Scribd Logo
Upload

Welcome to Scribd!

  • DGB93 CurrSc 65 827to35 (Large Scale Structure in The Universe)

    Uploaded by

    Dilip G Banhatti
    26 views9 pages
    A survey of som cosmographic methods is presented here, with examples of results on the large-scale structure in the universe using these methods. Showing structure upto a few 100 Mpc, the distribution of light tends to uniform random (i.e., Poisson) beyond that scale. The recently discovered substantial large-scale coherent flow of all visible matter relative to the rest frame of the 2.73 K cosmic background radiation is not discussed, nor are models & their predictions, including the dynamical importance of dark matter. The material presented forms part of the background needed to meaningfully discuss the 10 ppm ripples COBE observed in the 2.73 K radiation in 1992.

    Original Title

    DGB93 CurrSc 65 827to35 {Large Scale Structure in the Universe}

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    A survey of som cosmographic methods is presented here, with examples of results on the large-scale structure in the universe using these methods. Showing structure upto a few 100 Mpc, the distribution of light tends to uniform random (i.e., Poisson) beyond that scale. The recently discovered substantial large-scale coherent flow of all visible matter relative to the rest frame of the 2.73 K cosmic background radiation is not discussed, nor are models & their predictions, including the dynamical importance of dark matter. The material presented forms part of the background needed to meaningfully discuss the 10 ppm ripples COBE observed in the 2.73 K radiation in 1992.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    26 views9 pages

    DGB93 CurrSc 65 827to35 (Large Scale Structure in The Universe)

