/ Ê1*Ê/ Ê "- "-Ê “iÛ>Ê-ÌÀÕVÌÕÀiÊ/iiÃVœ«iÊ­*>-/®Ê>˜`]ʈ˜ÊÌ iʓœÀiÊ`ˆÃÌ>˜ÌÊ vÕÌÕÀi]ÊÌ iÊ-µÕ>Àiʈœ“iÌiÀÊ
ÀÀ>ÞÊ­-
®° ˜ÊÌ iÊLi}ˆ˜˜ˆ˜}ʜvÊÌ iÊ >ÀŽÊ
}iÃ]ÊiiVÌÀˆV>ÞʘiÕÌÀ>Ê ÞÊÃV>˜˜ˆ˜}Ê>VÀœÃÃÊÜ>Ûii˜}Ì Ã]ÊÌ iÃiÊ>ÀÀ>ÞÃÊ܈Ê«ÀœLiÊ Þ`Àœ}i˜Ê}>ÃÊwi`ÊÌ iÊ՘ˆÛiÀÃi°Ê
ÃÊÃÌ>ÀÃÊvœÀ“i`]ÊÌ iÞÊ Ì iÊÓ£‡Vi˜Ìˆ“iÌiÀÊi“ˆÃȜ˜Ê>ÌÊ`ˆvviÀi˜ÌