Filename
stringlengths 22
64
| Paragraph
stringlengths 8
5.57k
|
|---|---|
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | fast part of the outflow, but instead a slower wide-angle wind, or dense material in the outflow cavity walls. In- terestingly, another example of where C34S was found to be tracing a wide-angle outflow was the Class 0 pro- tostar HH212 (Codella et al. 2014), a source that also has signatures of a Keplerian disk. Codella et al. (2014) find that the C34S emission becomes more collimated at higher velocities, which can be also be seen to the north- west of mm1 in Fig. 21 at -52.71 km s−1 and again more collimated at -54.12 km s−1. They suggest this demon- strates that the wide-angle outflow of HH212 has an onion-like structure, with higher-velocity material closer to the jet axis, which may also be what we are seeing hints of here for AFGL 4176. |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | Figure 21. Channel map of the C34S (5-4) line, made using a combination of ALMA and APEX data, imaged at 1.41 km s−1 spectral resolution between -56.93 and -47.09 km s−1. Contours show 5, 10, 25, and 50 × 4 mJy beam−1 (the local rms noise), and the contours are filled with a colorscale to represent the channel velocity. The positions of mm1 and mm2 are shown as large and small plus signs, respectively, and the dashed line shows the position angle of the disk seen in 1.21 mm dust continuum (59◦). The red contour shows the 5σ 1.2 mm continuum emission. The gray ellipse traces the loops seen in the SE side of the outflow (see Section 5.1). |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | and -49.90 km s−1 channels by approximating the emis- sion by an ellipse with a fixed aspect ratio (shown in Fig. 21). This ellipse was measured to have semi-major and -minor axes of ∼2.2 × 1.9(cid:48)(cid:48) or ∼9200 × 8000 au, re- spectively. The distance on the sky of mm1 to the center of the ellipse is ∼1.4(cid:48)(cid:48). Assuming an inclination of 30◦, the true distance between the center of the ellipse and mm1 is therefore z0 ∼2.8(cid:48)(cid:48) or 12000 au at a cylindrical radius of (cid:36)0 =9200 au. From this, we could determine that the half-opening angle of the outflow or cavity wall at 1.5×105 au is ∼19◦. This is in reasonable agreement with the half-opening angle of 10◦ at 1.5×105 au that was assumed for our previous modeling, as this provided a good fit to the SED. |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | tracing dynamics driven by a separate source within this HII region, and/or by champagne flows due to varying density in the surrounding cloud. This seems the most likely explanation, as the bright bar of 1.23 cm contin- uum emission in the south of Fig. 17 is coincident with the peak of the integrated H68α emission seen in Fig. 22 and is centered on the velocity gradient within it, indi- cating that an independent source associated with this bar of continuum emission is probably producing the H68α emission. |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | We were only able to detect the remaining RRLs observed in December 2012 (H64α, H65α, H67α, and H68α) by finding their mean spectra over the area of sky covering the April H68α emission. We also fit these lines with Gaussians, and determine their peak flux densities to be 13.3±1.4, 12.9±1.9, 14.6±1.9, and 14.9±1.5 mJy beam−1, respectively, which are con- the velocities of sistent to within 1σ. the lines are −43.9±1.3, −47.4±2.4, −41.9±1.9, and −44.5±1.2 km s−1, and the linewidths are 25.5±3.2, 33.7±5.7, 30.5±4.5, and 24.4±2.9 km s−1, which are con- sistent to within 2σ. As no trends can be deduced from these, the line properties are thus consistent with a sin- gle flux, velocity, and linewidth and therefore likely trace a similar range of densities within the HII region. |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | We also detect H29α with ALMA. This H29α line emission is extended and faint, so that the resulting images are not useful. However, we measured a mean spectrum in an 5(cid:48)(cid:48)(R.A.) × 3.5(cid:48)(cid:48)(Dec.) ellipse centered on the same position as the circular aperture used for the ATCA RRLs. This resulted in a 9.3σ detection; the peak flux density, central velocity and linewidth from a gaus- sian fit are 0.58±0.1 mJy beam−1, -47.8±2.1 km s−1, and 25.7±5.0 km s−1, respectively. Although, due to large uncertainties, the linewidths of the H29α and H64, 65, 67, and 68α lines are consistent within errors, we calcu- lated the probability distribution of electron density ne for each pair of lines (i.e. H29α paired with one of the remaining lines) using Monte Carlo statistics and the method outlined in Keto et al. (2008). We then mul- tiplied these distributions to find the combined proba- bility distribution for ne and found the 99.7th percentile provided an upper limit of ne < 6.6 × 105 cm−3, which given the size of the emission would be consistent with a compact HII region. |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | Figure 22. First-moment map of the ATCA H68α emis- sion shown in colorscale. Black contours show integrated H68α emission (local σ = 30 mJy beam−1 km s−1× -5, 5, 6, 8, 10, 12, 14, 16). The black plus sign marks the position of AFGL 4176 mm1. The beam is shown in the bottom right corner; its properties are given in Table 3. |
Processed_A_Detailed_View_of_the_Circumstellar_Environment_a.txt | Figure 23. Ratio of the freefall to rotation timescales against Mgas for rotating structures around forming stars. The purple star represents AFGL 4176. The red, green, and black points represent the high-mass, intermediate-mass, and Herbig Ae sources presented in Beltr´an & de Wit (2016). The blue points represent new or updated sources, which are listed along with AFGL 4176 in Table 7. From bottom to top, black dashed lines indicate theoretical values of tff /trot for M(cid:63)=0, 10, and 25 M(cid:12) (see Section 6.1 for further details). |
Processed_Coding_Schemes_for_Line_Networks.txt | Abstract— We consider a simple network, where a source and destination node are connected with a line of erasure channels. It is well known that in order to achieve the min-cut capacity, the intermediate nodes are required to process the information. We propose coding schemes for this setting, and discuss each scheme in terms of complexity, delay, achievable rate, memory requirement, and adaptability to unknown channel parameters. We also briefly discuss how these schemes can be extended to more general networks. |
Processed_Coding_Schemes_for_Line_Networks.txt | nodes would have no processing or memory requirements, and would not introduce delay. This scheme would further adapt to unknown channel parameters. However, the achievable rate can only approach the end-to-end capacity of the overall channel, which is in general less than the min-cut capacity of the network. |
