On the relation between Lya absorbers and local galaxy filaments
AAstronomy & Astrophysics manuscript no. Lya_in_fil © ESO 2021February 4, 2021
On the relation between Ly α absorbers and local galaxy filaments S. J. D. Bouma, P. Richter, and M. Wendt
Institut für Physik und Astronomie, Universität Potsdam, Karl-Liebknecht-Str. 24 /
25, 14476 Golm, Germanye-mail: [email protected]
Received XXX; accepted YYY
ABSTRACT
Context.
The intergalactic medium (IGM) is believed to contain the majority of baryons in the universe and to trace the same darkmatter structure as galaxies, forming filaments and sheets. Ly α absorbers, which sample the neutral component of the IGM, have beenextensively studied at low and high redshift, but the exact relation between Ly α absorption, galaxies and the large-scale structure isobservationally not well-constrained. Aims.
In this study, we aim at characterising the relation between Ly α absorbers and nearby overdense cosmological structures (galaxyfilaments) at recession velocities ∆ v ≤ − by using archival observational data from various instruments. Methods.
We analyse 587 intervening Ly α absorbers in the spectra of 302 extragalactic background sources obtained with the CosmicOrigins Spectrograph (COS) installed on the Hubble Space Telescope (HST). We combine the absorption-line information with galaxydata of five local galaxy filaments originally mapped by Courtois et al. (2013).
Results.
Along the 91 sightlines that pass close to a filament, we identify 215 (227) Ly α absorption systems (components). Amongthese, 74 Ly α systems are aligned in position and velocity with the galaxy filaments, indicating that these absorbers and the galaxiestrace the same large-scale structure. The filament-aligned Ly α absorbers have a ∼
90 percent higher rate of incidence ( d N / dz = N (H i ) ≥ .
2) and a mildly shallower slope ( − β = − .
47) of the column density distribution function than the general Ly α population at z =
0, reflecting the filaments’ matter overdensity. The strongest Ly α absorbers are preferentially found near galaxiesor close to the axis of a filament, although there is substantial scatter in this relation. Our sample of absorbers clusters more stronglyaround filament axes than a randomly distributed sample would do (as confirmed by a Kolmogorov-Smirnov test), but the clusteringsignal is less pronounced than for the galaxies in the filaments. Key words.
Galaxies: halos – intergalactic medium – quasars: absorption lines – Cosmology: large-scale structure of Universe –techniques: spectroscopic — ultraviolet: general
1. Introduction
It is generally accepted that the intergalactic medium (IGM) con-tains the majority of the baryons in the universe (e.g. Shull 2003;Lehner et al. 2007; Danforth & Shull 2008; Richter et al. 2008;Shull et al. 2012; Danforth et al. 2016), making it a key compo-nent in understanding cosmological structure formation. It is es-timated that about 30 % of the baryons at low- z are in the form ofphotoionised hydrogen at a temperature of (cid:46) K (Penton et al.2000; Lehner et al. 2007; Danforth & Shull 2008; Shull et al.2012; Tilton et al. 2012; Danforth et al. 2016), while the col-lapsed, shock-heated Warm-Hot Intergalactic Medium (WHIM)at T ∼ − K contains at least 20 % (Richter et al. 2006a,b;Lehner et al. 2007). Cosmological simulations indicate that theWHIM may contribute up to 50 % of the baryons in the low- z universe (Cen & Ostriker 1999; Davé et al. 1999; Shull et al.2012; Martizzi et al. 2019). This makes the IGM an importantreservoir of baryons for the galaxies to fuel star formation. In-deed, to explain current rates of star formation, such a baryonreservoir is needed (e.g. Erb 2008; Prochaska & Wolfe 2009;Genzel et al. 2010). In this way, the evolution of galaxies is tiedto the properties and the spatial distribution of the IGM.The relation between the IGM and galaxies is not one-way,however, as galaxies influence their surroundings by ejectinghot gas into their circumgalactic medium (CGM) by AGN feed-back (e.g. Bower et al. 2006; Davies et al. 2019) and, partic-ularly in the early universe, by supernova explosions (Madau et al. 2001; Pallottini et al. 2014). Thus, there is a large-scaleexchange of matter and energy between the galaxies and the sur-rounding IGM and both environments influence the evolution ofeach other. One observational method to study the gas circula-tion processes between galaxies and the IGM is the analysis ofintervening Lyman α (Ly α ) absorption in the spectra of distantActive Galactic Nuclei (AGN), which are believed to trace thelarge-scale gaseous environment of galaxies. Generally, an an-ticorrelation between Ly α absorber strength and galaxy impactparameter is found for absorbers relatively close to galaxies (e.g.Chen et al. 2001; Bowen et al. 2002; Wakker & Savage 2009;French & Wakker 2017).The IGM is not only tied to the galaxies, but it is also ex-pected to trace the dark matter distribution and can, therefore,give insights into the large-scale structure of the universe. Thislarge-scale structure has been mapped by galaxy surveys like the2-degree Field Galaxy Redshift Survey and the Sloan Digital SkySurvey (SDSS) (Colless et al. 2001; York et al. 2000). Study-ing the IGM absorber distribution at low redshift allows for acomparison with data from these galaxy surveys. Already morethan 25 years ago, Morris et al. (1993) studied the spectrum ofthe bright quasar 3C 273 and mapped Ly α absorbers along thisline of sight together with galaxies in its vicinity. They foundthat Ly α absorbers cluster less strongly around galaxies than thegalaxies among themselves. This can be interpreted as most ofthe Ly α absorbers truly being intergalactic in nature, following Article number, page 1 of 16 a r X i v : . [ a s t r o - ph . C O ] F e b & A proofs: manuscript no. Lya_in_fil the filamentary large-scale structure rather than the position ofindividual galaxies.More recently, Tejos et al. (2016) studied Ly α and O vi ab-sorption in a single sightline in regions between galaxy clusters.The detected overdensity of narrow and broad Ly α absorbershints at the presence of filamentary gas connecting the clusters.A di ff erent approach was taken by Wakker et al. (2015). Insteadof mapping gas along an isolated sightline, they used severalsightlines passing through a known galaxy filament. By com-paring the relation of Ly α equivalent width with both galaxy andfilament impact parameters, Wakker et al. (2015) conclude thatLy α absorbers are best described in the context of large-scalestructure, instead of tracing individual galaxy haloes. Whilethere is a relation between strong ( N (H i ) > cm − ) absorbersand the CGM of galaxies, weak Ly α absorbers are more likelyto be associated with filaments. This view is also supported byPenton et al. (2002), who find that weak absorbers do not showa correlation between equivalent width and impact parameter tothe nearest galaxy, while stronger absorbers do. By comparingthe position of their sample of Ly α absorbers relative to galaxiesin filaments, they conclude that the absorbers align with the fil-amentary structure. Evidence for absorbers tracing an extensive,intra-group medium comes from other recent surveys of Stockeet al. (2013) and Keeney et al. (2018).While the correlation between Ly α equivalent width andgalaxy impact parameter seems to indicate that these absorbers somehow are associated with galaxies (e.g., by the gravitationalpotential), studies like Wakker et al. (2015); Tejos et al. (2016)show that at least some of the absorbers are associated with thecosmological large-scale structure. Others studies (Bowen et al.2002; Wakker & Savage 2009) conclude that their data simplydoes not yield any definite conclusions on this aspect (see alsoPenton et al. 2002; Prochaska et al. 2011; Tejos et al. 2014).Therefore, the question of how Ly α absorbers at z = / galaxy connectionare desired.In this paper, we systematically investigate the properties of z = α absorbers and their connection to the local galaxyenvironments and the surrounding large-scale structure. For this,we follow an approach similar to that of Wakker et al. (2015).We combine the information on local galaxy filaments mappedby Courtois et al. (2013) with archival UV absorption line datafrom the Cosmic Origins Spectrograph (COS) installed on the Hubble Space Telescope ( HST ).Information on the galaxy sample used in this study is pro-vided in Sect. 2. In Sect. 3, the HST / COS data are describedand information on the absorption line measurements are given.Details on the galaxy filaments are presented in Sect. 4. InSect. 5, we investigate the relation between absorbers and galax-ies, whereas in Sect. 6 we focus on the relation between ab-sorbers and filaments. In Sect. 7, we discuss our findings andcompare them with previous studies. Finally, we summarise andconclude our study in Sect. 8.