    Uploaded by

    Dilip G Banhatti
    A survey of som cosmographic methods is presented here, with examples of results on the large-scale structure in the universe using these methods. Showing structure upto a few 100 Mpc, the distribution of light tends to uniform random (i.e., Poisson) beyond that scale. The recently discovered substantial large-scale coherent flow of all visible matter relative to the rest frame of the 2.73 K cosmic background radiation is not discussed, nor are models & their predictions, including the dynamical importance of dark matter. The material presented forms part of the background needed to meaningfully discuss the 10 ppm ripples COBE observed in the 2.73 K radiation in 1992.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    of 9
    REVIEW ARTICL Large-scale structure in D. G. Banhatti Schiol of Physies, Madurai Kamarij University, Madura 625 021. dia ‘A survey of some cosmographie methods is presented here, with examples of results on the large-scale ure in the universe using these methods. Showing structure up to few 100 Mpe, the distri- bution of light tends to uniform random (ie, Poisson) heyond that seale, The recently discovered substantial large-seale coherent flow of all visible matter relative to the rest frame of the 2.73 K cosmic background radiation is not discussed. nor are models and their predictions, including the d: mieal importance of dark matter. The material presented forms part of the background needed 10 meaningfully diseuss the 10 ppm ripples COBE observed in the 2.73 K radiation in 1992. THE large-scale structure in the universe, ay we obser~ vationall ye it, consists of the distribution and motion galaxian objects and the distri- bution of diffuse radiation over the various windows: through the clectfomagnetic spectrum, At present this means: (i) the distribution and motion of galaxies seen in the optical band over most of the sky, to a depth of 20.1 (Zwichy and Abell surveys of clusters, Center for Astrophysics (CIA) survey, ete.)!; (i) the distribution of galaxies seen in the optical band over a portion of the celestial sphere deeper than == 0.1, perhaps to =>1 (Cambridge Automatic Plate-measuring Machine (APM) survey Oxford-Toronto (QDOT QIGC) survey ii) the distribution of infrared sources over most of the sky (InfuRed Astronomical Satellite (IRAS) survey)": liv) the diffuse X-ray back- ground which maps the overall distribution of clusters of galaxies and active galaxies; the nearer clusters identified with the broad maxima (mainly High Energy Astronomical Observatory 2 (IEAO 2 Observatory) satellite results)': (v) the distribution of extragalactic radio sources (galaxies and quasars) over the celestial sphere much deeper than == 0.1, perhaps to 23 (many low-frequency (< | GHz) radio surveys)* and (vi) the 2.73 K cosmic blackbody radiation peaking at microwaves (ie, emA) and its (en)isotropies. obser ved on scales from few arcminutes to few degrees, including the recent very sensitive Cosinic Background Explorer (COBE) satellite results? showing an ani: tropy of 10 ppm on scales > 7%. All these indicators instein Presented at a one-day meeting on COBE Results om 22 January 1993 a the fnsttute of Mathematical Sciences, Madras 600 113, Inala CURRENT SCIENCE, VOL. 63. NO. 11, 1ODECEMBER 1993 the universe Giveanalh) (QDOT atte Zwicky COA | an] Jima APNE x. Ray | [cone iteynes J" once ho SHALLOWEST 4-4 — eH rH 5 + beePEsT FOSSIL) igure |, Surveys of targe-seate sirueture and determiaants of have been listed above roughly in order of increasing depth: from the shallowest to the deepest possible (see Figure 1), The cosmic blackbody radiation is the deepest possible we can see. Matter was opaque to radiation before radiation decoupled from it and started expanding dependently. No imprint of any structure before this decoupling epoch (= = 150) is retained in the light we receive. Information about earlier epochs does come to us indirectly, ¢-g. from the abundance of light elements, -sele structure in the universe Dark matter On scales from galaxies to groups to clusters to superclustets and so on, dynamical mass as inferred from observed speeds of constituents is more than the visible mass inferred from their mass-to-Light ratios, The discrepancy increases with Inereasing seal. For structures on the various scales to be sufficiently long lived (or stable), dark mater must be present, and dominate the dynamics more and more on larger scales, So models must incorporate dark matter in some way However, | only treat different ways of describing the observed large-seale structure, Ieavit the most part. models alone for ‘Methods Analysis of observations of discrete objects (galaxies, groups of clusters of galaxies, radio sources, voids, etc.) is essentially the examination of the properties (Tike clustering) of (possibly ted) point seis. If the distance (ie known, we have a three-dimensional point set. If not, the point set is two-dimensional, consisting of the project. ions of the objects’ positions on the celestial sphere Some methods use these point sets as such. while others luse smoothed versions. I'sketch below the methods that redshift) of each object in a survey is REVIEW ARTICLE: Fable 1. Source densities for U1 flelds. (Banhatti ‘9OMN246 7P-10P) No Texas