aligned collections. The second observation is a moderate increase in the uniformity of magitudes in the aligned collection, shown in figure 8b. This is consistent with the core aim of our procedure, which translates the underlying time series such that the L2 norm with the current governing process, G, is minimized. To further explore this phenomenon, we compute the variance of time-varying eigenvector coefficients among both collections. The variance among the collection of initial time series is 1.03 x 10−5, while the aligned collection is 9.43 x 10−6. The modest reduction in variance is consistent with the increased uniformity observed in the figure. The reduction in variance of ∼ 7.5% may indicate that the initial mean of the underlying time series, prior to the optimisation procedure, serves as a good estimate of an underlying physical process. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | In keeping with our assessment of time-varying dynamics, we now turn to analyzing the power spectrum of the governing hydrological process. We wish to answer two main questions. First, do the spectral dynamics of the overall process resemble the dynamics of the underlying time series? Second, do we see differences over time in the power spectrum? Figure 9 shows the time-varying power spectrum of our governing process G before and after our alignment procedure. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 8: Evolution of first eignenvector magnitude of coefficients. The figure shows the magnitude of the coefficient of each element along the basis for each eigenvector propagated forward in time. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | periodic component. Like the underlying time series, lower power is exhibited at higher frequency components. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Visual inspection of Figure 9 suggests that the power spectrum does not display significant variability over time, for either collection (figure 9a or figure 9b). At the beginning of our time window in 1980, the spectrum displays significant power at low frequencies, with decreasing power at higher frequencies. Although the precise amplitude at various frequencies does oscillate over time, the general pattern is relatively stable. The spectral similarity between the initial and aligned collections once again suggests that simply computing an average of the initial time series could serve as a good approximation to the underlying physical process. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | In Section 3 we studied the similarity between hydrological stations’ temporal dynamics. Our trajectory analysis and associated hierarchical clustering revealed the existence of three predominant clusters. The three clusters shown in Figure 2 display a high level of intra-cluster similarity, with surprisingly low inter-cluster similarity. The temporal affinity norm of 0.26 highlights the low degree of affinity exhibited between trajectories belonging to separate clusters. Subsequent experiments reveal that these varied cluster behaviours were not related to stations’ spatial location, ruling out the assertion that the temporal dynamics of Australian streamflow may be related to geographic location. Despite the clear cluster separation, visual inspection suggests that most streamflow time series possess broadly similar features - small oscillations around a level close to zero, with sharp spikes which appear to occur on an annual periodicity. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Section 4 builds upon this analysis, introducing a new framework to explore similarity in periodic behaviours. Our experiments reveal uniform behaviours in the spectral dynamics of underlying stations’ streamflow. Figure 4 shows the existence of a majority cluster of spectral trajectories, and the spectral affinity norm of 0.79 confirms the profound similarity in streamflow spectral dynamics. This is primarily due virtually all locations displaying most significant power at at the frequency ν = 0.0027, corresponding to an annual streamflow cycle. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Given the general similarity in the temporal domain and the profound similarity displayed in the frequency domain, in Section 5 we introduce an iterative algorithmic procedure to determine a governing hydrological process. We introduce an L2 based objective function, where each time series is translated to minimize an average loss. Figures 5 and 6 reveal several notable findings. First, the mean of all streamflow time series does a reasonably good job in estimating the general behaviour of the collection, and is similar to our estimated process after the application of our alignment procedure. Figure 6a shows that the majority of estimated offsets are in fact 0, confirming the sample mean’s utility as a rough estimate. Figure 6b demonstrates that the power spectrum of the governing process resembles that of the underlying time series, and therefore still resembles the overarching theme of periodic streamflow behaviour. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 9: Rolling log power spectral density of governing temporal process. The constant nature of the log PSD suggest that both spectra are stationary. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | We round out our analysis in Section 6 where we examine the stationarity of such behaviours previously identified in the prior sections. In our time-varying PCA, we study the explanatory variance exhibited by the correlation matrix’s first eigenvalue, which is shown in Figure 7. The figure demonstrates a similar level of explanatory variance exhibited by the first eigenvalue in both the initial and aligned collections of streamflow time series. To further investigate, we accompany this analysis with a time-varying study of the magnitude of the first eigenvectors’ coefficients displayed by both collections. Both collections display relative uniformity in the coefficient magnitudes over time, indicating that there is no element, or collection of elements, along the vector accounting for the majority of the variance. Figure 9 shows the evolutionary power spectral density for both collections. It is clear that in both cases, the spectrum displays pronounced stationarity over time, exhibiting the same key features of significant power at low frequencies and decreasing power along higher frequency components. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | There are several avenues for future work. First, one could explicitly compare the offset methodologies introduced in this paper to other techniques used in various areas of the physical sciences. Furthermore, one could study the sensitivity of the results as various objective functions are used instead of those introduced in our experiments. The spectral analysis techniques introduced in this manuscript could be implemented differently, specifically, one could do so in a Bayesian framework so that uncertainty is quantified when evaluating the power spectrum. Although many of the techniques used in this paper are relatively standard (spectral analysis, L1 normalized trajectory modelling, hierarchical clustering, PCA, etc.) there are alternative methods that could be used to address the same data-driven questions. Future work could explore the stability of our results when alternative techniques are used. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | In section 3 and section 4 of the paper, four hydrological stations’ time series are displayed (Paddy’s River, Goodradigbee, Bluewater Creek and Gascoyne). The first two correspond to hydrology stations on the East Coast of Australia, while the third and fourth correspond to stations in the Northern and Western parts of Australia respectively. Our analysis throughout the paper demonstrates that these time series all display high similarity in temporal and spectral dynamics. Given the geographic spread of such stations, this inspired us to explore the spatial distribution of hydrology stations across Australia. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | We generate a distance matrix between all n = 414 hydrology stations, where the distance between any two candidate stations is the geodesic distance between two tuples of latitude and longitude data. We then apply K-means clustering to determine hydrology stations’ cluster memberships, where we use the elbow method to determine the existence of K = 3 clusters. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | The algorithm’s objective function seeks to minimize an appropriate sum of square distances. With k chosen a priori based on the elbow method aforemen- tioned, we examine all possible disjoint unions C1 ∪ C2 ∪ · · · ∪ Ck of {x1, . . . , xn}. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 10: Location of hydrological stations annotated on map of Australia. Colours denote cluster labels, which are determined solely based on geodesic distance between hydrological stations. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | For a space of dimension 2 or greater, it is NP-hard to find the globally optimal solution of this problem. The K-means algorithm [80] is an iterative solution that enables fast convergence to a locally optimal solution. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 10 shows the two respective sub-figures relating to our cluster analysis. Figure 10a corresponds to the geographic location of hydrology stations (with their colour denoting cluster membership) and the elbow curve (shown in Figure 10b displays the elbow curve plot. One can clearly identify three separate geographic clusters in Figure 10a. However the preceding analysis in this paper demonstrated that despite clear geographic separation, temporal and spectral dynamics of such stations were largely indistinguishable. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | To further explore the nonstationarity in the initial and aligned governing processes, we apply a reversible jump Markov Chain Monte Carlo algorithm (AdaptSPEC [81]) to identify the number and location of change points in each time series. The algorithm determines changepoints based on changes in the time-varying power spectrum for any candidate time series. That is, a nonstationary time series is assumed to exhibit local stationarity, where each (stationary) partition possesses a unique power spectral density. The algorithm determined the existence of 9 changepoints in both the initial and aligned time series. However, the location of change points within each time series were quite different. Changepoint indices and the respective location (date) for both time series are shown in Table 3. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | Figure 11: Time-varying power spectral density estimation for initial and aligned collections of hydrological time series. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | with high similarity in key trends. That is, significant power at low frequencies (indicative of a strong annual periodicity), and decreasing power at higher frequency components. Although 10 (quite different) segments are identified within each time series, the consistency in the shape and amplitude of the power spectrum between all adjacent segments supports the general notion of stationarity for both time series. |