2. Galaxy data
Courtois et al. (2013) used the V8k catalogue of galaxies to mapgalaxy filaments in the nearby universe. This catalogue is avail-able from the Extragalactic Distance Database (EDD Tully et al.2009). It is a compilation of di ff erent surveys, including John http://edd.ifa.hawaii.edu/ Huchra’s ‘ZCAT’ and the IRAS Point Source Catalog redshiftsurvey with its extensions to the Galactic plane (Saunders et al.2000a,b). In total, the catalogue consists of ∼
30 000 galaxies,all with velocities less than 8000 km s − . It is complete up to M B = −
16 for galaxies at 1000 km s − , while at 8000 km s − ,it contains one in 13 of the M B = −
16 galaxies. A radial ve-locity of 8000 km s − corresponds to a cosmological distanceof d ∼ (114 km s − ) h − . The distance to the Centaurus Cluster( v ∼ − ) is ∼
40 Mpc. As described in Sect. 3, the ve-locity range studied in this work extends up to v ∼ − ,which corresponds to λ ∼ − in the V8k catalogue are ad-justed to match the Virgo-flow model by Shaya et al. (1995). Therelatively uniform sky coverage (except for the zone of avoid-ance, ZOA) of the V8k survey combined with the broad range ofgalaxy types make it suitable for qualitative work (Courtois et al.2013).The distribution of apparent and absolute B -band magnitudesas well as log( L / L ∗ ) for all galaxies of the V8k catalogue is pre-sented in Fig. 1. As can be seen from this distribution, the V8kcatalogue is largely insensitive to dwarf galaxies with luminosi-ties log( L / L ∗ ) ≤ − .
5. This needs to be kept in mind for ourlater discussion of the absorber-galaxy relation in Sects. 5 and6. We decided to not add supplementary galaxy data from othersurveys, because the sky coverage of such a mixed galaxy sam-ple would be quite inhomogeneous, which would introduce anadditional bias to the galaxy-absorber statistics.In Fig. 2, upper panel, we show the sky distribution of thegalaxies in the various filaments, such as defined in Courtoiset al. (2013). The galaxies in these filaments have radial veloci- B N u m b e r AllFilaments B N u m b e r ∗ )10 N u m b e r Fig. 1.
Histogram of apparent and absolute B -band magnitudes andluminosities for all galaxies of the V8k catalogue.Article number, page 2 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments ties in the range v = − − . All filaments are feedinginto the Centaurus Cluster located at l ∼ ◦ and b ∼ ◦ . Thelarge concentration of galaxies in the green filament, between l ∼ − ◦ and b ∼ − ◦ is due to the Virgo Cluster.
3. Absorption line data
In this study we make use of ancillary HST / COS data, as re-trieved from the HST Science Archive at the Canadian Astron-omy Data Centre (CADC). The total sample consists of 302AGN sightlines, all reduced following the procedures describedin Richter et al. (2017).Since the Ly α absorption ( λ = .
67 Å) in the spectrastudied here falls in the wavelength range between 1215 and1243 Å, we make use of the data from the COS G130M grating.This grating covers a wavelength range from 1150 − R = , − ,
000 (Green et al. 2012;Dashtamirova et al. 2018). The data quality in our COS sampleis quite diverse, with signal-to-noise (S / N) ratios per resolutionelement varying substantially (between 3 and 130; see Fig. A.1in the Appendix.)We also checked for metal absorption in the Ly α ab-sorbers, considering the transitions of Si iii λ .
50, Si ii doublet λ .
42; 1193 .
29, Si ii λ .
42, Si ii λ . iv doublet λ .
76; 1402 .
77, C ii λ .
53, C iv doublet λ .
20; 1550 .
77. For the lines at λ > λ = − For all 302 COS spectra, the wavelength range between 1220 − α in the velocity range v ≈ − − .At velocities < − , Ly α absorption typically is stronglyblended with the damped Ly α absorption trough from the fore-ground Galactic interstellar medium (ISM). To ensure consis-tency, we do not further consider any absorption feature below1220 Å.Each detected absorption feature at 1220 − α absorption by ruling out Galactic foregroundISM absorption and other, red-shifted lines from interveningabsorbers at higher redshift. As for the Galactic ISM absorp-tion, this wavelength range contains only the N v doublet (1238,1242 Å) and the weak Mg ii doublet (1239, 1240 Å) as potentialcontaminants and the regions were flagged accordingly. Poten-tial red-shifted contaminating lines that were ruled out include: the H i Lyman-series up to Ly δ , Si iii (1206.50 Å), and the twoO vi lines at 1037 .
62 and 1031 .
93 Å. Whenever possible, we alsoused the line-list of intergalactic absorbers from Danforth et al.(2016), which covers a sub-sample of 82 COS spectra. All in all,we identify 587 intervening Ly α absorbers along the 302 COSsightlines in the range λ = − span code (Richter et al. 2011) in theESO-MIDAS software package, which also provides veloci-ties / redshifts for the absorbers. To derive column densities of H i (and the metal ions) for a sub-sample of the identified Ly α ab-sorbers we used the component-modelling method, as describedin Richter et al. (2013). In this method, the various velocity sub-components in an absorber are consistently modelled in all avail-able ions (H i and metals) to obtain column densities ( N ) andDoppler-parameter ( b -values) for each ion in each component.Throughout the paper, we give column densities in units [cm − ]and b -values in units [km s − ]. The modelling code, that is alsoimplemented in ESO-MIDAS, takes into account the wavelengthdependent line-spread function of the COS instrument. Wave-lengths and oscillator strengths of the analysed ion transitionswere taken from the list of Morton (2003).The total sample of 302 COS sightlines was separated intotwo sub-samples, one with sightlines passing close to a filament,and the other with sightlines that do not. To account for the oc-casionally seen large projected widths of the filaments (see, e.g.,part of the dark blue filament in Fig. 2) and to be able to mapalso the outer parts of the filaments, a separation of 5 Mpc to thenearest galaxy belonging to a filament was chosen as dividingdistance in this selection process. One sightline (towards 4C–01.61) was categorised as belonging to a filament – although itsnearest galaxy distance is as large as 7.9 Mpc – because it passesa filament that is very poorly populated. In total, our selectionprocesses lead to 91 sightlines that are categorised as filament-related, while the remaining 211 sightlines are categorised assightlines that are unrelated to the filaments studied here. Thetotal redshift pathlength in our COS data set can be estimatedas ∆ z = . N , with N being the number of sightlines. Thisgives ∆ z = .
72 and 3 .
99 for the sightline sample belonging tofilaments and the one unrelated to filaments, respectively. Thiswill be further discussed in Sect. 7.Within the for us relevant sub-sample of the filament-related sightlines, 12 spectra were unsuited for measurementsfor absorption-line measurements due to various di ff erent dataissues, such as an indeterminable continuum, or heavy blendingfrom various lines. Of the remaining 79 spectra, 9 had no Ly α ab-sorption features detected in the studied wavelength range. Thisimplies a Ly α detection rate of ∼
90 % (we will later further dis-cuss the number density and cross section of Ly α absorbers inthis sample). The signal-to-noise ratios for these 79 spectra varybetween 5 and 92 per resolution element. In this sub-sampleof 79 filament-related sightlines, we identify 215 Ly α absorp-tion systems that are composed of 227 individual components.For these 215 (227) absorbers (components), we have derivedH i column densities and b -values via the component-modellingmethod, as described above.In the other sightline sample, that we categorise as unrelatedto the galaxy filaments, 25 spectra were unsuited for measure-ments for the same reasons as described above. Of the remaining186 spectra, only 24 show no Ly α absorption in the range con-sidered above, resulting in a 87 % detection rate for Ly α in thissample. Article number, page 3 of 16 & A proofs: manuscript no. Lya_in_fil b ( ◦ ) VirgoCentaurusFornaxNorma l ( ◦ )9060300306090 b ( ◦ ) v fl ( k m s − ) Fig. 2.