couns él counts ea of Seedensiy (x) ste density Feld (ee) ses (ee) e (Te (dee?) ao $0 1ast02R 027 +033 M31 = 0041 = 41 2 si 2onta2s 026 +106 121° 236 ~ 19084 351 ea 9 .2ar034 Ot $124 2282035 = 190" + 54 DA 240 o788 4 56 2 iasso29 0a 095 Van + 1331-099, 97 271378026 030 Las - snot si8 Whe 20 41 1682026 026 038 1602025 1740-4978 Ga" + 10° 240 39 1S6025 0.26 ~ 068 = 104251 Ws 25° 20 32 Laes023 0.26 in = 212° +65 @ +00 20 53212029 26 rida 113-099 246 41 1664026 029 -030 250 50 1098028 0.26 +098 2000.29 10.08 dew In the fitst column, the source same or field designetion approximate galuctc exordinates are given inthe second In Figure 2: Disteibutian oF aver 180 voids inex have been used, and indicate broadly the kind of information that each method can give, with examples of results obtained. | have described the newer methods in more detail than the older ones | Binning or counting the number of objects in diffe rent parts of a survey is the simplest method. This may be done in three dimensions if redshifts (indicators of 28 hen, followed by the R.A, and Dee. (epoch 1950). The tol coordinates, Limiting teashunt 30,000 km depth) are available, or else in two dimensions (on the celestial sphere). Sometimes the (optical) apparent magnitude or (radio) flux density may serve a8 a crude indicator of distance (i.e. depth). For uniformly and randomly distributed points, Poisson statistics is expected ~ the population of a bin being proportional to its size, and the fluctuation in the number equal to its square root. Such an analysis of the 2nd Bologna (B2) CURRENT SCIENCE, VOL, 65, NO. 11, 10 DECEMBER 1993 survey of (extragalactic) radio sourees down to 0.25 Jy at 408 MHiz shows that the number ef radio sources withina cube of size | Gpe or more varies by"< 3% as the cube is moved from place to place®, From the numbers of sources down to 0.25 Jy at 365 MHz (from Texas survey) in 11 randomly selected fields (avoiding any Galaetic Features) of sizes 18 to 2S deg’, the distri- bution of (extragalactic) radio sources on the celestial sphere is fully consistent with Poisson statistics CTable 1), Extragalactic radio sources are generally associated with faint distant optical galaxies. In contrast, the distribution of the brighter (and nearer) galaxies is Zwicky: amd Abell clusters). Somewhat deeper. but still shallower than radio surveys. the optical CEN survey shows clusters of galavies arranged in sheets clustered (2-8 enclosing voids, Data on these «and. other voids (Figure 2)" show more distant voids to be bigger (Figure 3 and Table 2)!" The sizes of voids range from 27 Mpe 16) 270 Mpe and the distances’ trom == 0.003 to 0.1. The enumeration of these voids is subjective. tind needs confirmation by an appropriate objective procedure. similar to an algorithm which broke up the istribution Of optical galaxies into groups and clusters!! > Power specirum anatvsis consists of resolving the sum of possibly weighted defunetions at the source positions on the celestial sphere into orthogonal sets of functions (Fourier series or spherical hurmonies), and using the sum of the squares of the amplitudes of the different components as relevant slatisti¢s. In practice, this method has given lower limits (generally = degrees) above which any clustering occurs. Although in principle it is possible to use this method for thrse- dimensional data, a convenient method of covariance fumetion analysis (sce below) is used for surveys with known redshifts, The results of applying power spectrum analysis to many radio surveys are summarized in Table 3 (ref. 5), The general conclusion is that tadio sources are distributed independently, uniformly at random on the celestial sphere on scales of 1 dearee and above, correspunding to more than a few 100 Mpe at estimated depths, 3. Covariance function analysiy (oF use of 1sv0-paint vrtelation funciions) gives the excess density of sources above that expected for a uniform random distribution at different linear or angular distances from each other, averaged over all the sources. Using to: dimensional surveys, this method ha (edi wen upper limits ) below which any clustering occurs. ‘This method and the previous, one can be considered a Fourier pair, If the 6o-polnt spatial correlation function has a power-law form with index 7. the corresponding jinaular finetion also has'a power-law form: with index 1p, Observations give y= 1.7 from both spacial and lar surveys up 60 10 Mpe amd 1? ang respectively SCTPNCE VOLCBRNO. EL. [0 DECEMBER 1993 REVIEW ARTICLES. Table 2. Other possime Felaisonshans foe te plot of ene 8 Tyne Equation Correa Power Dads LOG (RADIAL DISTANCE INK igure 5. Covecition beth volumes of Sil sit Chae desta Te carve is log I ihlpel) 0.208 exp 40.b6 Tow ims Sh th correlation = 0 658 Recent messurements ofthe angular gorrelatitn Function on seales from 0.01 slope 0.7 to 1? und show s gradually. sivepening decline 10 Zero at @ (re 2 Miree-poin and further correlation functions may also be used, but are not as well-determined a§ the wvo-point function. Ta point correlation functions are different types of Surveys coverin to 10° confirm the power-law of sho used far tv 4 the same region of 200) da not show pemselves on Seales oF a few dees sky. Thus. the 3 radio sources ( clustering among 1 but 3C radio galaxies