Processed_Temporal_and_spectral_governing_dynamics_of_Austra.txt | In search of peak human athletic potential: a mathematical investigation, Chaos: An Interdisciplinary Journal of Nonlinear Science 32 (2022). doi:10.1063/5.0073141. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | I show how one can use lattice methods to calculate various continuum properties of SU(N ) gauge theories; in part to explore old ideas that N = 3 might be close to N = ∞. I describe calculations of the low-lying ‘glueball’ mass spectrum, of the string tensions of k-strings and of topological fluctuations for 2 ≤ N ≤ 5. We find that mass ratios appear to show a rapid approach to the large–N limit, and, indeed, can be described all the way down to SU(2) using just a leading O(1/N 2) correction. We confirm that the smooth large–N limit we find is confining and is obtained by keeping a constant ’t Hooft coupling. We find that the ratio of the k = 2 string tension to the k = 1 fundamental string tension is much less than the naive (unbound) value of 2 and is considerably greater than the naive bag model prediction; in fact it is consistent, within quite small errors, with either the M(-theory)QCD-inspired conjecture that σk ∝ sin(πk/N ) or with ‘Casimir scaling’. Finally I describe calculations of the topological charge of the gauge fields. We observe that, as expected, the density of small-size instantons vanishes rapidly as N increases, while the topological susceptibility appears to have a non-zero N = ∞ limit. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | How SU(N ) gauge theories approach their N = limit and what that limit is, are interest- ∞ ing questions [1], in themselves, whose answers would, in addition, represent a significant step to- wards addressing the same question in the context of QCD. Accurate lattice calculations in 2+1 di- mensions [2] show that in that case the approach is remarkably precocious in that even N = 2 is close to N = . Such calculations have to be very accurate because for each value of N one has to perform a continuum extrapolation of var- ious mass ratios and then these are compared and extrapolated to N = . Earlier D=3+1 calcula- tions [3,4] were much too rough for this purpose even if their message was optimistic. Recently, however, this situation has been rapidly improv- ing and here I want to describe some of the things that one has already learned. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | We would like a non-perturbative confirmation of this expectation. Again, we would like a confirmation of the assumption that the large- N theory remains confining, since much of the phenomenological argument that QCD might be relies on this being so [1]. Since close to N = quarks are in the fundamental representation, the leading correction to N = is expected to be O(1/N ). In the pure gauge theory it is expected to be O(1/N 2). (Again an expectation that one would wish to test non-perturbatively.) Thus if one would expect the QCD is close to N = SU(3) gauge theory to be close to SU( ) as well. We can test this by calculating dimensionless ra- tios of physical masses as a function of N . There are also arguments and speculations (e.g. [5]) that topological fluctuations are very different at large N and at small N ; this too we would like to test in lattice simulations. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | tinuum) properties of SU(2), SU(3), SU(4) and SU(5) theories and directly compare them. Since N = 5 is large in the sense that the leading cor- rection 1, it is not surprising ≪ that such a calculation turns out to suffice for our purposes. These calculations were intended as exploratory. A much larger calculation is now underway and this will provide much more infor- mation on the mass spectrum and its dependence on the number of colours. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Thus by varying the inverse lattice coupling β we can vary the lattice spacing a. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Path Integrals using standard Monte Carlo tech- niques. On the lattice t = na and so we obtain the energies from eqn(4) as aEn i.e. in units of the lattice spacing. In practice one needs to use several carefully chosen operators, and a varia- tional calculation [6,7]. Since our volume is finite we must make sure that finite volume corrections are negligible. For the calculations I describe here such checks have been made. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Having calculated some masses ami at a fixed value of a (i.e. at a fixed value of β) we can re- move lattice units by taking ratios: ami/amj = mi/mj. This ratio differs from the desired contin- uum value by lattice corrections. For our action the functional form of the leading correction is known to be O(a2). Thus for small enough a we can extrapolate our calculated mass values mi(a) mj(a) where using different mk will make differences at O(a4) (as does the a-dependence of mk(a)). We shall later show some explicit examples of such extrapolations. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Figure 1. The mass of the lightest scalar glueball, m0++, expressed in units of the string tension, σ, is plotted against the latter in lattice units. The a 0 continuum extrapolation, using a leading lattice correction, is shown. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Fig.1 I show you how this works for the lightest scalar glueball for N = 2, 3, 4, 5. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | While the mass of the scalar glueball is the most accurately calculated, because it is the light- est, it is also the one with (by far) the largest lat- tice corrections. Despite this it is already clear from Fig.1 that for small a there is very little N - dependence for N 3. To be quantitative we perform continuum extrapolations using eqn(5), as shown in Fig.1. The results are shown in Fig.2 where we also show the extrapolations to N = using eqn(6). |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | The extra factor involving the plaquette is a mean-field (or tadpole) improved version of β and the naive λ(a) we would derive from it [11]. It is necessary [11] because the naive lattice coupling is known to be very poor in the sense of having very large higher order corrections. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | values should fall on a universal curve. In Fig.3 we plot all our calculated values of a√σ, against the value of λI (a). We observe that, within cor- rections which are small except at the largest value of a, we do indeed see a universal curve, and so the diagrammatic expectation is supported by our non-perturbative calculations [6]. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | of k into a charge of k′ if zk . Thus there should be a different stable string for each such k, with string tension σk. (Conjugate strings will have the same tensions.) Of course, it might be that we just have k fundamental strings joining such sources, in which case one will find σk = kσ and we have no new string. As we shall see, this is not the case: we find new tightly bound k-strings. The first k = 2 string appears in SU(4) and the first k = 3 one in SU(6). The values of σk are interesting because they carry information about confinement. Also because there are specific con- jectures about their values from M(-theory)QCD [8], from arguments about Casimir scaling [12], and from certain models, such as the bag model [13]. Moreover such strings may have striking im- plications [6,7] for the glueball mass spectrum as a function of N . |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Clearly the k = 2 string is tightly bound, and the string tension is incompatible, for example, with the naive bag model. What we find is that it falls between the MQCD and Casimir scaling predictions and, at the two sigma level, is con- sistent with both. We remark that a very recent higher statistics calculation [16] favours MQCD over Casimir scaling. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Figure 4. The D=3+1 effective string correction coefficient estimated from the masses of flux loops of different lengths (indicated by the span of the horizontal error bar). The solid line is what one expects for a simple bosonic string. For compar- ison the dashed line indicates the value for the Neveu-Schwartz string. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Figure 5. The ratio of k = 2 and k = 1 string tensions in our D=3+1 SU(4) ( ) ) and SU(5) ( ◦ • lattice calculations plotted as a function of a2σ. Extrapolations to the continuum limit, using a leading O(a2) correction, are displayed. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | Figure 6. The continuum topological susceptibil- ity in units of the string tension plotted against 1/N 2. A linear extrapolation to N = is shown. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | where λ is the ’t Hooft coupling that we keep . In the real vacuum this argu- fixed as N 1 ‘fermi’ and ment holds for instantons with ρ so these instantons are exponentially suppressed as N increases. Now the Monte Carlo changes Q by an instanton shrinking through small values a where it can vanish through of ρ down to ρ (Or the reverse process.) However the lattice. |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | This procedure is clearly of ambiguous validity for broad instantons, but is unambiguous for the high narrow peaks that correspond to the smaller instantons of interest to us here. In Fig.7 we dis- play the result of such a calculation, comparing the instanton densities in SU(2), SU(3) and SU(4) on lattices with (almost) equal volumes and lat- tice spacings when expressed in units of the string tension. We clearly see a drastic suppression of small instantons with increasing N . |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | are becoming available to many groups, would be to calculate the quenched hadron spectrum at large N – which is interesting since it approaches the correct spectrum of SU(N ) QCD as we take N at fixed non-zero quark mass. And then, → ∞ eventually, to dynamical quarks at large N and the η′ ... |
Processed_Glueballs,_strings_and_topology_in_SU(N)_gauge_the.txt | I am very grateful to the organizers for inviting me to participate in this interesting (and enjoy- able) workshop, and to my fellow participants for many useful comments and discussions. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | ABSTRACT Here we describe the Compact Array Broadband Backend (CABB) and present first results obtained with the upgraded Australia Telescope Compact Array (ATCA). The 16-fold increase in observing bandwidth, from 2 × 128 MHz to 2 × 2048 MHz, high bit sampling, and addition of 16 zoom windows (each divided into a further 2048 chan- nels) provide major improvements for all ATCA observations. The benefits of the new system are: (1) hugely increased radio continuum and polarization sensitivity as well as image fidelity, (2) substantially improved capability to search for and map emission and absorption lines over large velocity ranges, (3) simultaneous multi-line and con- tinuum observations, (4) increased sensitivity, survey speed and dynamic range due to high-bit sampling, and (5) high velocity resolution, while maintaining full polarization output. The new CABB system encourages all observers to make use of both spectral line and continuum data to achieve their full potential. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Given the dramatic increase of the ATCA capabilities in all bands (ranging from 1.1 to 105 GHz) CABB enables scientific projects that were not feasible before the upgrade, such as simultaneous observations of multiple spectral lines, on-the-fly map- ping, fast follow-up of radio transients (e.g., the radio afterglow of new supernovae) and maser observations at high velocity resolution and full polarization. The first sci- ence results presented here include wide-band spectra, high dynamic-range images, and polarization measurements, highlighting the increased capability and discovery potential of the ATCA. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Figure 1. The Australia Telescope Compact Array (ATCA), located at the Paul Wild Observatory near Narrabri, some 550 km northwest of Sydney. Displayed are five of the six 22-m Cassegrain antennas, here arranged in one of the hybrid configurations which include two antennas on the North-South spur. The latter was opened on 26 Nov 1998, just over ten years after the ATCA opening on 2 Sep 1988. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The Australia Telescope National Facility (ATNF) provides open access to a large range of radio telescopes: the 64-m Parkes telescope, the Australia Telescope Compact Array (ATCA), the 22-m Mopra telescope, and the Long Baseline Array (LBA); see Frater, Brooks & Whiteoak (1992). Fur- thermore, CSIRO’s ‘Australian SKA Pathfinder’ (ASKAP), consisting of 36 × 12-m dishes, is currently under construc- tion in Boolardy, Western Australia. With its expected 30 square degrees field of view at 1.4 GHz, provided by novel phased array feeds, ASKAP will be a fast 21-cm survey machine, designed to carry out large-scale H i spectral line, transients and radio continuum surveys of the sky (Johnston et al. 2007, 2008). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | In this paper we focus on the ATCA and its recent up- grade in bandwidth from 2 × 128 MHz to 2 × 2048 MHz. While greatly increasing the sensitivity of continuum and spectral line observations (a factor 16 in bandwidth gives a factor four in continuum sensitivity) or, alternately, reducing the required observing time for a given continuum sensitivity by a factor 16, this upgrade also enables simultaneous (full Stokes) multi-line observations, on-the-fly mapping, as well as searching for emission and absorption lines over large ve- locity ranges. The high-bit sampling further improves sensi- tivity, mapping speed and dynamic range for radio frequency interference (RFI) suppression. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | hybrid arrays†, one antenna (CA06) is fixed at a distance of 3-km from the western end of the track. Each antenna currently has a set of six cryogenically cooled low noise re- ceivers sampling the frequency range from 1.1 to 105 GHz (i.e., wavelengths from 3-mm to 30-cm), apart from CA06 which does not have a 3-mm receiver system (see Table 1). — A comprehensive description of aperture synthesis and radio interferometry techniques is given in the expert lec- tures at the regularly held Radio Synthesis Schools (e.g., Perley, Schwab & Bridle 1988, and Taylor, Carilli & Perley 1998) and also Thompson, Moran, & Swenson (2002). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Figure 4. ATCA primary beam FWHM as a function of fre- quency; values were computed with the miriad task pbplot. The currently available frequency coverage across all ATCA bands from 1.1 to 105 GHz range is indicated by the grey-shaded bands (see also the linear display of the ATCA frequency coverage in Fig. 3). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Figure 2. The Australia Telescope Compact Array (ATCA). Dis- played are several antennas on the East-West track pointing at a source near the horizon. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | of two independently tunable intermediate frequency (IF) bands, while CABB also increased the velocity resolution and delivers full Stokes parameters in all observing modes. Furthermore, the digitisation level has improved from 2-bits to 9-bits, increasing correlator efficiency and consequently lowering Tsys. In the following we list the advances provided by the new system. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | In this paper, we describe the design and construction of the Compact Array Broadband Backend (CABB), the first observational results, and the scientific potential of the up- graded instrument. The sections are as follows: (2) CABB overview, (3) CABB design, (4) CABB installation and op- erations, (5) CABB data reduction software, and (6) first CABB results. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The correlator is supplied with an 8 GHz wide IF band from the front-end. Within this, the observer can tune (within the receiver limits, see Table 1) two independent 2048 MHz windows (dual polarization) for correlation. This represents a factor of at least 16 increase in the useable bandwidth over the original ATCA correlator. Each 2 GHz window can be split into 2048, 512, 128, or 32 primary chan- nels (see Table 2). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Up to 16 high velocity resolution zoom windows can be placed anywhere within each 2048 MHz band. Each zoom window covers the width of one ”continuum channel” in the primary band and further splits it into 2048 high-resolution spectral channels (as illustrated in Fig. 9). Basic correlator configurations are listed in Table 2. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The design and construction of other correlators for radio interferometers such as Expanded Very Large Ar- ray (EVLA) and ASKAP are described in the respective overview papers by Perley et al. (2009) and DeBoer et al. (2009). SKA correlator advances are discussed by Bunton (2005) and software correlators by Deller et al. (2011). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | CABB combines high spectral resolution with full Stokes imaging, e.g., needed for the study of Zeeman split- ting in masers. All Stokes parameters are computed by the new correlator in all available modes. The original ATCA correlator was unable to provide cross-polarization correla- tions when used in modes with high spectral resolution. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The Compact Array Broadband Backend (CABB) upgrade, described here, has provided a new wide-bandwidth correla- tor for the ATCA, which is significantly more versatile and powerful than the original correlator (Wilson et al. 1992). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Sampling resolution and digitisation accuracy has in- creased from 2-bits with the original correlator (Wilson et al. 1992) to 9-bits with CABB (see § 3.2), increasing the correlator efficiency from <0.88 to ≈1 (i.e., reducing the ATCA Tsys by ∼14%), the dynamic range and tolerance to radio frequency interference (RFI). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Figure 3. ATCA system equivalent flux density (in Jy, log scale), for each observing band, obtained at high elevation and under reasonable observing conditions. — Note that one to four 2 GHz IF bands can be placed in each row, separated by no more than 8 GHz, for simultaneous single or dual polarization observations. — The measurements at frequencies from 1.1 to 50 GHz were made by Jamie Stevens in Jul/Aug 2010 using the CABB system, while the 3-mm measurements (80 to 105 GHz) were made by Tony Wong in Sep 2004. The values are based on hot-cold load measurements and include the atmosphere at the time of the observation. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Table 1. Some ATCA properties. — See the ATCA Users Guide (Table 1.1) for more details and up-to-date information. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Notes: ATCA observing information can be found at www.narrabri.atnf.csiro.au/observing, including a link to the CABB Sensitivity Calculator which is highly recommended to obtain observing characteristics (e.g., Tsys) at specific frequencies and correlator settings (see also Fig. 3). The ATCA primary beam size (in arcmin) can be approximated by 50/ν where ν is the observing frequency in GHz; the miriad task pbplot provides details of the primary beam model (see Fig. 4). ⋆ In 2010 the 1.5 GHz (20-cm) and 2.3 GHz (13-cm) bands were combined into one broad band covering the frequency range from 1.1 to 3.1 GHz (now referred to as the 16-cm band). Note that the 3- and 6-cm bands can be used simultaneously. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | the ringing that was commonly seen with the original ATCA and other correlators while observing narrow spectral lines. • Modes providing high velocity resolution (for spectral line studies), high time resolution (for the study of fast tran- sients), or pulsar binning come as an addition to the basic wide-bandwidth modes. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | ity, with its capability of being configured into many differ- ent operating modes. An important goal was to improve the spectral line capability of the instrument, particularly at the higher observing frequencies. Technologies employed, such as multi-bit digitising, high bandwidth digital data transfer over fibre optic cables, and advanced signal processing, were all considered to be applicable to the SKA. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The first stage of the CABB backend selects two inde- pendently tuneable 2 GHz bands from the active receivers. Two orthogonal linear polarizations are available from each band. The four resulting 2 GHz wide IF bands are digitised and sent over fibre optic cables to the central control build- ing. Here they enter the correlator, where the online signal processing takes place. A block diagram of the system is shown in Fig. 5, and details of the signal path and the pro- cessing are described in the following sections. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | the correlator can be configured to provide the most suitable compromise of sensitivity, speed and resolution for any particular observation. This is achieved by varying the width of each channel in the primary bands (see Table 2). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Associated analogue-to-digital conversion (ADC) re- quirements are pushing the limits of high-speed sampling and quantisation techniques. Nine bit data samples and 17- bit (minimum) internal signal paths through to the integra- tors provide performance approaching analogue correlation but with the obvious advantages of digital processing. Com- bined with excellent channel selectivity the system is signif- icantly more robust against strong interference than 2-bit correlators (Ferris & Wilson 2002). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | These new technologies are direct candidates for signal processing on the proposed Square Kilometre Array (SKA) and its precursors, such as ASKAP, the Australian SKA Pathfinder. CABB includes a total of eight antenna ports, allowing two antennas to be added to the existing six anten- nas of the ATCA, providing both a more powerful instru- ment and a mature test bed for the SKA pathfinders. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | transport and reception (§3.3), the digital filter bank cor- relator (§3.4), the signal processing (§3.5), the correlator hardware implementation (§3.6), the rear transmission mod- ule (§3.7), the front board (§3.8), and the control software (§3.9). In §3.10 we briefly look at the role of CABB in demonstrating novel technologies for SKA. The last sub- section (§3.11) gives an overview of the CABB installation. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The modular CABB IF conversion system resides in a shielded rack at the vertex of each antenna of the ATCA. The functionality of this system (see Fig. 6) includes pri- mary band filtering 4.2 to 12 GHz, stepped level control of 16 dB, sub-band selection for image management and fre- quency translation to suit the digitiser requirement. Defini- tion of the IF band is achieved by anti-aliasing filters cover- ing 2.048 to 4.096 GHz, corresponding to the second Nyquist zone of the analog to digital converter. |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | Slope compensated low noise amplification, digitally controlled step attenuation, image suppressing sub-band se- lection, and frequency translation are achieved in the sig- nal processing part of the conversion module. Support for maintenance is provided by low level monitoring outputs. Suppression of potential local oscillator (LO) leakage con- tamination from the split frequency design is provided by both fixed and tracking filtering. The provision of a YIG (yttrium iron garnet) tracking LO rejection filter is only made tractable by the integrated tuning management incor- porated into the optically slaved LO system (see below). |