Upper panel:
Sky distribution of galaxies from V8k belonging to filaments as defined in Courtois et al. (2013). The di ff erent coloursindicate di ff erent galaxy filaments. Several important clusters are noted. Lower panel:
Sky distribution of HST / COS sightlines passing close toa filament (black circles) and HST / COS sightlines not belonging to a filament (grey circles) plotted together with the galaxies from the V8kcatalogue belonging to filaments (colour-coded according to velocity).
Metal ions (Si ii , Si iii , Si iv , C ii or C iv ) were detected for26 of the 215 Ly α filament absorbers, giving a metal detectionfraction of ∼
12 %. Two example HST / COS spectra are shownin Fig. 3 (black) together with synthetic model spectrum (red).These example spectra give an indication of the characteristicdi ff erences in S / N in the COS data used in this study.Figure 4, upper panel, shows the distribution of H i Ly α equivalent widths for the detected absorbers in the two sub-samples and in the combined, total sample. The lower panelinstead shows the distribution of H i column densities in thefilament-related absorbers, as derived from the component mod-elling. Both distributions mimic those seen in previous Ly α stud-ies at z = α absorbers obtained with HST / COS showsa similar distribution with a peak in equivalent width just be-low 100 mÅ. The H i column-density distribution falls o ff belowlog N (H i ) = . i Ly α absorbers. Note that because of the limitedspectral resolution and S / N many of the broader Ly α lines mostlikely are composed of individual, unresolved sub-components.The H i column-density distribution function will be discussedin Sect. 7. λ ( Å )0.00.51.01.5 N o r m a li s e d f l u x Fig. 3.
HST / COS G130M spectra of the QSOs VV2006-J131545.2 + α absorbers are seen in these spectra. For a better visualisa-tion, both spectra are binned over two pixels.Article number, page 4 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments λ (m Å )0.00.51.01.52.02.5 F r e q u e n c y
1e 1
OutsideInside N ( HI )01234 F r e q u e n c y
1e 1
Fig. 4.
Histogram of equivalent widths of Ly α absorbers (upper panel)and log N (H i ) in the filament-related absorbers, as derived from the com-ponent modelling (lower panel). Errors of the measured equivalent widths have been derivedwith the span code (Richter et al. 2011), which takes into ac-count the S / N around each line, the uncertainty for the localcontinuum placement, and possible blending e ff ects with otherlines / features. Typical 1 σ errors for the equivalent widths liearound 20 mÅ. The errors in the column densities were derivedbased on the component-modelling method Richter et al. (2011).Here, the typical errors are on the order of ∼ . b . With the majority of b -values falling be-tween 10 and 30 km s − , the errors are typically ∼ − , withlower errors for the low end of the range of b and slightly highererrors for larger b -values. Tabulated results from our absorption-line measurements can be made available on request.
4. Characterisation of galaxy filaments
To study how the IGM is connected to its cosmological envi-ronment, it is important to characterise the geometry of the fil-aments, their galaxy content, and their connection to the overalllarge scale structure. In Fig. 5, we show the position of the galax-ies in the filaments together with their radial extent in 1.5 virialradii (1.5 R vir ). Gas within this characteristic ‘sphere of influ-ence’ can be considered as gravitationally bound to that galaxy.This plot therefore gives a first indication of how much uncov-ered sky there is between the galaxies and their spheres of in-fluence, indicative for the projected intergalactic space in the fil-aments (compared to the projected circumgalactic space within1.5 R vir ). The Virgo Cluster clearly stands out, as many galax-ies are overlapping in their projected spheres at 1.5 R vir , whilein most other filaments, there are both regions with strong over-lap and regions without overlapping halos. In Sect. 6 and in theAppendix, we will discuss also other virial radii as selection cri-teria.
To define an axis for each filament, a rectangular box was gener-ated per filament containing the galaxies therein. The dark bluefilament (see Fig. 2) was split into two individual boxes becauseof geometrical reasons. Widths and lengths of the boxes vary forthe di ff erent filaments, as they scale with the filament’s projecteddimensions.After defining the boxes sampling the individual filaments,they were each sub-divided into segments with the full widthof the box and a length corresponding to 20 ◦ on the sky. Eachsegment overlaps with the previous one with half the area (10 ◦ length). The average longitude and latitude of the galaxies withineach segment was then determined and used as an anchor pointto define the filament axis. All these anchor points were con-nected in each filament to form its axis.In this way, the definition of the filament axis on the skyallowed us to calculate impact parameters of the COS sightlinesto the filaments. In addition, we calculated velocity gradients inthe filaments, by taking the average velocity of all galaxies ineach segment as velocity anchor point.The method of using overlapping segments to determine thefilament axis is similar to the approach used by Wakker et al.(2015). A di ff erence with their approach is that they first deter-mined which galaxies were part of the filament by looking at thevelocities. We did not do this as the filaments were already de-fined by Courtois et al. (2013). The uncertainty on the placementof the filament axes is no more than 1.5 ◦ on the sky, less for mostfilaments.The characteristics of each filament will be discussed sepa-rately in the following subsections. The orange or ‘4 clusters’ fil-ament from Courtois et al. (2013) is not discussed here, as thereare no available COS sightlines nearby. Perhaps the most notable of the filaments discussed here isthe one containing the Virgo Cluster, located at a distance of ∼ l ∼ − ◦ and b ∼ − ◦ . The Virgo area has the highestgalaxy density of the regions studied here.The axis of the green filament as well as the galaxy ve-locities are indicated in Fig. 6a. The velocities range from ∼ − at the Centaurus Cluster to ∼ −
400 km s − How-ever, the velocities of the galaxies in the Virgo Cluster reach upto ∼ − , indicating a large spread in velocities, just asexpected for a massive galaxy cluster.Figure 7 shows that the density along the filament variesgreatly, with the Virgo Cluster being the densest region (sub-boxes 6 − As mentioned in Courtois et al. (2013), the purple filament isthe longest cosmological structure in space from those studiedhere. In projection, however, it is one of the shorter filaments onthe sky. This filament was discussed in detail by Fairall et al.(1998), who named it the ‘Centaurus Wall’. A striking lack ofgalaxies in the regions around b ∼ ◦ is evident in Fig. 6b dueto the ZOA caused by the Milky Way disk in the foreground. Article number, page 5 of 16 & A proofs: manuscript no. Lya_in_fil l ( ◦ )9060300306090 b ( ◦ ) Fig. 5.
Galaxies belonging to all filaments considered in this study plotted together with their projected 1.5 virial radii.
Just below this scarcely populated region is the Norma Cluster( l ∼ ◦ ), followed by the Pavus II Cluster ( l ∼ ◦ ).The purple filament contains the galaxies with the highestvelocities in our sample, with v reaching up to 6500 km s − (seeFig. 6b). These high velocities indicate distances of ≤
85 Mpc.It is the only filament in which the galaxy velocities stronglyincrease when moving away from Centaurus. As such, it extendsbeyond the velocity range considered in Courtois et al. (2013).Here, we consider only the part of the filament indicated by theirwork.The purple filament is the densest of our defined filaments,which is not surprising as it hosts two galaxy clusters and theprojection e ff ect makes it visually compact on the sky. A total of351 galaxies from the V8k catalogue belong to this filament, butonly 2 COS sightlines, which are both shared with the dark bluefilament. The dark blue filament represents one branch of the SouthernSupercluster filament, defined in Courtois et al. (2013). Since itis clearly separated on the sky from the other branch (the cyanfilament), these two branches are treated as individual filamentsin this study. Starting from the Centaurus Cluster, the dark bluefilament is entangled with the purple filament, but it continuesto stretch out as a rather di ff use cosmological structure over therange l ∼ − ◦ in the southern hemisphere. Because of thelow galaxy density, the filament axis of the dark blue filamentis not well defined and unsteady compared to other filaments, ascan be seen in Fig. 6c. The dashed portion of the axis indicatedin the figure is a result of the small number of galaxies foundin this region, so the exact filament geometry in this part of thefilament remains uncertain.Figure 6 further indicates that average velocities in the darkblue filament are much lower than in the purple filament, mak-ing the two filaments easy to distinguish. The dark blue filamentalso exhibits two distinct velocity branches: one with veloci- ties ∼ − and one with v ∼ − (see Fig. 6),further underlining the inhomogeneous morphology of this fila-ment. This filament has only 180 galaxies and 21 COS sightlines. The second branch of the Southern Supercluster filament is in-dicated by the cyan colour in Fig. 2. Compared to the dark bluefilament, this branch is rather densely populated and the corre-sponding filament axis is well defined (Fig. 6d).As with the green and dark blue filaments, the highest veloc-ities in the cyan filament are found near the Centaurus Cluster,with velocities decreasing as one gets closer to the Fornax Clus-ter. However, Fig. 6 suggests that there is a slight increase invelocity near the end of the filament at l < ◦ .The cyan filament is made up of 289 V8k galaxies and thereare 20 COS sightlines passing though it. This filament (magenta coloured in Fig. 2) contains the AntliaCluster and also crosses the ZOA. While it is densely populatedfor b > ◦ (near Centaurus), it is underdense near the ZOA andalso only moderately populated at negative Galactic latitudes.This makes the transition of the filament axis from positive tonegative latitudes hard to define.As can be seen in Fig. 6e, the velocities in this filament rangefrom 3000 km s − near the Centaurus Cluster to 1400 km s − nearits end at l = ◦ and b = − ◦ . It has 143 galaxies and 2 usableCOS sightlines.