occur preferentially 0 bright galaxy clusters (as seen in the. optical bund), while 3€ quasars (whieh are an order of ma distant) show no such preference® 4. Disteibation functions are af two types. One, denored by f(A), measures the probability of finding different numbers of objects in a given area of sky oF a given Volume of space specified by R. This way first applicd systematically to galaxies by Hubble. who used samples with large values of V. The other complementary distribution funetion fy (/&) measures the probability of a given number V of abjects acewpying different ar sky or volumes of space given by R bach may be computed for angular sind spatial surveys with a specific procedure, and compared with the two-parameter prediction for relaxed gravitational clustering, In this article, I have avoided any discussion of models. but make an exception for this simple ankl powerful model REVIEW ARTICLES spectrin analysis applied t» many zadio surveys Feequeney Hus density No.of beams Survey iGita, Time (iy) per source Rests and eemarhs w 0178 anu a ‘No clustering on alt seates = 4 Shght anvelustering ae 1a as 0.09 9 Novelusteving on all veaey = 4 Shah annclustering 2 90 Weak clustering un seales> 10,7 wih 17 sources/eioaer Mer odos. 690 a 330 No elustening on all ales = 1 oo 90 ‘No clustering om al seales = 18 PRS 3s 1560 No clustering on all scales = 5 — 27 22700 _—_ aa PRs 028 1190 No clustenng on all sales © 8 2 ons «770 02 Pl No clustering on all seales = 0:7 Slight anticlusterins 4 0408 oo 40 Nu clustering on all scales = Wa One model parameter is the average number of objects inthe given area or volume. and the other, denoted by h essentially equals half the ratio of the gravitational cor- ‘elation energy to the kinetic energy of peculiar mation. For no clustering, Le, uniform random (= Poisson) distribution, #0. and 6 = 1 ifall objects are clustered. Application of this method to Zwicky and Abell surveys of clusters, and to. a small deep |.4 GHz, radio. survey (207 sources down to Smly ina 4° x-4° area) gives N(Awicky)~ 0.704 0.05 on seales 1 to 10 Mpe, (Abell) — 0.29 £0.08 (10 to SOMpe} and Airadio) 0.01 £0.01 & 100 Mpe)"®. Thus gravitational clustering decreases with scale. diseppearing compleiely beyond 100 Mpe 5 Fractal geometry! may be applicable on Seales’ from a few Mpc to a few 100Mpe. de Va hierarchical structure in the distribution of galaxies yor in pairs or multiplets (0.1 Mpe) forming Foups and small clusters (few Mpe) as welll as rich clusters. (fens ef Mpe) and all these forming superclusters. (few 100 Mpe). to the depth of Abell survey!” Rich clusters have up to a thousand galaxies, although most galaxies are in smaller groups. The density of visible matter against the scale on which it is determined isa powertaw of slope —1,7 (ref. 16) consistent with the hierarchical (or fractal) structure, and also with the slope of the two-point correlation function of the distribution of gatasies (see (3) above) The fractal dimension of the structure is thus, 3-1.7= 1.3. Deeper. especially radio, surveys have shown that the higrarehy probably does not continue, radio sources being distributed uniformly at random (see couleurs found 80 (2) and (4) above). But since actual distances to radio sources (ie. their redshifis) are largely not yet deter mined, it is well to Keep an open mind In the method of analysis using the covariance function or the two-point (or pair) correlation function (G) above) there is an implicit assumption of uniformity (or homogeneity) 09 a scale within the survey extemt, An alternative equivalent method without this assumption is possible, and necessary if fractal structure persists to the largest seale accessible to a survey. This method uses the conditional number density instead of the covariance function, To bring out clearly the difference and the relation- ship between these 1wo methods. | give details. For a spatial survey over volume V having V objceis of equal weights at positions r,./= 1, ....N, the number density n(r)=E8\8(r—r), Defining average by p= de’, where dr’ denotes an element of volume, th cross-correlation function of (r) is ie Layer) Ges (ace ynteeey), = $35 (r-( 1) Satn-= The two-point (er pair) correlation funetion Er) defined by THE Cr} lene +e (a)? where, clearly, (n) = N/V. So 198 = GO? = Ene ry CURRENT SCIENCE, VOL. 63, NO. 11.10 DECEMBER 1993 REVIEW ARTICLES: Thus E(r) involves the survey size V7. (Note that the cross-correlation function G(r) also involves 1.) Consider now the conditional density Per) Sdn(r?)n(e" EY) y (nd This function is calculated to be i re 2 riey-L Ying r= Yay) & yen © ing Ph 2 Speen 1 ya and does not involve the survey volume , Note that Sr) and I(r) are related by Er) =F) dnp Pe 8000 km/s Tery=(oou+g ey. -20.0 + Slog h Assuming that [(e) depends only on the magnitude + of r. and denoting it by T(r), an average conditional ; ‘ ts oglr (hb Mpe! density I'+(r) is defined as at pe] 4% gue as'Fne age ante (satus, ee 206 genio res dean ere’) EiRisuvey nth ex 00 kal saa 475-200" § lo janet a The results of analysing a volume-timited sample of CIA survey using E(), P(r) and 7%") are shown in 2h & re (re) = 0 4| Figure 4 (ref 17), Although £ (7) drops sharply, I(r) and '*(7) continue scale which can be probed using the sample. As seen for