Processed_The_Australia_Telescope_Compact_Array_Broadband_Ba.txt | The IF module provides slope compensated amplifica- tion, low level monitoring outputs and anti-aliasing band filtering. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | Within the framework of the Covariant formulation of Light-Front Dynamics, we develop a general non- perturbative renormalization scheme based on the Fock decomposition of the state vector and its truncation. The explicit dependence of our formalism on the orientation of the light front is essential in order to analyze the structure of the counterterms and bare parameters needed to renormalize the theory. We present here a general strategy to determine the dependence of these quantities on the Fock sectors. We apply our formalism to QED for the two-body (one fermion and one boson) truncation and recover analytically, without any perturbative expansion, the renormalization of the electric charge according to the requirements of the Ward Identity. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | x = 0, where ω is plane given by the equation ω · the four-vector with ω2 = 0. The particular case where ω = (1, 0, 0, 1) corresponds to standard LFD. We shall show in this study how one should determine, in a true non-perturbative way, the renormalization condition for systems composed of one fermion and at most N 1 bosons. The first calculation of the purely scalar system in CLFD, for N = 3, has been done in [4]. Non-perturbative calculation in a gauge theory (for N = 2) is given in [3,5] and in Yukawa theory (for N = 3) – in [6]. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | G The operator ˆτ acts on a given Fock component (p) is thus decomposed in an φi to give τ φi. infinite sum of the vertex functions Γi. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | re- component n with components the lates 2, ... depending on the system un- 1, n n, n der consideration. It is expressed in terms of bare quantities like the bare coupling constant g0, and/or physical quantities like the mass m of the constituent fermion with a mass counterterm δm. In order to calculate physical observables, these bare quantities and counterterms should be fixed from physical conditions, in the limit m. If we restrict ourselves to QED where M in order to compare with well established results in perturbation theory, the mass counterterm is fixed to get P 2 = M 2 for the physical state, and e0 is fixed to get the physical charge of the g0 ≡ electron, with α = 1 4π . |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | n > N . This is done ”by hand” in the coupled eigenvalue equations for the vertex parts Γ(N ) n . We have explicitly mentioned the dependence of Γ on N by a superscript. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | ii) When the bare coupling constant g0 and mass counterterm δm are fixed from physical ob- servables, this is done within the approximation In each intermediate state, of Fock truncation. for the calculation of a given Fock component, the total number, l, of particles in which a given state can fluctuate should be less or equal to N . Let us look for example at the renormaliza- tion of the fermion propagator in first order perturbation theory. The three contributions to the physical fermion propagator are indicated in Fig.1. These are from left to right the free propagator, the self energy contribution, and the contribution from the mass counterterm δm. The sum of these three contributions should be equal, at p2 = m2, to the free propagator. This fixes 0Σ(p2 = m2). This equation links two g2 δm = diagrams with two different Fock components. To keep track of the number of Fock component the counterterm is associated to (here 2), we use the superscript (l), and note δm(l), g(l) 0 . We shall see in the following section how this dependence should be calculated in practice. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | the electromagnetic form factor in the case of QED. In the determination of the state vector, the exchanged meson is indeed not an external particle and thus should participate to the count- ing rule of the total number of particles either in the initial or final state. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | To illustrate this fact, let us write down for in- stance the first contributions to the self energy involving at most two mesons in flight. These are indicated in Fig.2. The vertices (1) and (2) in Fig.2.(a) involve bare coupling constants in order to cancel divergences arising in Fig.2.(b) and (c). However, the vertex (3) in Fig.2.(b) should not be corrected in the same way as the vertex (4) since the vertex (3) can not be modified anymore by radiative corrections, while the vertex (4) can, as indicated in Fig.2.(c). The vertex (3) would however be corrected if one takes as bare cou- pling constant the one fixed from the calculation of the electromagnetic form factor, as indicated in Fig.3. We shall denote by ¯g(l) the ”Amputated” 0 Bare Coupling Constant (ABCC) to be used in the equation of motion. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | sponds just to a free fermion. The case N = 2 is calculated in section 4 for QED. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | For the calculation we consider in this work, we do not need to discuss higher n-point Green function. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | then determined from the equation of motion. In the approximation where one considers only one fermion and one boson state, these components are constants. This is however not the case in general. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | where k is the photon momentum, and L is a cut-off needed to give a meaning to the integral. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | For our purpose, we do not need to precise the regularization scheme we use. I(L) is a logarith- mically divergent quantity as a function of L. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | We recover exactly Eq.(15). In the r.h.s. of Fig.4, the final state can fluctuate into a fermion plus a boson. It is thus given by Γ(2) 1 . This is how- ever not the case for the initial state since there is already a boson in flight, so that it is given 1 , i.e. a free fermion state with Z (1) by Γ(1) 2 = 1. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | Since the contribution from the r.h.s. of Fig.4 is already ω-independent, we should not consider any ω-dependent counterterms, so that δ¯g(2) ω,i = 0. This validates the use of an ω-independent cou- pling constant in the N = 2 calculation of [3]. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | We would like to emphasize here that we did not do any perturbative expansion in order to express Z (2) in Eq.(19) starting from Eq.(14). The first 2 one is expressed in terms of the physical coupling constant g while the second one is a function of the ABCC ¯g0. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | factor are indicated on Fig.6. Direct calculation of the contributions from fermionic contact inter- actions gives no contribution at Q2 = 0 to the ω-independent contribution F1. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | By definition, the vertex counterterms δg(2) ω are chosen to cancel the ω dependent contributions arising from the diagrams in Fig.6, so that B1,2,3 in Eq.(23) are equal to zero. We do not detail here their calculation. They depend directly on the regularization scheme used in the evaluation of the loop diagrams. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | where the factor F (2) is the contribution coming from the ω-independent part of the loop diagram in Fig.6.(b) for N = 2. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | This is in complete agreement with the Ward Identity. This originates from two particular fea- tures of our strategy: the ABCC from (16) which implies (19) instead of (14), and the definition of the Fock-dependent BCC in (26) and (24). |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | We have presented in this contribution a sys- tematic strategy to calculate physical observables for fermionic systems composed of a fermion and N 1 bosons. This implies to implement a renor- malization scheme in a non-perturbative frame- work. Within the covariant formulation of Light Front Dynamics, we have shown how to fix the mass counterterm and bare coupling constant of the elementary Hamiltonian in a consistent way. As a check of our formalism, we treat the case of QED in the two body approximation. We have been able to recover, for the first time, the standard renormalization of the electromagnetic charge according to the Ward Identity (i.e. the bare charge should be equal to the physical one), without any perturbative expansion. This shows that no divergences are left uncancelled. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | used in the calculation of the state vector. This part is explicit in our formalism since it should be independent of ω. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | ii) Since we truncate the Fock expansion, the physical content of the bare coupling constant to be used in the state vector should be different from the bare coupling constant calculated from external meson coupling (by looking at the elec- tromagnetic form factor for instance if we deal with QED). These two couplings are of course the same in the limit of infinite Fock decomposition. |
Processed_Non-perturbative_renormalization_in_Light_Front_Dy.txt | Our results are very encouraging in the per- spective of doing true non-perturbative calcula- tions of bound state systems in a field theoretical framework. |
Processed_Supporting_Qualitative_Analysis_with_Large_Languag.txt | Qualitative analysis of textual contents unpacks rich and valuable information by assigning labels to the data. However, this process is often labor-intensive, particularly when working with large datasets. While recent AI-based tools demonstrate utility, researchers may not have readily available AI resources and expertise, let alone be challenged by the limited generalizability of those task-specific models. In this study, we explored the use of large language models (LLMs) in supporting deductive coding, a major category of qualitative analysis where researchers use pre-determined codebooks to label the data into a fixed set of codes. Instead of training task-specific models, a pre-trained LLM could be used directly for various tasks without fine-tuning through prompt learning. Using a curiosity-driven questions coding task as a case study, we found, by combining GPT-3 with expert-drafted codebooks, our proposed approach achieved fair to substantial agreements with expert-coded results. We lay out challenges and opportunities in using LLMs to support qualitative coding and beyond. |
Processed_Supporting_Qualitative_Analysis_with_Large_Languag.txt | CCS Concepts: • Human-centered computing → HCI design and evaluation methods; • Computing methodologies → Natural language processing. |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.