V2'#$0D*:$0()%"*,-.!/ K'(B5*2#*0D; T2&3B5U

V2'#$0D*:$0()%"*,-.!/ K'(B5*2#*0D; T2&3B5U

2 S-Cam NB101 Observations KONNO ET AL. tion. The previous studies suggest that the fraction of strong LAEs in Lyman break galaxies (LBGs) decreases from z 6 to 7 in contrast to the increase from z 3 to 6 (Pentericci et al. 2011; Schenker et al. 2012; Ono et al. 2012). This is because the neutral hydrogen fraction will have increased in z = 6 7. By the comparison with theoretical models, these results suggest xh I! 0.5. There are some other observational studies to probe when and how cosmic reionization took place. Observations of the Gunn & Peterson (1965) (GP) trough in quasar (QSO) spectra indicate that xh I 10 4 at z 6 (Fan et al. 2006), which suggest that cosmic reionization has been completed at this redshift. Measurements of the polarization of cosmic microwave background (CMB) by WMAP constraints the optical depth of Thomson scattering, τel = 0.081 ± 0.012, and suggests that if cosmic reionization has occurred instantaneously, the universe would be reionized at zinst = 10.1 ± 1.0 (Hinre Konno+ 14 shaw et al. 2013; Bennett et al. 2013). In the recent meafigure 1. Filter response curves of NB101 (red), NB1006 (blue) and z! surements by Planck, the electron scattering optical depth is (black) bands. All of these solid lines denote the response curves whose τel = 0.089+0.012 peaks are normalized to 1.0. These response curves include the quan 0.014, and instantaneous reionization would take inst tum efficiency of Hamamatsu CCDs (Kamata et al. 2008), airmass and place at zre = 11.1 ± 1.1 (Planck Collaboration et al. 2013). transmission+reflectionbandwidth of instrument and telescope optics. Note that LAEs (1) Custom filter sharp/narrow Totani et al. (2006) measure xh I using the shape of Lyα dampat z 7.3 can be identified with NB101, while LAEs in the redshift range ing wing appearing in the optical afterglow spectrum of GRB z = 7.2 7.3 can be detected with NB1006. The redshift of Lyα is shown 2 Subaru ultra-deep on the top in axis. SXDS+COSMOS ( 0.5 deg ) 050904 at(2) z 6.3, and estimate xh I < 0.17 (68%NB101 confidence survey level) at this redshift. In the recent study with a GRB, the z = 7.3 LAE survey and selection of the LAE candidates in unprecedentedly bright optical afterglow spectrum of GRB Section 2. We derive z = 7.3 Lyα LF and compare it with 130606A at z 5.9 suggests xh I = 0.1 0.5 (Totani et al. previous studies of z 7.3 ± 0.442 in Section 3. We examine 2013). Mortlock et al. (2011) report observations of a QSO the evolution of Lyα LF at z = 5.7 7.3, and discuss cosmic at z = 7.085, ULAS J1120+0641, and estimate xh I > 0.1 at reionization by the comparison of electron scattering optical this redshift the near-zone transmission profile. Bolton from Comparable with z=3-6 Subaru surveys depth with CMB results in Section 4. Throughout this paper, et al. (2011) use radiative transfer simulations taking into acwe adopt AB magnitudes (Oke 1974), and concordance cos- To detect very faint LAEs at z=7.3 NB101 Total exp. time : 106hr! 5-sigma limiting luminosity L(Lya) = 2.4 x 10 erg/s

!"! "! #! $! %&! '&! (! )!!"#$#%&'(&%)*$+!!"#$#%&'(&%+,-.)!!"#$#%&'(&%/*+$-!!"#$#%01&21&%/#/,!!"#$#%01&21&%3-$/$!!"#$#%01&21&%3-/+.!!"#$#%01&21&%#$3*,,!