5. Ly α absorption and its connection to galaxies To learn about the relation between intervening Ly α absorption,nearby galaxies, and the local large scale structure, in which theabsorbers and galaxies are embedded, we first look at the con- Article number, page 6 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments b ( ◦ ) (a) (d) b ( ◦ ) (b) l ( ◦ )6045301501530 (e) l ( ◦ )907560453015015 b ( ◦ ) (c) 01000200030004000300040005000600001000200030004000 v fl ( k m s − ) v fl ( k m s − ) v fl ( k m s − ) Fig. 6.
Galaxies belonging to di ff erent filaments [(a): green; (b): purple; (c): dark blue; (d): cyan; (e): magenta] with their velocities colour-coded.Grey dots show galaxies belonging to one of the other filaments. The filament axes is indicated with the black solid line. nection between Ly α absorption in the COS data and individualgalaxies.In Fig. 8, upper panel, we have plotted the equivalent widthsof all Ly α absorbers against the line-of-sight impact parame-ter to the nearest galaxy, ρ gal , that has a radial velocity within400 km s − of the absorber. For this plot, all V8k galaxies havebeen taken into account (not just the ones in filaments), as someof the absorbers might be related to galaxies outside of the maincosmological structures. We indicate absorbers that are within1000 km s − of the nearest filament in red, and those that havelarger deviation velocities in blue. Non-detections have been in-dicated by the black crosses. The corresponding sightlines do notshow any Ly α absorption in the wavelength range 1220 − W λ <
200 mÅ. This overabundance of weak absorbersclose to galaxies might be a selection e ff ect. Prominent regions,such as the dense Virgo Cluster, receive more attention by re-searchers and are sampled by more sightlines (and by spec- tral data with better S / N) compared to underdense cosmologi-cal regions, which typically are not as well-mapped. The highestequivalent widths of the absorbers ( W λ >
500 mÅ) typically arefound closer to the galaxies, in line with the often observed anti-correlation between Ly α equivalent width and impact parameter(e.g. Chen et al. 2001; French & Wakker 2017). There is, how-ever, a large scatter in this distribution, such as seen also in otherstudies (e.g. French & Wakker 2017). This scatter most likelyis related to filament regions that have a large galaxy densityand overlapping (projected) galaxy halos, such as indicated inFig. 5. Ly α absorption that is detected along a line of sight pass-ing through such a crowded region cannot unambiguously be re-lated to a particular galaxy (such as the nearest galaxy, which isassumed here), but could be associated with the same likelihoodto any other (e.g., more distant) galaxy and its extended gaseoushalo that is sampled by the sightline.The lower panel of Fig. 8 shows the Ly α equivalent widthplotted against ρ gal / R vir . Again, we see the same trend forstronger absorbers to be closer to a galaxy. Out of the 208 Ly α absorption components, 29 are within 1 . Article number, page 7 of 16 & A proofs: manuscript no. Lya_in_fil G a l a xy d e n s i t y ( N r / ◦ ) Fig. 7.
Galaxy density along the green filament. The galaxy densityindicates the number of galaxies within 5 ◦ on the sky for each galaxy. ρ gal (Mpc)02004006008001000 W λ ( m Å ) ∆ v > − ∆ v ≤ − ρ gal /R vir W λ ( m Å ) Fig. 8.
Equivalent width of Ly α absorbers (blue dots) plotted againstthe impact parameter to the nearest galaxy (upper panel) or against theimpact parameter in units of the galaxy’s virial radius (lower panel).The sample has been split into components that lie within 1000 km s − of the nearest filament segment (red) and ones with a larger velocitydi ff erence (blue). Black crosses indicate sightlines that exhibit no sig-nificant Ly α absorption in the analysed spectral region. For these, wegive the distance to the nearest galaxy in the velocity range v = − − . nearest galaxy. Following Shull (2014); Wakker et al. (2015),this is the characteristic radius up to which the gas surround-ing a galaxy is immediately associated with that galaxy and itscircumgalactic gas-circulation processes (infall, outflows, merg-ers). It corresponds to ∼ − − of the galaxy’s velocityto associate each absorber with either the galaxy or the filament.This velocity range (which we also adopt here; see above) isjustified in view of other dynamic processes that would cause l o g N ( H I ) ∆ v > − ∆ v ≤ − ρ gal (Mpc)0102030405060 b ( k m s − ) Fig. 9.
Logarithmic H i column density and the Doppler parameterof Ly α absorbers are plotted against ρ gal for the same two samples asshown in Fig. 8. a Doppler shift of the gas in relation to the galaxy’s mean ra-dial velocity, such as galaxy rotation, velocity dispersion of gas-structures within the virialised dark-matter of the host galaxy, aswell as in- and outflows.In Fig. 9 we show how the H i column density (log N (H i ))and the Doppler parameter ( b value) vary with ρ gal . Similarly to W λ , the largest values for log N (H i ) and b are found at smallerimpact parameters, but (again) the scatter is large.Wakker et al. (2015) have also plotted the equivalent widthversus impact parameter to the nearest galaxy for their sample.Although there are some high equivalent width absorbers at large ρ gal (out to 2000 kpc), the average equivalent width decreaseswith increasing ρ gal . Similar to our sample, Wakker et al. (2015)find the majority of the absorbers within 1 Mpc of a galaxy. Oursample, however, has a larger scatter and more strong absorbersat larger distances. Prochaska et al. (2011) also conclude there isan anti-correlation between equivalent width and galaxy impactparameter for their sample that has a maximum ρ gal of 1 Mpc. Inaddition to stronger absorbers having lower impact parameters,their sample shows an increase of the number of weak absorbers( W λ <
100 mÅ) with increasing impact parameter.
6. Ly α and its connection to filaments In Fig. 10, the filaments are plotted together with the positionof the COS sightlines (filled squares) and the velocities of thedetected Ly α components colour-coded (in the same way asthe galaxies). Only those absorption components are consid-ered that have velocities within 1000 km s − of the nearest fila-ment segment. These plots are useful to visualise the large-scalekinematic trends of the absorption features along each filament,while at the same time the spatial and kinematic connection be-tween Ly α components and individual galaxies can be explored.In the green filament (a), Ly α absorption is predominantlyfound near 1500 km s − . This holds true for both the sightlinesat the outskirts of the filament and those going through the VirgoCluster. For the latter, this indicates the gas has a higher velocitythan the typical velocity of galaxies in the Virgo Cluster (as men-tioned earlier, the V8k catalogue takes into account the Virgo-flow model by Shaya et al. (1995)). Due to the extended Ly α Article number, page 8 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments b ( ◦ ) (a) (d) b ( ◦ ) (b) l ( ◦ )6045301501530 (e) l ( ◦ )907560453015015 b ( ◦ ) (c) 01000200030004000300040005000600001000200030004000 v fl ( k m s − ) v fl ( k m s − ) v fl ( k m s − ) Fig. 10.
Same as Fig. 6, but here with Ly α absorbers (coloured squares) overlaid that fall within 1000 km s − of the filament’s velocity. Multipleabsorbers along the same sightline have been given spatial o ff set. COS sightlines that do not exhibit Ly α absorption in this range, are indicated byblack crosses. trough of the Galactic foreground ISM absorption, interveningLy α absorption below ∼ − cannot be measured in ourCOS data set, so that our absorption statistics is incomplete at thelow end of the velocity distribution. Still, the trend of decreasinggalaxy velocities with increasing distance to the Centaurus Clus-ter (see above) is not reflected in the kinematics of the detectedLy α absorbers in this filament, which appears to be independentof the large-scale galaxy kinematics.The purple filament (b) and first section of the dark blue fil-ament (c) overlap on the sky and have two COS sightlines incommon. The di ff erent filament velocities allow us to assign thedetected Ly α absorption in one of the sightlines to the purplefilament, while the other sightline has one absorption compo-nent that we associate with the dark blue filament. With only 2sightlines available for the purple filament, no clear trends canbe identified.As the dark blue filament continues, the di ff erent ‘branches’noted earlier in Sect. 4.4 are also reflected in the velocities ofthe Ly α absorption components. This trend might be partly a result of our original selecting criterion for filament-related ab-sorbers (absorption within 1000 km s − of the closest filament-segment velocity; see above). However, because of the large ve-locity range used, the selection criterion cannot account for theentire branching e ff ect. Obviously, in this filament, the gas tracesthe velocities of the galaxies. Since this is the most di ff use fila-ment, the chance of finding a Ly α absorber, that is not directlyassociated with a galaxy but rather traces the large-scale flow ofmatter in that filament, is higher.The cyan filament (d) instead is well-populated with galax-ies, while also being relatively long and broad. It thus has a highcross-section and there are several sightlines that pass throughthis structure. Also in this case, the Ly α absorption appears tofollow the velocity trend of the galaxies in the filament here.Starting from Centaurus, the absorbers first exhibit velocitiesaround 1800 km s − , then the velocities decrease several hun-dred km s − , to rise again slightly at the end of the filament, inline with the galaxies’ velocity pattern. Article number, page 9 of 16 & A proofs: manuscript no. Lya_in_fil ρ fil (Mpc)0100200300400500600700800900 W λ ( m Å ) Fig. 11. Ly α equivalent width versus filament impact parameter for ab-sorbers with velocities within 1000 km s − of the nearest filament seg-ment. Open circles indicate absorbers that are associated with a galaxy(passing it within 1.5 R vir and ∆ v <
400 km s − ), closed circles are notassociated with a (known) galaxy. The di ff erent colours indicate the in-dividual filaments. Most of the sightlines that pass the magenta filament (e) arenot suited for a spectral analysis. The one sightline that has beenanalysed, shows no significant absorption in the relevant velocityrange, implying that no useful information is available for themagenta filament.In analogy to Fig. 8, Fig. 11 shows the equivalent width ofthe Ly α absorbers, but now plotted against the filament impactparameter, ρ fil . To evaluate whether the absorbers are related toa nearby galaxy, those absorbers that pass within 1.5 R vir and ∆ v <
400 km s − of a galaxy are shown as open circles, whereasabsorbers not associated with a (known) galaxy are indicatedwith closed circles. In the Appendix we show in Fig. B.1 the ef-fect of varying the impact-parameter criterion for absorbers to beassociated with a galaxy between 1 . . vir (which leadsto no further insights, however).While some of the absorbers with the highest equivalentwidths are associated with a galaxy, this is not true for all strongabsorbers. Neither sub-sample shows a clear, systematic trendfor the equivalent width scaling with ρ fil , except that the maxi-mum Ly α equivalent width in a given ρ fil bin decreases with inincreasing distance. However, both sub-samples show a higherabsorber density within ρ fil < ρ fil =
13 Mpc, but these absorbers are unlikely to be part of thegreen filament, as the typical width of a cosmological filament isa few Mpc (Bond et al. 2010). But even if we limit our analysisto absorbers with ρ fil < α equivalent widths versusfilament impact parameter remains.The velocity trends for galaxies and absorbers along four fil-aments (green, purple, dark blue, cyan) are shown in Fig. 12.Starting point for each filament is the Centaurus-Cluster region.Here, each sub-box (segment) is defined to have a length of 10 ◦ on the sky. This is half the length of the sub-boxes (segments)used to define the filament axes (see Sect. 4.1), because here,sub-boxes (segments) do not overlap. Only for the second partof the dark blue filament (sub-boxes 12 − ◦ was chosen to have a su ffi cient number of galaxies available for thedetermination of a meaningful average velocity.Both the green and cyan filaments (Fig.12a and d) show aclear decrease in velocity as they extend further away from theCentaurus Cluster. For these two filaments, the velocities of thedetected Ly α absorbers all lie above the lower limit of the galaxyvelocity-dispersion in each sub-box, with only one exception(see sub-box 3 in the green filament, where one absorber fallsjust below the shaded area). This could possibly mean that thereis a void of absorbers in the region between the filament andthe Milky Way. However, it is important to recall that absorberswith velocities less than ∼ − could not be measureddue to the Galactic foreground absorption. This limit could alsoexplain why there is no absorption in the lower velocity range ofthe Virgo Cluster (see sub-boxes 6 and 7 in Fig. 12a).Furthermore, most absorbers in sub-boxes 2 − ρ fil ( >
7. Absorber statistics
In quasar-absorption spectroscopy, the observed relation be-tween the number of H i absorption systems in the column den-sity interval ∆ N ( N H I to N H I + dN ) and the absorption-path lengthinterval ∆ x ( X to X + dX ) is commonly characterised by the dif-ferential column density distribution function (CDDF), f ( N H I ).We use the formalism described in Lehner et al. (2007, and ref-erences therein) and adopt the following expression to describethe di ff erential CDDF of our Ly α absorbers: f ( N H I ) dN H I dX = C H I N − β H I dN H I dX . (1)Following e.g. Tytler (1987), absorption path ∆ x and redshiftpath ∆ z at z ≈ ∆ X = . + ∆ z ) − , where we calculate the redshift pathlength ∆ z for the vari-ous sightline samples as described in Sect. 3.2. The slope of theCDDF is given by the exponent β , while the normalisation con-stant, C H I can be calculated via the relation C H I ≡ m tot (1 − β ) / { N − β max [1 − ( N min / N max ) − β ] } Here, m tot is the total number of absorbers in the column-density interval N min to N max .The column density distributions for our sample of Ly α ab-sorbers are shown in Fig. 13. The CDDFs were fitted for Ly α components with log N (H i ) ≥ . N (H i ) = Article number, page 10 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments v fl ( k m s − ) (a) (c) v fl ( k m s − ) (b) (d) N u m b e r o f g a l a x i e s N u m b e r o f g a l a x i e s Fig. 12.
Average galaxies velocities along four filaments (dots, (a): green; (b): purple; (c): dark blue; (d): cyan) plotted together with the velocitiesof Ly α absorbers (squares) for each sub-box (segment). The velocity dispersion is indicated by the colour-shaded area. The grey bars indicate thenumbers of galaxies belonging to each sub-box (segment) in the filament.
12 13 14 15 16log N ( HI )20191817161514 l o g [ f ( N )] β =1.63 β =1.47 All ∆ v ≤ − Fig. 13. H i column-density distribution function for the Ly α absorbersin sightlines close to filaments (blue) and the absorbers falling within1000 km s − from the filament velocity (red). Errors in log[f( N )] arefrom Poisson statistics. β = . ± .
12, while for the sub-sample of absorbers within 1000 km s − of the filament velocity we obtain β = . ± .
24. Using high-resolution STIS data, Lehner et al. (2007) derived β = . ± . ≤
40 km s − , 110 ab-sorbers) and β = . ± .
06 for b ≤
150 km s − (140 absorbers),thus slightly steeper slopes than the distributions found here.Note that we do not split our sample based on b -values, as thefraction of absorbers with b >
40 km s − is small ( ≤ N ( HI )11.512.012.513.013.5 l o g N ( S i III ) Fig. 14.
Relation between log N (Si iii ) and log N (H i ) for 13 systems inour absorber sample. A comparison with other recent studies can be made, for ex-ample with the Danforth et al. (2016) COS study of the low-redshift IGM, which yields β = . ± .
02 for a redshift-limitedsub-sample of 2256 absorbers. Other results are: β = . ± . β = . ± .
07 by Penton et al. (2000) (187 absorbers from GHRSand STIS), and β = . ± .
03 by Tilton et al. (2012) (746absorbers from STIS). Most of these values are consistent with β = . − .
70, whereas higher values ( β > .
70) may indicatea redshift evolution of the slope between z = . − .
4. Suchan evolution was discussed in Danforth et al. (2016), who find asteepening of the slope with decreasing z in this redshift-range. Article number, page 11 of 16 & A proofs: manuscript no. Lya_in_fil
For our study, with z ≤ . z = β . For instance, the spectral resolution of COS( R ≈ , R ≈ , α absorption components might be composedof several (unresolved) sub-components with lower column den-sities. The limited S / N in the COS data additionally hampersthe detection of weak Ly α satellite components in the wings ofstronger absorbers (see also Richter et al. 2006a, their Fig. 1).In this series of results, the shallower CDDF ( β = . ± .
24) for the sub-sample of velocity-selected absorbers within1000 km s − of the filaments stands out. Although this low valueis formally in agreement within its error range with the canoni-cal value of β = .
65, it may hint at a larger relative fraction ofhigh-column density systems in the filaments, reflecting the spa-tial concentration of galaxies and the general matter overdensityin these structures. A larger sample of Ly α absorbers associatedwith filaments would be required to further investigate this inter-esting aspect on a statistically secure basis.Like previous studies, our sample o ff ers an opportunity tostudy the number of Ly α absorbers per unit redshift ( d N / dz ).Table 1 gives the Ly α line density for the entire sample, as wellas for several subsamples. Those subsamples separate the sam-ple into di ff erent column-density bins, allowing us to directlycompare the results to the high-resolution Lehner et al. (2007)absorber sample and other studies.For the full absorber sample (including filament and non-filament related sightlines), the Ly α line density is 116 ± . (cid:46) log N (H i ) (cid:46) . . (cid:46) log N (H i ) (cid:46) .
5, we derive Ly α line densities of 88 ± ±
8, respectively. These values are in good agreementwith those reported by Lehner et al. (2007), who derive numberdensities of 80 ± ± d N / dz in the form d N ( > N ) / dz ≈
25 ( N / cm − ) − . . For absorbers withlog N (H i ) ≥ .
2, this leads to d N / dz ∼
83, mildly lower thanthe values derived by us and Lehner et al. (2007), but still in fairagreement.If we take the velocity-selected absorber sample, which po-tentially traces the Ly α gas associated with the filaments, we ob-tain a significantly higher line density of d N / dz = ± ± − around the center-velocity for the fila-ment segment that was closest to that sightline.The value of 189 ±
25 for the velocity-selected filament sam-ple is 93 percent higher than the value derived for the totalfilament-absorber sample (along the same lines of sight). Thisline overdensity of the Ly α forest kinematically associated withfilaments obviously reflects the matter overdensity of baryons inthe potential wells of these large-scale cosmological structures.For the sake of completeness, we also show in Fig. 14 therelation between log N (Si iii ) and log N (H i ) for absorbers in oursample for which both species are detected. Only for a smallfraction (8.4 %) of the Ly α components, Si iii can be measured,which is partly because of the velocity-shifted Si iii falling inthe range of Galactic Ly α absorption. Generally, log N (Si iii ) in-creases with log N (H i ), as expected from other Si iii surveys in Table 1. Ly α line density for the full sample (filament and non-filamentrelated sightlines), filament related sightlines, and for the velocity se-lected absorber sample ( ∆ v < − ). log N (H i ) N d N / dz Full sample 579 116 ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± iii / H i absorbers in our sam-ple does not allows us to draw any meaningful conclusions onthe metal content of the absorbers in relation to their large-scaleenvironment.
8. Discussion on Ly α absorbers and theirenvironment In their study, Prochaska et al. (2011) correlated galaxies andLy α absorbers at z = . − .
57 and found that for weakabsorbers (13 ≤ log N (H i ) ≤
14) less than 20 % of the sys-tems were associated with a known galaxy, while for strong ab-sorbers (log N (H i ) ≥ ff erence between absorber and galaxy of ≤
400 km s − , and ii) an impact parameter of ρ ≤
300 kpc. Usingthe same criteria, we derive for our sample that 10 % (40 %) ofthe low (high) column density absorbers are associated with agalaxy in the V8k galaxy sample.Therefore, and in agreement with Prochaska et al. (2011),we find that high column density Ly α absorbers are four-timesmore often associated with a galaxy than low column densityabsorbers, but the overall fraction of absorbers for which an as-sociated galaxy is found is only half of that in the Prochaska et al.(2011) sample. This can be attributed to the fact that the V8k cat-alogue is incomplete for M B < −
16 and v > − , whilethe Prochaska et al. (2011) galaxy sample is complete up to atleast z = . L < . L ∗ . By comparing theirobserved covering fractions with a filament model, Prochaskaet al. (2011) conclude that all Ly α absorbers are associated witheither a galaxy or a filament. This view is debated by Tejos et al.(2012), however, who argue that there is an additional populationof ‘random’ Ly α absorbers that reside in the underdense large-scale structure (voids).The idea of Ly α absorbers belonging to di ff erent populations(and thus di ff erent environments) was proposed more than 25years ago by Morris et al. (1993). By analysing Ly α absorbersin a single sightline and comparing the location of the absorberswith locations of galaxies, these authors found that the absorbersdo not cluster around galaxies as strongly as galaxies clusteramong themselves. On the other hand, they also found the trendthat the absorbers do cluster around galaxies. From this, theyconcluded that there could be two populations of Ly α absorbers:one that is associated with galaxies and one that is more or lessrandomly distributed. Article number, page 12 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments To test whether the Ly α absorbers in our sample resemblea ‘random population’, we generated two artificial populationsof Ly α absorbers, both with random sky positions, random ab-sorption velocities within the assumed v fil ± − velocityrange of a filament, and random H i column densities weightedby the H i CDDF. For the one population, we have restrictedthe sample from Lehner et al. (2007) (hereafter abbreviated withL07) to the redshift range spanned by the filaments in our studyand used the slope of their CDDF (resulting in 39 absorbers, β = . β = . ≤ ff erent Ly α absorber samples (observed sample, ran-dom sample with own statistics, random sample with L07 statis-tics) depend on ρ gal . Like in Sect. 5, ρ gal was calculated for thenearest galaxy on the sky within a velocity interval of 400 km s − from the absorber. Clearly, the measured absorbers cluster morestrongly around galaxies than both random samples. This in-dicates that at least some of the absorbers are associated withgalaxies, as expected from previous studies (e.g. Morris & Jan-nuzi 2006; Prochaska et al. 2011; Tejos et al. 2014; French &Wakker 2017).A very rough estimate of the fraction of absorbers associ-ated with a galaxy can be made by comparing the fraction of ab-sorbers within 1.5 Mpc of a galaxy in di ff erent samples. For themeasured absorbers 82 % of the absorbers have ρ gal ≤ . α absorbers to thenearest filament axis, as shown in Fig.16, a similar, albeit lesspronounced trend can be seen. Measured absorbers are gener-ally found closer to the filament axis than a random distributionshows. One may argue that this could be a selection e ff ect, e.g.,from targeting particularly interesting areas such as the VirgoCluster, which would result in more sightlines near the Virgofilament. However, Fig.11 showed the break-down of absorbersinto di ff erent filaments, demonstrating that the absorbers belong-ing to the Virgo Cluster filament (green) are in fact more spreadout than absorbers in other regions, speaking against such a se-lection e ff ect.To further investigate the possible clustering signal of Ly α absorbers near the filament axis, we have plotted in Figure 17 thecumulative distribution function for ρ fil for the three previouslymentioned absorber samples (observed sample, random samplewith own statistics, random sample with L07 statistics) as wellas the galaxies that constitute the filament. We also have addedanother absorber test sample (D16) generated from the Ly α col-umn density distribution of absorbers reported in Danforth et al.(2016). The cumulative distribution of galaxies set the referencepoint, as these define the filament. As can be seen, the observeddistribution of absorbers cluster more strongly around the fila-ment axis than the three random absorber test distributions, but l o g N ( H I ) This work13141516 l o g N ( H I ) Random, this work0 2 4 6 8 10 ρ gal (Mpc)13141516 l o g N ( H I ) Random, L07
Fig. 15. H i column density versus ρ gal for Ly α absorbers, i) as mea-sured in the COS data (upper panel), ii) for a randomised sample fol-lowing the CDDF in this work (middle panel), and iii) for a randomisedsample following the number statistics and CDDF from Lehner et al.(2007) (lower panel). l o g N ( H I ) This work13141516 l o g N ( H I ) Random, this work0 2 4 6 8 10 12 14 ρ fil (Mpc)13141516 l o g N ( H I ) Random, L07
Fig. 16.