Processed_Coding_Schemes_for_Line_Networks.txt | In [5] the authors examined the benefits of intermediate node processing from an information theoretic point of view. Our work can be viewed as approaching the same problem from a coding theory point of view. |
Processed_Coding_Schemes_for_Line_Networks.txt | Networked systems arise in various contexts such as the public internet peer-to-peer networks, ad-hoc wireless net- works, and sensor networks. Such systems are becoming central to our everyday life. The networked systems today employ traditional coding schemes for end-to-end connections and are generally not tailored to the network environment. For example, for reasons of design simplicity, intermediate nodes at a network are only allowed to forward and not to process incoming information flows. However, as the size of communication networks grows, it becomes less clear if the benefits of the simple end-to-end approach outweigh those of coding schemes that employ intermediate node processing. |
Processed_Coding_Schemes_for_Line_Networks.txt | is well-known that if intermediate nodes are allowed to decode and re-encode the information sent by the source, –with no constraints on complexity and/or delay,– then the information capacity between a sender and a receiver is upper bounded by the min- cut capacity of the network, as described in [2]. A crucial in making schemes that employ intermediate node- point processing practical and attractive, is in realizing benefits without incurring excessive complexity and delay. |
Processed_Coding_Schemes_for_Line_Networks.txt | In this paper we propose coding schemes that employ intermediate node processing and discuss their performance. These schemes are based on fountain codes, a set of rate- less codes recently proposed [4], [7] that have a number of desirable properties for networked environments. We compare different coding schemes based on their complexity, delay, memory requirement, achievable rate, and adaptability; we will define these metrics precisely in Section II. |
Processed_Coding_Schemes_for_Line_Networks.txt | is organized as follows. In Section II we present our model and performance metrics in more detail. In Section III we describe our proposed coding schemes. In Section IV we discuss generalization to other networks; In Section V we compare our results with some related work in more details, and finally we conclude the paper in Section VI. |
Processed_Coding_Schemes_for_Line_Networks.txt | We consider a linear network that models a path between a source and a destination. The corresponding graph is com- prised of a source node, a destination node and a series of 1 intermediate nodes. The L edges between the nodes L correspond to independent memoryless erasure channels, and the information units sent over the ith link are erased with probability ǫi. |
Processed_Coding_Schemes_for_Line_Networks.txt | Fig. 1. A path between a source A and a receiver C with L=2 links. |
Processed_Coding_Schemes_for_Line_Networks.txt | assume that a packet transmitted at time d, if not erased, is received immediately at the next node in the chain. |
Processed_Coding_Schemes_for_Line_Networks.txt | We observe that, although it is possible to design a code over a single link that is both adaptable and is optimized for achievable rate and delay, the overall coding scheme cannot be adaptable if we want to jointly optimize for achievable rate and delay. Indeed, assume that ǫ2 = 0. Then the scheme that jointly optimizes the delay and the achievable rate requires node B to transmit (forward) only when it receives a new packet. However, if ǫ1 and ǫ2 are equal and large, then a large fraction of the packets will get erased. In order to optimize 1−ǫ2 packets for for delay, node B should transmit about each packet it receives, without waiting to receive the next packet from node A. Therefore a single scheme cannot be rate-optimal for both cases. |
Processed_Coding_Schemes_for_Line_Networks.txt | placed on these performance metrics. For example, consider a real-time application, where information is collected into blocks of k packets that are encoded and sent over the channel. In other words, we want to transmit the real-time information from a source, as it is produced. Assume that we have M such blocks. Then the delay overhead at intermediate nodes can be considered to be a “set-up” delay for the connection, experienced only once, and hence insignificant if M is large. On the other hand, the memory requirements at intermediate nodes may be restrictive. Indeed, there might exist a large number of connections (paths) that share an intermediate node that performs processing. Thus, the memory available for each individual connection might need to be scaled down accordingly. |
Processed_Coding_Schemes_for_Line_Networks.txt | Recall that our notion of delay is linked with the optimal time of communication over a single channel with equivalent min-cut capacity. Note however that with this definition, it is impossible to achieve a ‘zero delay’ scheme even for the simple network of Fig. 1. In fact, even if both links AB and BC provide perfect feedback, there is an inherent delay to be suffered due to the existence of sequential links. As we will see, even in this perfect setting, there is also a need for memory storage, in amounts that grow with k. In this section we will calculate the memory requirements, as well as the minimal delay which is incurred when perfect feedback exists; certainly no coding scheme that does not rely on feedback can transmit in less time. |
Processed_Coding_Schemes_for_Line_Networks.txt | The obvious optimal scheme in the presence of feedback is one where each node repeats transmission of each packet until it is successfully received at the destination. Node A then completes transmitting in time n ǫ). The operations at node B can be described using a Markov chain with states , indicating the number of received packets xi still to be sent at each time; therefore at each time i, xi packets need to be stored in memory. At each time (when xi = 0), ǫ) the state is unchanged, and with a probability 1 − ǫ), the state is increased or decreased with a probability 2ǫ(1 by 1, with equal probability. Therefore, after n time slots, the dynamics of this system resembles that of a random walk with a reflecting boundary at 0, over n′ = 2ǫ(1 ǫ)n steps; (there is slight correction, due to ‘longer stays’ at state 0, but for large n, the probability of being at that state is insignificant.) Thus the expected value of xn is the expected value of the absolute value of a random walk after n′ steps. Therefore E[xn] = O(√n′) = O(√2ǫk), where we have used that n ǫ). Node B then completes transmitting the remaining xn packets in a time d ǫ). Therefore, the ‘delay’ of this scheme ≈ is O(√ǫk/(1 ǫ)), while the expected memory requirement is O(√ǫk). |
Processed_Coding_Schemes_for_Line_Networks.txt | in linear network with L similar links, where L is a fixed finite number, each intermediate node incurs a delay of O(√ǫk/(1 ǫ)) and requires O(√ǫk) units of memory. |
Processed_Coding_Schemes_for_Line_Networks.txt | In this section we describe and compare a number of coding schemes for a line network with L links. In the next section we will discuss how these schemes can be extended to more general settings. |
Processed_Coding_Schemes_for_Line_Networks.txt | We will use the configuration in Fig. 1, with L = 2, as the illustrating example, and assume for simplicity that ǫ1 = ǫ2 =: ǫ, in which case n1 = n2 =: n. In all schemes below we will use as code C1 over the link AB, a fountain code, such as an LT-code or a Raptor code; as demonstrated in [4] and [7], these codes are low complexity, rate optimal, adaptable codes over erasure channels. Then for each different coding scheme, we will specify the code C2 over the link BC. A summary of the properties of all these schemes will be provided in Table I. |
Processed_Coding_Schemes_for_Line_Networks.txt | An obvious scheme is to use a separate code for each of the L links of the line network, and have each intermediate node completely decode and re-encode the incoming data. Then it is obvious that we can achieve the min-cut capacity by using optimal codes (e.g. LT-codes) over each link. However, the system suffers a delay of about kǫ/(1 ǫ) time-slots due to each intermediate node. Indeed, at node B, we can directly forward the (1 ǫ)n received coded bits without delay, and then, after decoding, create and send an additional ǫn bits over the second channel. |
Processed_Coding_Schemes_for_Line_Networks.txt | This straightforward scheme imposes low complexity re- quirements. We only need O(k log(k)) binary operations at each intermediate node to decode and re-encode an LT- code, and the complete decoding and re-encoding scheme has memory requirements of the order O(k). Moreover, LT-codes adapt to unknown channels in the sense defined previously. |
Processed_Coding_Schemes_for_Line_Networks.txt | The complete decoding and re-encoding scheme of the previous section is adaptable, rate optimal and has low com- plexity. However it requires each intermediate node to store in memory the entire k packets of information in order to re- encode. We propose a class of coding schemes, which we call systematic schemes, which minimize the memory requirement at the intermediate nodes, but require the knowledge of the erasure probabilities of the links. |
Processed_Coding_Schemes_for_Line_Networks.txt | the non- systematic packets are formed as random (sparse) linear com- binations of the systematic ones. More precisely, whenever a new packet is received at B, it is added to the storage space allocated to each of the non-systematic packets independently and with a (small) probability p. |
Processed_Coding_Schemes_for_Line_Networks.txt | Theorem 1: With p = (1 + δ) log(ǫk)/(ǫk) for δ > 0, the described systematic random code asymptotically achieves the capacity over the channel BC. Proof: [Sketch] Suppose k′ ǫ) systematic sym- ǫk. We will then bols are received at C, and let l = k wait for a further l + c log2(l) non-systematic symbols to be also received at C, where c > 1 is a constant. After eliminating the received systematic symbols, these linear combinations can l binary matrix, be described by a random (l + c log(l)) with i.i.d. entries which are nonzero with probability p = (1 + δ) log(ǫk)/(ǫk). The results of [1] can be extended to show that, if p > log(l)/l, the probability that such a matrix is not full-rank approaches zero polynomially fast with l. Using this and the law of large of numbers then, with high probability C can retrieve all the k symbols received at B, –e.g. by applying Gaussian elimination to this sparse matrix,– which can then be used to decode the fountain code C1. This code can decode the k information symbols from an average of k + c log(kǫ) received symbols at C, and hence this scheme rate optimal for large k. |
Processed_Coding_Schemes_for_Line_Networks.txt | The complexity of decoding this code is that of inverting kǫ matrix, which is O((kǫ)2 log(kǫ)). In fact, it the sparse kǫ can be shown that O(log(k)/k) is the smallest possible value for the probability p, and equivalently the density of the non- systematic part of the code, if the code is to be decodable with negligible overhead. In that sense, the scheme provided here offers the lowest decoding complexity for any such random code where the parity bits are chosen as linear combinations of the systematic bits with i.i.d. distribution. |
Processed_Coding_Schemes_for_Line_Networks.txt | The main advantages of this random scheme are its adapt- ability and optimality in terms of delay. The drawbacks are large memory requirement, and high decoding complexity, which is O(k2 log k) XOR operations on packets. |
Processed_Coding_Schemes_for_Line_Networks.txt | A closer examination of the proof of Proposition 1 reveals that, in order to make E[ 1 converge to zero, it is ] | sufficient that, for each column i, the matrix A contains at least i rows with Bernoulli(1/2) random variables at k + c log(k) the ith position; this will then guarantee that the size of K Vi is no more than 1/kc, for some c > 1, and we use (1) to obtain the desired result. The interpretation of this statement in the context of our coding scheme is that, in order to be able to decode with high probability at C, it is sufficient that for each i packets are successfully i = 1, transmitted over BC after B has received the ith coded packet from A. |
Processed_Coding_Schemes_for_Line_Networks.txt | Let αd and βd denote the number of packets successfully transmitted over links AB and BC respectively. Suppose now that we end transmission at a time n when C has received k + l packets, i.e., βn = k + l, where l = o(k) will be appropriately chosen. Then the number of packets that will βd; this, be received by C after a time d is equal to k + l αd. In other we would like to be at least k + c log(k) words, the sufficient conditions above require that at each time d = 1, βd be greater than βd behaves similar to a symmetric c log(k) in 1 one-dimensional random walk: ǫ) fraction of the time slots, xd remains unchanged, while in ǫ) fraction, it increases or decreases by 1 the other 2ǫ(1 with probability 1/2. Therefore, in n ǫ) time it takes to complete transmission as described above, xd’s movements are identical to n′ steps of a random walk , } where n′ = 2nǫ(1 2kǫ. Straightforward calculation ≈ kǫ log(kǫ)), then shows that with l = O( p n′ goes below the probability that l is polynomially small in k. This proves that, with high − probability, the k packets of information can be retrieved at C ) received packets. The overhead goes from k(1 + q to zero as k becomes large, and hence this coding scheme is asymptotically rate optimal. |
Processed_Coding_Schemes_for_Line_Networks.txt | mC0 for some integer m. It is straightforward to see that if we are employing a capacity-achieving coding scheme, it is sufficient to route the information along m parallel paths , Pm, where each path Pi consists of Li links. We can P1, then directly apply the coding schemes previously described to each path separately. |
Processed_Coding_Schemes_for_Line_Networks.txt | In practice, since coding schemes will employ codewords of finite block lengths, there might exist benefits in combining independent information streams [6]. Moreover, not all edges might be used at the same rate, for example because of cost considerations. |
Processed_Coding_Schemes_for_Line_Networks.txt | Consider a routing scheme that observes the flow conserva- tion principle and utilizes each edge at rate smaller or equal to its capacity. Since all the component codes are linear, the received symbols along a link l in the network can be described using an (nl kl) matrix, where kl is the number of information symbols sent along the link, and nl is the number of received symbols. The point we make in this section is that, as long as all such matrices corresponding to the intermediate links have full column rank, the end-to-end matrix that the receiver will have to decode in order to retrieve the information bits, will also be full rank and hence decodable. |
Processed_Coding_Schemes_for_Line_Networks.txt | Partitioning a matrix into parts, to create the equivalent matrix corresponding to splitting an input stream to multiple outgoing streams, such as node A in Fig. 2. • Multiplication of matrices, in order to create the equiva- lent matrix corresponding to serially concatenated chan- nels, such as nodes B and C in Fig. 2. |
Processed_Coding_Schemes_for_Line_Networks.txt | Finding the direct sum of matrices, to create the equiv- alent matrix corresponding to merging multiple input streams of a node into a single outgoing stream, such as node D in Fig. 2. |
Processed_Coding_Schemes_for_Line_Networks.txt | All these operations preserve the full-rank property. Thus, all coding schemes described in Section III can be directly applied over a more general network. However, for this general case, a thorough study of the delay and memory requirements for each scheme is not provided here. |
Processed_Coding_Schemes_for_Line_Networks.txt | takes advantage of intermediate node processing to approach the min-cut capacity. The authors model the departures and arrivals at nodes as Poisson processes and work out the queuing-theory aspects of the problem. The employed coding scheme allows inter- mediate nodes to transmit random linear combinations of the incoming packets over a finite field GF(q). The transmission rate approaches the min-cut capacity as q goes to infinity. This 1 scheme, as described in [8], requires O(k2(1 q )) operations to encode k symbols at the transmitter, O(k3) operations for 1 decoding at the receiver, and O(k2(1 q )) operations at each intermediate node. Moreover, the operations are over GF(q) that are more complex than binary operations. Intermediate nodes require storage capabilities for k packets over GF(q). The main benefit of the scheme in [8] is in terms of delay as we do not decode at each intermediate node. Indeed, complete decoding and re-encoding requires a delay of ǫn time-slots. However, note that the scheme in [8] achieves the min-cut rate for large q, i.e., assuming that we are able to send log2(q) bits per time-slot instead of one bit per time-slot as we assume. Thus in this sense it is not clear that the comparison is fair. |
Processed_Coding_Schemes_for_Line_Networks.txt | In fact, the coding scheme employed in [8] can be thought as employing the greedy random codes in Section III-C, where the linear combinations are performed over GF(q) instead of the binary field, and where the encoding matrix is not sparse. Thus our results can be viewed as an improvement over the coding scheme proposed in [8]. |
Processed_Coding_Schemes_for_Line_Networks.txt | In this paper we have examined the problem of communi- cation over a line network, where processing of information at the intermediate nodes is required in order to achieve the min- cut capacity, and we have included guidelines to extend our results to general networks. We have proposed coding schemes based on fountain codes. Each scheme has been analyzed and evaluated in terms of complexity, delay, memory requirement, achievable rate, and adaptability (see Table I). In general, there is a trade-off between these desirable properties, and an absolute best scheme is not claimed. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | As part of an extensive radio–IR–optical–X-ray study of ROSAT clusters of galax- ies in the Hydra region we have observed the bimodal Abell cluster A3528, located in the core of the Shapley Supercluster (z ≃ 0.053), with the Molonglo Observatory Synthesis Telescope at 843 MHz and the Australia Telescope Compact Array at 1.4 and 2.4 GHz. This is part I in a series of papers which look at the relationship between the radio and X-ray emission in samples of ROSAT selected clusters. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The radio source characteristics — tailed morphologies and steep spectra — are consistent with the effects of a dense intra-cluster medium and the pre-merging envi- ronment of A3528. In particular, we present evidence that the minor member of the radio loud dumbbell galaxy located at the centre of the northern component of A3528 is on a plunging orbit. We speculate that this orbit may have been induced by the tidal interactions between the merging components of A3528. In addition, the radio source associated with the dominant member of the dumbbell galaxy exhibits many of the characteristics of Compact Steep Spectrum sources. We argue that the radio emission from this source was triggered ∼ 106 yr ago by tidal interactions between the two members of the dumbbell galaxy, strengthening the argument that CSS sources are young. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Re-analysis of archive pointed PSPC data using multi-resolution filtering suggests the presence of an AGN and/or a cooling flow in the southern component of A3528. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | RXJ 12 54.6 −29 13 in the ROSAT All Sky Survey (RASS). The cluster parameters are given in Table 1. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | A3528, along with other clusters in the Shapley region, is included in a statistically complete, flux-limited sample of 42 ROSAT clusters located in a contiguous area covering ∼ 1750 deg2 in the Hydra region. These clusters have been the subject of an extensive study at radio–IR–optical–X-ray wavelengths (Pierre et al. 1994a). One of our objectives is to assess the extent to which bias may enter estimates of cluster masses and luminosity functions derived from X-ray surveys. This has important implications for studies of large scale structure formation in the universe. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | CentreACO: optical cluster centre from ACO in J2000 equatorial coordinates (RA, Dec) and galactic coordinates (l, b); AT: Abell Type; BM: Bautz-Morgan class; R: richness class; D: distance class; m1 m3 m10: weighted mean total V magnitude estimate for the first, third and tenth ranked cluster members uncorrected for Galactic extinction; C: background-corrected count of cluster members in the magnitude range m3 to m3 +2; cz: mean recession velocity; σ: velocity dispersion; and MV: virial mass, calculated from 30(39) galaxies within radius 2(3) Mpc from CentreACO; m∗: characteristic Schechter (1976) bJ magnitude. References: 1 =Abell, Corwin & Olowin 1989; 2 =Quintana et al. 1995; and 3 =Raychaudhury et al. 1991. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Case (i) probably corresponds to significant AGN contribu- tion to the X-ray flux, but higher resolution X-ray images are required to be conclusive. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Both cases are linked intimately to the dynamical evolution of clusters. Substructure in the X-ray brightness distribu- tion, and in the spatial and velocity distributions of galaxies, are signatures of clusters which are not virialised and are un- dergoing mergers. The existence of various types of extended and distorted radio sources in X-ray luminous clusters, no- tably the so-called tailed sources, is also strongly correlated with dynamic activity in clusters (eg. A2256, R¨ottgering et al. 1994). This is because their special characteristics (spec- tra and morphology) are believed to be the result of strong interaction with the cluster environment. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Radio observations therefore hold promise as a means of revealing clusters for which the gravitational potential based on X-ray data may be affected by AGN contami- nation or non-hydrostatic/non-isothermal conditions. Ulti- mately, it is hoped that our statistically complete sample of clusters, which is chosen independently of their radio prop- erties, will clarify the relationship between cluster X-ray and radio emission. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The X-ray–optical analysis of the 40 brightest clusters in our sample has been completed (Pierre et al. 1994c) and the Shapley region in particular has been studied exten- sively at optical and X-ray wavelengths (Vettolani et al. 1990; Day et al. 1991; Fabian 1991; Raychaudhury et al. 1991; Nobuyoshi & Suto 1993; Bardelli et al. 1994; Metcalfe, Godwin & Peach 1994; Quintana et al. 1995). The southerly declinations (−40 <∼ δ <∼ −20◦) of 27 of the clusters from this sample, including A3528, are well suited to radio obser- vations with the Molonglo Observatory Synthesis Telescope (MOST: Mills 1981; Robertson 1991) and the Australia Tele- scope Compact Array (ATCA; Frater & Brooks 1992). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | of the PSPC image to build a more comprehensive multi- wavelength picture relating the radio emission in A3528 to the pre-merging cluster environment. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | We describe the radio observations and data reduction in section 2. In section 3 we present our radio source lists and images, discuss morphologies of the detected sources, and compare them with previous observations. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | In section 4 we discuss and relate the optical, X-ray and radio characteristics of A3528. We look at the spatial and velocity distribution of the luminous mass and make radio-optical identifications. A re-analysis of the X-ray data allows us to identify possible excess emission from A3528S. We then discuss the radio sources in A3528, differentiat- ing between those associated with brightest cluster members (BCMs) and non-BCMs, and explore their relationship with the pre-merging cluster environment of A3528. Our conclu- sions are summarized in section 5. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The field of A3528 was observed with the MOST at 843 MHz and the ATCA at 1.4 and 2.4 GHz. Details of the observa- tions are summarized in Table 2. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Follow-up observations with the ATCA at 1.4 and 2.4 GHz used the 6C array configuration (interferometer spacings 153 m–6 km) in ‘snapshot’ mode (Burgess & Hun- stead 1995), time-shared with five other targets in a 1-hour cycle. Total on-source integration time was ∼2 hours. Two orthogonal linear polarizations were recorded. The observa- tions were centred on A3528N so primary beam attenuation has lowered the sensitivity for A3528S, especially at 2.4 GHz ′ where the primary beam is 22 FWHM. The primary ATCA flux density calibrator, PKS B1934−638, was observed at the start of the run and the secondary gain and phase cali- brator B1320−446 was observed every 30 mins. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The MOST data were reduced using custom software (Cram & Ye 1995). The ATCA snapshot observations were reduced predominantly in aips using the techniques of Burgess & Hunstead (1995). However, miriad (Sault, Teuben & Wright 1995) was used specifically to calibrate the visibility data because of the special polarization characteristics of the ATCA. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | analysed using aips. All images were first corrected for pri- mary beam response. Source lists were then generated for each radio image. Elliptical gaussians were fitted, as de- scribed by Condon (1997), to all unresolved and slightly resolved (less than two beamwidths in size) sources in the ra- dio images. The task VSAD (W. Cotton, priv comm), writ- ten for the NRAO VLA Sky Survey (NVSS), was used to give source positions and peak and integrated flux densities. Parameter errors were calculated using formulae similar to those used for the NVSS (Condon et al. 1996). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | In cases where the emission was well resolved we used in- stead the task TVSTAT to sum inside an irregular polygon, tightly defined around the source, and obtain the integrated flux density. An estimate of the background was made from a second measurement using a larger concentric polygon but excluding any other sources. In these cases the largest angu- lar size and position angle were measured directly from the image. The position corresponds to the location of the peak in the emission. There was no obvious way of accurately and objectively estimating the errors in these parameters and so none are given. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Source lists at each frequency were then correlated. As the field of view at each frequency is different we do not provide a catalogued entry at some frequencies. For every catalogued source the quoted position is measured from the highest frequency image in which the source was detected, taking morphology into account. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Spectral indices α (where Sint ∝ ν α and Sint is the integrated flux density at frequency ν) were calculated for sources detected at two or more frequencies. It is important to note, however, that no attempt was made to match the uv coverage of the observations and, therefore, values for extended sources should only be regarded as indicative. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Table 3 lists all the detected radio sources with peak flux densities down to 5σ; non-detections were assigned a 5σ up- per limit. Table 4 lists parameters for all of the sources which could be deconvolved and/or identified with cluster galaxies (See Sec. 4.1.2). As the cluster radio sources are at known distances this allows us to derive physical quantities such as the integrated spectral power. The largest projected linear size was also determined for those cluster sources that could be deconvolved. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | In A3528N there is a strong radio source A coincident with the BCM. MOST detects an additional source B of comparable strength to the south and a weak source E to the East. In A3528S we also find a strong radio source C coincident with the BCM and a second weaker source D to the south-east. Optical identifications will be discussed in detail later (see Sec. 4.1.2). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Notes to Table 3: Parameter errors for the sources fitted with elliptical gaussians (∆RA, ∆Dec, and quantities in brackets) were based on formulae similar to those used for the NVSS (see text). A1 and A2 are blended at 0.8 GHz, so the numbers in columns 13 and 14 are repeated. D was too weak and too far down the primary beam for reliable measurements at 1.4 and 2.4 GHz. (1) catalogue ID in order of RA; (2–5) equatorial coordinates, epoch J2000; (6) & (7) projected offset from the centre of A3528N; (8) & (9) projected offset from the centre of A3528S; (10) spectral index α defined by S ∝ να where S is the integrated flux density at frequency ν; (11) observing frequency; (12) flag: 1 =TVSTAT used to calculate the peak and integrated flux, 2 =deconvolution possible (see table 4), 3 =sources blended; (13) peak radio flux density (per beam); (14) integrated radio flux. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Notes to Table 4: Parameter errors for the sources fitted with elliptical gaussians (quantities in brackets) were based on formu- lae similar to those used for the NVSS (see text). A1 and A2 are blended at 0.8 GHz, so the numbers in columns 3–6 are re- peated. D was too weak and too far down the primary beam for reliable measurements at 1.4 and 2.4 GHz. (1) catalogue ID from table 3; (2) observation frequency; (3) log10 of the integrated spectral power; (4) and (5) largest angular size and corresponding physical size; (6) projected position angle. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The ATCA observations give structural information on scales <∼ 10 and allow us to assign morphological classifica- tions. We find an exceptional number and variety of tailed sources. In particular, it is now clear that A is actually two separate sources, A1 and A2 in Fig. 2, each identified with one of the nuclei of this putative dumbbell galaxy. E was not detected at either 1.4 or 2.4 GHz while D was too weak and too far down the primary beam at 2.4 GHz to obtain a reliable image. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | A1 is a symmetric double source but has a spectral power at 1.4 GHz which is below the FR I/II break (Fanaroff & Riley 1974) in rich clusters (Burns et al. 1994a). Its projected linear extent of ∼ 30 kpc is well below the typical size (150– 300 kpc) for an edge brightened double (Miley 1980). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The morphology of B is reminiscent of the λ21 cm observa- tion of the prototypical narrow-angled tailed (NAT) source 3C 83.1B/NGC 1265 in the Perseus cluster (O’Dea & Owen 1986). B also has a similar linear extent but appears to more asymmetric and lacks a clear core component. The differ- ences may be due to a combination of projection effects and the poorer linear resolution of our observations. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Figure 2. ATCA contours of source A overlaid against a background of the DSS-II showing the two separate sources A1 and A2, each associated with one of the components (d1, d2) of the putative dumbbell galaxy. Left: 1.4 GHz, contour levels −1.5, 1.5, 3, 6, 10, 20, 40, 80 mJy/beam. Right: 2.4 GHz, contour levels −1, 1, 2, 4, 8, 16, 32 mJy/beam. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | C is a classic example of an FR I (edge darkened) wide- angled tailed (WAT) source similar in the bending sequence to B1610 − 608 (Jones & McAdam 1992). Its radio power is just below the FR I/FR II break which is typical of WATs (O’Donoghue et al. 1990). It has a relatively small extent (∼ 50 kpc) although the size is less well defined than A1. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | did not separate the two individual sources A1 and A2. Fi- nally, Schindler (1996) observed this cluster using the VLA in its CnB configuration at both 1.4 and 0.327 GHz. The 0.327 GHz observations indicate that D has a long tail ex- tending to the W. Our 843 MHz MOST image shows only a small extension in the direction of the tail suggesting that the tail has a very steep spectrum. Schindler’s observations at 1.4 GHz have lower resolution than the ATCA images shown in Figs. 2–5. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The morphology of D is not clear from the 1.4 GHz. We appear to resolve a core and stronger emission to the W. This unusual morphology may be due to the chance superposition of unrelated sources, but is more likely the result of missing extended flux. The 2.4 GHz image is not shown. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | In addition, this cluster has been observed at 1.4 GHz as part of the NRAO VLA Sky Survey (NVSS). These images have a sensitivity and beam size (45′′FWHM) similar to the MOST but with better N-S resolution. The NVSS image shows E quite clearly. D shows a vestigial tail while C shows an extension to the south. In fact all four main sources are clearly extended. We measured the flux densities of sources A–E using TVSTAT, and a comparison of the NVSS results with our ATCA measurements and those of Schindler (1996) in Table 5 confirms that we are missing extended flux. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Previous radio observations of A3528 available in the liter- ature are summarized in Table 5. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | ing only the 1.4 and 2.4 GHz ATCA images. The sources are clearly separated in the 2.4 GHz image. In the 1.4 GHz image there is some blending at the lowest levels but most of the emission is clearly separated. To estimate the uncertainty due to blending we summed over different polygonal regions around the two sources. The resultant integrated flux den- sities where never more than 10% from the quoted values, corresponding to an error of ∼ 0.1 for the spectral index. Both A1 and A2 have a steep spectrum (α ∼ −0.7). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The spectrum of B appears to be curved and this may be an indication of spectral aging in this source. Using the NVSS data point and the low frequency data (327 and 843 MHz) we found that D has a steep overall spectral index (including the tail) of α ∼ −1.0. E has an even steeper index of α ∼ −1.2 between 0.8 and 1.4 GHz. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Figure 4. ATCA contours of C overlaid against a background of the DSS-II. right: 1.4 GHz, contour levels −3, 3, 6, 10, 20, 40, 80, 160 mJy/beam Right: 2.4 GHz, contour levels −3, 3, 6, 10, 20, 40, 100 mJy/beam. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Figure 5. ATCA contours of D overlaid against a background of the DSS-II. 1.4 GHz, contour levels −2, 2, 2.5, 3, 4, 5 mJy/beam. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Figure 6. Plot of the radio spectra for sources A(top), B(middle) and C(bottom). Triangles: literature values (Table 5); asterisks: this paper (Table 3); solid line: best fit power law giving αA = −0.8, αB = −0.8 and αC = −1.3. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Figure 7. Luminous mass distribution for A3528 overlaid on the DSS-II. (a) Luminosity weighted galaxy surface density plot; the contour increments are equal in flux and correspond to 16.8, 16.1, 15.6, 15.3, 15.1 and 14.9 bJ mag arcmin−2. (b) Galaxy surface density plot; the contour levels are 0.9, 2.6, 4.3, 6.0, 7.7 and 9.4 galaxies arcmin−2. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Notes to Table 6: B is identified with galaxy g1 (see Fig. 3). Since COSMOS has difficulty in deconvolving partially blended objects we relied on the literature for the optical positions of d1 and d2 (see Fig. 2). (1) catalogue ID from Table 3; (2) COSMOS position with PPM correction (Drinkwater et al. 1995); (3) COSMOS apparent magnitude (accurate to about 0.5 mag.); (4) redshift: 1=Quintana et al. 1995, 2=Postman & Lauer 1995; (5) semi-major axis; (6) semi-minor axis; (7) major axis position angle E of N. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | We carried out a search for identifications for the radio sources in Table 3 with COSMOS objects which were con- sistent with being cluster members. The results are given in Table 6. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | were made from an examination of the radio-optical overlays in Figs. 1–5. We have already noted (Sec. 3.2) that A1 and A2 are identified with the two members, d1 and d2, of the putative dumbbell galaxy (and BCM of A3528N) included in a study undertaken by Gregorini et al. (1992, 1994). A1 is centred directly on the dominant component d1 and, in projection at least, the radio emission does not appear to extend beyond the optical halo. There is a small but signif- icant positional offset between d2 and the peak of the radio emission of A2, in the direction of the radio tail. d1 has a line of sight peculiar velocity relative to the mean velocity −1 (Postman & Lauer 1995). The of A3528N of ∼ +128 km s velocity of d2 with respect to d1 is consistent with being −1). On the other hand the radio zero (∆V = 108 ± 141 km s morphology suggests d2 has a substantial transverse velocity with respect to d1 and the cluster. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | In the case of B the 1.4 GHz image does not distinguish clearly between the two potential galaxy identifications, g1 and g2 in Fig. 3. The redshift of g1 confirms it as a clus- ter member, while the apparent magnitude of the brighter galaxy g2 is consistent with it being a cluster member. Based on the 2.4 GHz image we conclude that B is most likely as- sociated with g1. However, we do not observe a clear radio core, so it is still possible that B is associated with g2 and the radio emission has become completely detached from the host galaxy; a similar configuration is seen in the source J1324−3138 in A3556 (Venturi et al. 1997). Better radio data are needed to test this hypothesis. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | C is identified with the brightest galaxy in A3528S, a confirmed cluster member. The eastern knot of D is coin- cident with the core of another galaxy with an apparent magnitude consistent with it being a member of A3528S. E was the only unresolved or slightly resolved source identified with a cluster galaxy. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Figure 8. Multi-resolution filtered contour image of A3528 (not exposure corrected, see text) overlaid on the exposure map (greyscale). The image is 42′ × 42′, North is up and East is to the left. The contours are drawn in logarithmic intervals: 23 and 362 raw counts/arcmin2 for the lowest and highest contours re- spectively. The pattern of dark grey lines in the exposure map corresponds to obscuration by the rib structure. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | tures on different scales, as well as to enhance extended emis- sion (a simple gaussian filter was used by Schindler 1996). ′ at At an off-axis distance of 30 1 keV. We have tuned the filter to fully exploit this limited spatial resolution, and the lowest contours — except those close to the obscured regions — are at least 3.7σ. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | On the other hand, it is extremely difficult in practice to produce an exposure-corrected image with this technique since our analysis is based on individual photon statistics, whereas the exposure map is averaged over time. The sharp shadowed boundaries present in these observations would produce singularities. In particular, the region between the two clumps is severely affected by a reduction in exposure (a factor of ∼ 2 for the interface region and ∼ 10 for the nodes of the support). This means that contours across the support should be about twice as high as shown in the figure, and the outermost contours are unreliable. Despite this, our re-analysis of the data suggests the presence of a pointlike source in the southernmost maximum. A S-N cut through each peak is displayed in Fig. 9 and these show significant differences in shape. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | As to the optical counterparts of these X-ray features, we are unable to rely directly on the given satellite attitude solution (which often shows slight inaccuracies) since a few arcsec offset at the detector centre — as well as roll angle uncertainty — would result in much higher displacements at 30′off-axis distances. Indeed, we find a 20–30′′offset between the two X-ray centroids and the expected values correspond- ing to the radio positions. Unfortunately, no point sources were present outside the cluster at both wavelengths to pro- vide independent astrometry. Close correspondence between ′′ ′′ the separation of the X-ray (828 ) ) and radio centroids (813 — a difference of 15′′, small compared with the X-ray PSF — makes it likely that the X-ray maxima are associated with the BCMs of A3528N and A3528S and their associated ra- dio sources, A and C respectively. Furthermore, the excess component of X-ray emission in A3528S, whether it is due to the AGN or a cooling flow, is most likely to be coincident with the radio-loud BCM. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | Such conditions may partly explain discrepancies be- tween the determination of the cluster potential from grav- itational lensing (Miralda-Escud´e & Babul 1995) and from spectroscopic and X-ray studies. It may also explain why some clusters (eg. Hydra I: Fitchett & Merritt 1988) do not fit well on the LX – σ relationship (Quintana and Melnick 1982). Other examples of clusters displaying all or most of the characteristics of an advanced merger are A754 (Henry and Briel 1995; Zabludoff & Zaritsky 1995) and the Jones & Forman (1984) nXD clusters Coma (Biviano et al. 1995), A2255 (Burns et al. 1995) and A2256 (Edge, Stewart & Fabian 1992; R¨ottgering et al. 1994). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | A3528 fails to display many of the characteristics of a merger. For instance Raychaudhury et al. (1991) indi- cated likely cooling flows in both A3528N and A3528S; while Schindler (1996) was only able to specify upper limits of a Hubble time for the cooling time, it is likely to be much less than this (Sec ??). Temperature substructure, however, is possibly the most sensitive probe of cluster-cluster merg- ing (Roettiger et al. 1996). Indeed, Schindler (1996) found evidence for temperature substructure in A3528 and, based upon the results of N-body and hydrodynamic simulations, concluded that gas is being heated in the merging front be- tween A3528N and A3528S prior to collision. While the ev- idence appears to favour a pre-merging scenario we cannot totally discount the possibility that A3528N and A3528S have already passed through each other. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | There is now considerable evidence that tailed radio sources are associated with substructure and therefore mergers, in clusters (McHardy 1979; Valentijn 1979; Burns et al. 1994b; Doe et al. 1994; Venkatesan et al. 1994; Bliton et al. 1995; Gomez et al. 1997). Hydro simulations of jet propagation in merging clusters (Loken et al. 1995), as well as observations (Burns et al. 1994b), indicate that large-scale bulk flows and resulting shear discontinuities and shocks are likely to influence the morphology of extended radio sources. Tidal interactions between subclusters may also affect the peculiar velocities, and therefore the radio morphology, of galaxies through ram pressure effects. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | A3528 has an exceptional number and variety of tailed radio sources. Is this due to the “cluster weather” gener- ated by components which have already passed through each other, as has been proposed for Coma (Burns et al. 1994c), or, if A3528 is in a pre-merging state, is it the result of tidal interactions between the two subclusters beginning to affect the peculiar velocities of galaxies. In the following sections we attempt to identify the dominant physical mechanism shaping the morphology of each of the sources in A3528 and thereby determine if the morphologies are linked, either di- rectly or indirectly, to the dynamical state of the cluster. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The bright galaxies at the centres of A3528N and A3528S respectively are both moderately powerful radio sources, A1 and C. Their formation and evolution are likely governed by their location as their radio jets interact with or are impeded by the high ambient pressure of the ISM/ICM at the bottom of the cluster potential well (Rudnick & Owen 1977; Simon 1978; Burns, Gregory & Holman 1981; Ball, Burns & Loken 1993). It is also possible that a cooling flow may in turn fuel the radio core directly through mass accretion (Burns et al. 1981; Valentijn & Bijleveld 1983; Zhao, Burns & Owen 1989; Burns 1990; Ball et al. 1993). |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The radio emission from A1 appears, at least in pro- jection, to be contained completely within the optical halo of the galaxy. Its small size and symmetric double structure may be due to recent triggering (Sec. 4.4.4) so there has not been sufficient time for the jets to push the working surface outside the ISM of the host galaxy. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | In addition to the two BCM sources A1 and C there are three extended radio sources associated with non-BCMs: A2, B and D. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | by the observations and therefore appear to be a single tail. This is probably the case for A2. |
Processed_Radio_Study_of_X-ray_Clusters_I:_A3528_---_a_pre-m.txt | The radio observations reveal, therefore, that d2 is not bound to d1 and that this is not a true dumbbell system. Galaxies are often seen close in projection to BCMs at the centres of clusters but with a significantly different redshift implying high peculiar velocities along the line of sight (Mer- rit 1985). In the case of d2, however, the peculiar velocity is inferred to be almost entirely in the plane of the sky. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.