appropriately simulated samples, P(r) and Fr) would level at the seale at which homogeneity (or uniformity) sets in, As ABE £ (7), thete has been a tendency to derive ecreasing smoothly to the largest Ero (ro) = 1 ‘characteristic seales’ ry and rj from E(r)=!andE(rj) Sg | = 0, although a power-law variation implies a scale-fee é behaviour. If the derived scales 7, and 1 are really S characteristic of the distribution, they should be indepedent of the sample size. However. for volume- 6h limited samples of CfA survey ry andr depend on the sample size (Figure 5)!” (ref. 17) precisely in the way expected if there is no characteristic scale up to the largest sample size. For E(r) a power-law, say Ar-?, rp defined by E (ra) = | equals 4" so that & (7) = (rina)? This latter way of writing a power-law gives a false impression of there being a characteristic scale rp glevurnueeeesiags | ‘An analysis of the eluster-cluster correlation in Abell act survey using conditional densities shows that there isno Ra (h” Mpc) 7 88 1084 # uniformity to the largest accessible scale (Figure 6)!’..A Figure 8. ry defined by E (rg)= 1 andr bY a rr) 45 + 2 eo P94 T(r) 90 %eet, hb = ass ae UL 1 1 1 6 Y 16 = log(r) [log(h’ Mpe)} Figure 6. Fev and * (oy fora sample of Abel surve} certainly deeper, but whose exact depth is not known ue to lack of redshift measurements, indicate that the Jange-scale structure is uniform beyond a few 100 Mpe Visual impressions. from maps’ have shown superclusters, voids, filaments and sheets in the distri- bution of galaxies on scales of a few tens 10 few 100 Mpe. Fractal geometry may provide a uood description of these structures 6 Graph theory deals with graphs, which are sets of points, some of them connected pairwise. Graphs with out loops are called trees, All the points of a set may be interconnected in a single tree in many different ways Fach gives a certain length of the spanning tree. spanning tree of the minimum leagth iy mininwl spanning tree. The visual impression of Filaments In a set of points can be quantified using minimal spannine irees of its subsets, pruning them, and rearran: random manner to see that filamenis disappeis. Thy filaments have been shown to ‘objectively’ exist in galaxy surveys". 7 Topological studies are relevant since galasies ay embers of (possibly weighted) point sets imply an underlying smoother mass.density, certainly at the epoch fof structure formation, pethaps before redshift 5 (ie == 5). Isodensity surfaces of smoothed point sets with Jifferent smoothing scales and density levels (or thres hold densities) may be profitably studied and compared ‘with predictions of structure formation models, includ= ing the simplest possibility. viz. Gaussian (or random Pe Enc eLUsTeR COM haan wate STR CHEE py any ap sinha ATTMACTOR / SUTERCLIISTIRS © REAM SUREEYS pe recite tt rte $A 4 81 My NEW ANALYSIS (WITHOUT ASSUMPTIONS ) >? > $$ iti a —— ihe 25 Mpe 50-10) Mpe 150-300 Mpc (EG Littisesay cummins 0000 Mp R SMpc 25Mpe 50-100Mpc 180-300 Mpc 3 - S000 Mpc CURRENT SCIENCE, VOL. 65, NO. 11.10 DECEMBER 1993 IEW ARTICLES Abell Clesters a ° ey ak ae Vou = 22600 t/a Se 1 = 6.000 hae mt ab 4 001 ‘rally = + bells af ja # 2 § Sips of 43 448) MH 1-6 F 5 sere ceEacl 1 ! 1 ist ! ok x a # 100 1000 10000 A (cm/s) igure. g,(0) , Gr “sample from Abell_survey with igure 10, Amplitude of, (¥) for dtterert smoothing Kengths wed ¥ 600 km 3"! and smoothing seale 3000 km s-'* in deriving x; (¥}, Composite Date: CH. CIA, TS, and Tull ee re 1 1 7-73 for a composite of four samples in Figure 9 (ref. 19) A= 600 km/s The curves correspond t g, o¢ (I~ v2) exp(-v7/2) for 20 =] the Gaussian (i. random phase) distribution, Within sf v/< 1, the genus per unit volume is positive, indicating VIN a sponge topolagy. For v<—1, there are voids or I h | bubbles. while for p> +1, there are isolated clusters. \ Isolated voids surrounded by matter and isolated clumps of matter in empty space are the only two types of topology possible in two dimensions. In three dimens- —{ ions, one can, in addition, have sponge topology, ie. a connected matter region interlocking, with a connected as) void region. A two-dimensional slice of a (three-dimen- sional) sponge shows isolated voids or isolated clumps The amplitude of the g,() curve for galaxies decreases with increasing smoothing scales (Figure 10)", that is, the structure becomes less choppy (or smoother) as we average over larger volumes, in agreement with v (examples of) results cited in sections (2) and (4) above, genus igure 9, (0) for composite sample 8. Percolation study? of a point set is done by constructing a sphere of radius r around each point of phase) distribution. The genus of the isodensity surface the set, and increasing F, so that the sizes of connected (i.e. the number of holes it has) at a given density level groups of spheres increase, while, at the same time, the (or threshold) may be calculated as an integral over the qumber of groups decreases, AL the (critical) percolation susface of ity Gaussian curvature (by Gauss-Bonnet radius r, all spheres connect in a single group. Kor theorem). Superposing a grid of small cubes through the galaxies it is found that the observed distribution has, surface, one may, in practice, extract a topologically —r,

    You might also like