Same as Fig. 15, but now for ρ fil . not as strongly as the galaxies. Within the inner 2 Mpc, in par-ticular, the fraction of measured Ly α absorbers rises faster thanthe synthetic absorbers in the randomised samples.The cumulative distribution function as shown in Fig. 17 canbe compared with one for absorbers associated with galaxies.Fig. 3 in Penton et al. (2002) shows this function for 46 Ly α absorbers and subsamples thereof. Their full sample follows adistribution similar to our absorbers, with ∼
60 % found within2 Mpc of the nearest galaxy (Penton et al. 2002) or filament(this study). Both studies show the galaxies more strongly clus-tered than the Ly α absorbers. Stocke et al. (2013) compared theirabsorber-galaxy cumulative distribution function with a randomdistribution concluded that absorbers are associated with galax-ies in a more general way, i.e., tracing the large-scale structureinstead of individual galaxies. Penton et al. (2002); Stocke et al.(2013); Keeney et al. (2018) all conclude that high-column den-sity absorbers are more strongly correlated with galaxies than Article number, page 13 of 16 & A proofs: manuscript no. Lya_in_fil ρ fil (Mpc)0.00.20.40.60.81.0 C u m u l a t i v e f r a c t i o n This workRandom, this workRandom, L07Random, D16Galaxies
Fig. 17.
Cumulative distribution function for ρ fil for three di ff erent ab-sorber samples. The measured absorbers in the COS data are indicatedin blue, the random sample with own statistics is plotted in dashedgreen, the random sample with the L07 statistics is indicated in dot-ted black, and the random sample from Danforth et al. (2016) is addedin dashed magenta (D16). The distribution for the galaxies is shown inred. those with lower column densities. This is in agreement withwhat is found here: Ly α absorbers do not trace the same distri-bution as galaxies, but they are not randomly distributed aroundfilaments either.A Kolmogorov-Smirnov (KS) test confirms that the ab-sorbers found in the COS data are not drawn from a randomdistribution. Table 2 lists the KS values and p -values for the dif-ferent samples.A KS-test can be used to compare two di ff erent samples toevaluate whether if they are drawn from the same parent distribu-tion. A high KS value (maximum of 1) indicates a high probabil-ity that this is true. The p -value instead indicates the significancethat the null hypothesis is rejected. In this case, the null hypothe-sis is that ρ fil for the absorbers measured in this study follows thesame distribution as that of a random sample, or of the galaxiesin the filament. A p -value of < .
05 means the null hypothesiscan be rejected with a probability of >
95 %. In all three compar-isons between the measurements and the random absorber sam-ples (our own sample, randomised, and the randomised L07 andD16 samples), the KS-test indicates they do not follow the samedistribution.The low p -value that we obtain from comparing the COSabsorber sample with the V8k galaxy sample in the filamentsfurther indicates that the the Ly α absorbers are not drawn fromthe same distribution as the galaxies. Evaluating the hypothesisthat there are two populations of Ly α absorbers (e.g. Morris et al.1993; Tejos et al. 2012), we have removed from the sample (as atest) those absorbers that we have associated with galaxies. This Table 2.
Kolmogorov-Smirnov test for impact-parameter distributions
Sample compared KS statistics p -valueRandom, this work 0.28 0.003Random, L07 0.31 0.011Random, D16 0.32 0.013V8k galaxies 0.26 0.032 has, however, no significant e ff ect on the cumulative distributionfunction for ρ fil .In conclusion, the cumulative distribution functions for ρ fil show that galaxies are more strongly clustered in the filamentsthan the Ly α absorbers that belong to the same cosmologicalstructures. Ly α absorbers do not follow a random distributionand neither do they follow the same distribution as the galax-ies that constitute large-scale filaments. There might be two (ormore) separate populations of Ly α absorbers in filaments, but(from our study) there is no evidence that the Ly α absorbers thatare not directly associated with large galaxies are randomly dis-tributed in the field of the filament.Deeper insights into these aspects (including other importantcosmological issues such as overdensity bias-factors and howthey a ff ect the absorber / galaxy / filament statistics) are highly de-sirable, but will require a much larger observational data set incombination with numerical cosmological simulations.
9. Summary and concluding remarks
In this study, we have combined galaxy data of more than30,000 nearby galaxies from the V8k catalogue Courtois et al.(2013) with HST / COS UV spectral data of 302 distant AGN toinvestigate the relation between intervening H i Ly α absorbersand five nearby cosmological structures (galaxy filaments) at z ≈ v < − ).(1) All in all, we identify 587 intervening Ly α absorbersalong the 302 COS sightlines in the wavelength range between1220 and 1243 Å. For the 91 sightlines that pass the immediateenvironment of the examined galaxy filaments we analysedin detail 215 (229) Ly α absorption systems (components) andderived column densities and b -values for H i (and associatedmetals, if available).(2) For the individual galaxies in our sample, we have cal-culated the virial radii from their luminosities and the galaxyimpact parameters, ρ gal , to the COS sightlines. We assume 29Ly α absorbers to be directly associated with galaxies, as theyare located with 1.5 virial radii of their host galaxies and within400 km s − of the galaxies’ recession velocity.(3) We characterise the geometry of the galaxy filamentby considering the galaxy distribution in individual segmentsof the filaments. In this way, we define for each filament ageometrical axis that we use as reference for defining thefilament impact parameters, ρ fil , for those Ly α absorbers that arelocated within 1000 km s − of the filament.(4) We find that the absorption velocities of the Ly α ab-sorbers reflect the large-scale velocity pattern of the four galaxyfilaments, for which su ffi cient absorption-line data are available.74 absorbers are aligned in position and velocity with the galaxyfilaments, indicating that these absorbers and the galaxies tracethe same large-scale structure .(5) If we relate the measured Ly α equivalent widths (orH i column densities) with the galaxy and filament impactparameters, we find that the strongest absorbers (equivalentwidths W λ >
500 mÅ) are preferentially located in the vicinityof individual galaxies (within 3 virial radii) and / or in the vicinityof the filament axes (within 5 Mpc). The observed relations Article number, page 14 of 16. J. D. Bouma et al.: On the relation between Ly α absorbers and local galaxy filaments between W and ρ gal / ρ fil exhibit substantial scatter, however,disfavouring a simple equivalent width / impact parameter anti-correlation.(6) We find that the measured H i components follow a column-density distribution function with a slope of − β = − . ± . α forest. Only forthe sub-sample of absorbers within 1000 km s − of the filamentvelocity do we obtain a shallower CDDF with β = . ± . α forest.(7) The Ly α absorbers that lie within 1000 km s − of thenearest filament have a ∼
90 percent higher rate of incidence( d N / dz = ±
25 for log N (H i ) ≥ .