Figure 11. Evolution of Lyα and UV luminosity densities. The red circles are the Lyα luminosity densities obtained by this study, Ouchi et al. (2010), and Ouchi et al. (2008) for z =7.3, 6.6, and5.7, respectively. ThebluecirclesaretheUVluminositydensities given by Bouwens et al. (2014) for z =5.9, 6.8, 7.9, and10.4, andellisetal.(2013) forz =9.0. TheleftordinateaxisisreferredfortheLyα luminosity densities, and the right ordinate axis is for the UV luminosity densities. The Lyα luminosity density starts evolving acceleratingly at z 7, whiletheuvluminositydensityrapidlydecreases atz 8 and beyond. The ρ Lyα and ρ UV knees are indicated with the arrows. Table 6 Best-Fit Parameters for Pure-Luminosity and Number Density evolutioncases

ti- We use theoretical models to constrain x HI at z =7.3with la- our estimates of T IGM IGM Lyα,z=7.3 /TLyα,z=5.7 =0.29. In the analytic model of Santos (2004), the Lyα transmission fraction F as of IGM is related to x HI in two cases of no galactic wind g- and a galactic outflow that give shifts of Lyα line from a systemic velocity by 0 and 360 km/s, respectively.thevalueof re ve T IGM IGM Lyα,z=7.3 /TLyα,z=5.7 =0.29 corresponds to x HI 0.0 and n- 0.8 in the former and the latter case, respectively. Because s. recent studies have reported that the Lyα line emission of es LAE at z =2.2is redshifted by 200 km/s (Hashimoto et al..7 2013; Shibuya et al. 2014), we take x HI 0.5 that is the x HI b- value interpolated by the Lyα velocity shift in Figure 25 of 12 KONNO ET AL. Santos (2004). McQuinn et al. (2007) predict Lyα LFs for te various x HI values with radiative transfer simulations. By the comparison of our Lyα LF with these simulation results in Figure 4 of McQuinn et al. (2007), we obtain x HI 0.7. In the models of Dijkstra et al. (2007a,b), the Lyα transmission

Simulations for Clumpy HI Clouds x (h cmpc) 100 14 x (h cmpc) hybrid box 14 Choudhury et al. Ionized region vint =100[(1 + z)/7] 3 km s 1 Lya transmission ratio 80 y (h 1 cmpc) 60 40 20 HI clouds 0 0 20 40 Choudhury et al. Choudhury+14, arxiv:1412.4790 60 80 100 x (h 1 cmpc) 1.4 Konno+14 1.2 1.0 0.8 0.6 0.4 T (z)/t (z = 5.5) default T (z)/t (z = 6.0) default T (z)/t (z = 5.5) late T (z)/t (z = 5.7) Konno et al. 2014 0.2 0.0 4. 5.0 5.5 6.0 6.5 z 7.0 Redshift 7.5 8.0 8.5 Figure 10. Evolution of the mean Lyα transmissivity Figure of the IGM, normalizedoftothe that at alyα particular redshift. of The shift vint 10. Evolution mean transmissivity theredward IGM, normalized 1 of the Lyα line is 100km s 1 in the left panel and assumed to Lyα evolve with redshifts 1 as v 100km + z)/7] 3 in the right panel int = of the line is 100km in the left panels and[(1 assumed to evolve with redsh The observational constraint is from Konno et al. (2014). The error bars on the simulation data display the 68 per cent scatter for The observational constraint is from Konno et al. (2014). The error bars onthe th Figure 3. Illustration of the construction of the high-resolution hybrid box including the large scale ionization field at z = 7. The top 25000 sight-lines considered. 1 left panel shows the large scale density field obtained from the large box of size 100 h Mpc, as well as the boundaries of the ionized 25000 sight-lines considered. Choudhury+14 high res. hydrodynamical simulation + intermediate res. DM simulation. regions obtained using the analytic model. The top right panel displays the neutral hydrogen density in the small box with the map of the ionized regions from the large box applied to it. The lower panel shows the neutral hydrogen density in the hybrid box, constructed by populating the volume of the large box with 103 randomly shifted and rotated copies of the small box. 5 CONCLUSIONS Loeb 2013; Becker Boltonvisibility. 2013) and the expected lower Estimate spatial distribution & redshift evolution of&lya 5 CONCLUSIONS Loeb 2 neutral hydrogen column densities in high redshift galaxies neutral c 0000 RAS, MNRAS 000, 000 000 We have combined high resolution hydrodynamical simulaat z > 6. resolution hydrodynamical simulawe have combined high at z > tions with an intermediate resolution collisionless, dark A modelresolution where reionization completes somewhat later tionsmatwith an intermediate collisionless, dark mata ter only simulation and an analytical model for the and perhaps more rapidly assumed tergrowth only simulation and an also analytical model than for the growthin the HM2012 and per Reproduce our estimation of IGM Lya transmission ratio.