2) than the general Ly α absorber population in our sample ( d N / dz = ± N (H i ) ≥ . α absorbers perunit redshift most likely reflects the filaments’ general matteroverdensity.(8) We compare the filament impact-parameter distribu-tions of the galaxies, measured Ly α absorbers, and a (synthetic)Ly α absorber sample with randomised locations on the sky witheach other. We find that the galaxies are most strongly clusteredaround the filament axes, while the spatial clustering of theobserved Ly α absorbers around the filament axes is evident,but less pronounced. Using a KS test, we confirm that the Ly α absorbers neither follow the impact-parameter distribution ofthe galaxies, nor do they follow a random distribution, butrepresent an individual, spatially confined sample of objects.Taken together, the results of our study underline that therelation between intervening Ly α absorbers, large-scale cosmo-logical filaments, and individual galaxies (that constitute thefilaments) in the local universe is complex and manifold, anddi ffi cult to reconstruct with existing data.This complexity is not surprising, of course, if we recall,what Ly α absorbers actually are: they are objects that trace localgas overdensities in an extremely extended, di ff use medium thatis gravitationally confined in hierarchically structured potentialwells, and stirred up by large-scale matter flows and galaxy feed-back. In this picture, the spatial distribution of Ly α absorbers incosmological filaments is governed by both the distribution ofindividual sinks in the large-scale gravitational potential energydistribution (i.e., galaxies, galaxy groups etc.) and more (or less)stochastically distributed density fluctuations at larger scales thatreflect the internal dynamics of the IGM.For the future, we are planning to extend our study of therelation between intervening absorbers and cosmological fila-ments in the local universe by using a larger (and deeper) galaxysample and additional HST / COS spectra, in combination withconstrained magneto-hydrodynamic cosmological simulationsof nearby cosmological structures.
Acknowledgements.
The authors would like to thank the referee for his valuablecomments and suggestions which helped to improve the manuscript.
References
Bond, N. A., Strauss, M. A., & Cen, R. 2010, MNRAS, 409, 156Bowen, D. V., Pettini, M., & Blades, J. C. 2002, ApJ, 580, 169Bower, R. G., Benson, A. J., Malbon, R., et al. 2006, MNRAS, 370, 645Cen, R. & Ostriker, J. P. 1999, ApJ, 514, 1Chen, H.-W., Lanzetta, K. M., Webb, J. K., & Barcons, X. 2001, ApJ, 559, 654Colless, M., Dalton, G., Maddox, S., et al. 2001, MNRAS, 328, 1039 Courtois, H. M., Pomarède, D., Tully, R. B., Ho ff man, Y., & Courtois, D. 2013,AJ, 146, 69Danforth, C. W., Keeney, B. A., Tilton, E. M., et al. 2016, ApJ, 817, 111Danforth, C. W. & Shull, J. M. 2008, ApJ, 679, 194Dashtamirova, D., Fischer, W. J., et al. 2018, Cosmic Origins Spectrograph In-strument Handbook, Version 12.0 (Baltimore: STScI)Davé, R., Hernquist, L., Katz, N., & Weinberg, D. H. 1999, ApJ, 511, 521Davies, J. J., Crain, R. A., McCarthy, I. G., et al. 2019, MNRAS, 485, 3783Erb, D. K. 2008, ApJ, 674, 151Fairall, A. P., Woudt, P. A., & Kraan-Korteweg, R. C. 1998, A&AS, 127, 463French, D. M. & Wakker, B. P. 2017, ApJ, 837, 138Genzel, R., Tacconi, L. J., Gracia-Carpio, J., et al. 2010, MNRAS, 407, 2091Green, J. C., Froning, C. S., Osterman, S., et al. 2012, ApJ, 744, 60Keeney, B. A., Stocke, J. T., Pratt, C. T., et al. 2018, ApJS, 237, 11Lehner, N., Savage, B. D., Richter, P., et al. 2007, ApJ, 658, 680Madau, P., Ferrara, A., & Rees, M. J. 2001, ApJ, 555, 92Martizzi, D., Vogelsberger, M., Artale, M. C., et al. 2019, MNRAS, 486, 3766Mei, S., Blakeslee, J. P., Côté, P., et al. 2007, ApJ, 655, 144Morris, S. L. & Jannuzi, B. T. 2006, MNRAS, 367, 1261Morris, S. L., Weymann, R. J., Dressler, A., et al. 1993, ApJ, 419, 524Morton, D. C. 2003, ApJS, 149, 205Pallottini, A., Gallerani, S., & Ferrara, A. 2014, MNRAS, 444, L105Penton, S. V., Shull, J. M., & Stocke, J. T. 2000, ApJ, 544, 150Penton, S. V., Stocke, J. T., & Shull, J. M. 2002, ApJ, 565, 720Prochaska, J. X., Weiner, B., Chen, H. W., Mulchaey, J., & Cooksey, K. 2011,ApJ, 740, 91Prochaska, J. X. & Wolfe, A. M. 2009, ApJ, 696, 1543Richter, P., Fang, T., & Bryan, G. L. 2006a, A&A, 451, 767Richter, P., Fox, A. J., Wakker, B. P., et al. 2013, ApJ, 772, 111Richter, P., Krause, F., Fechner, C., Charlton, J. C., & Murphy, M. T. 2011, A&A,528, A12Richter, P., Nuza, S. E., Fox, A. J., et al. 2017, A&A, 607, A48Richter, P., Paerels, F. B. S., & Kaastra, J. S. 2008, Space Sci. Rev., 134, 25Richter, P., Savage, B. D., Sembach, K. R., & Tripp, T. M. 2006b, A&A, 445,827Richter, P., Wakker, B. P., Fechner, C., et al. 2016, A&A, 590, A68Saunders, W., d’Mellow, K. J., Tully, R. B., et al. 2000a, in Astronomical Societyof the Pacific Conference Series, Vol. 218, Mapping the Hidden Universe:The Universe behind the Mily Way - The Universe in HI, ed. R. C. Kraan-Korteweg, P. A. Henning, & H. Andernach, 153Saunders, W., Sutherland, W. J., Maddox, S. J., et al. 2000b, MNRAS, 317, 55Shaya, E. J., Peebles, P. J. E., & Tully, R. B. 1995, ApJ, 454, 15Shull, J. M. 2003, in Astrophysics and Space Science Library, Vol. 281, TheIGM / Galaxy Connection. The Distribution of Baryons at z =
0, ed. J. L. Rosen-berg & M. E. Putman, 1Shull, J. M. 2014, ApJ, 784, 142Shull, J. M., Smith, B. D., & Danforth, C. W. 2012, ApJ, 759, 23Stocke, J. T., Keeney, B. A., Danforth, C. W., et al. 2013, ApJ, 763, 148Tejos, N., Morris, S. L., Crighton, N. H. M., et al. 2012, MNRAS, 425, 245Tejos, N., Morris, S. L., Finn, C. W., et al. 2014, MNRAS, 437, 2017Tejos, N., Prochaska, J. X., Crighton, N. H. M., et al. 2016, MNRAS, 455, 2662Tilton, E. M., Danforth, C. W., Shull, J. M., & Ross, T. L. 2012, ApJ, 759, 112Tully, R. B., Rizzi, L., Shaya, E. J., et al. 2009, arXiv e-prints, arXiv:0902.3668Tumlinson, J., Shull, J. M., Rachford, B. L., et al. 2002, ApJ, 566, 857Tytler, D. 1987, ApJ, 321, 49Wakker, B. P., Hernandez, A. K., French, D. M., et al. 2015, ApJ, 814, 40Wakker, B. P. & Savage, B. D. 2009, ApJS, 182, 378York, D. G., Adelman, J., Anderson, John E., J., et al. 2000, AJ, 120, 1579
Article number, page 15 of 16 & A proofs: manuscript no. Lya_in_fil
Appendix A: Signal-to-noise N u m b e r OutsideInside 0 10 20 30 40 50W λ limit (m Å )05101520 N u m b e r Fig. A.1.
Left panel: distribution of the measured S / N ratios per resolution element near Å 1240 for the COS spectra that are filament-relatedand those outside of filaments.
Right panel: formal 3 σ detection limits for H i Ly α absorption in these spectra, based on the equation given inTumlinson et al. (2002). Note that these values reflect the detectability of Ly α absorption as a function of the local S / N under idealised conditions(no blending, no fixed-pattern noise, perfectly known continuum flux).
Appendix B: Absorbers associated with galaxies W λ ( m Å ) ρ gal ≤ R vir ρ gal ≤ vir ρ fil (Mpc)0100200300400500600700800900 W λ ( m Å ) ρ gal ≤ vir ρ fil (Mpc)0100200300400500600700800900 ρ gal ≤ vir Fig. B.1.
Same as Fig. 11, but now using di ff erent impact-parameter criteria ρ galgal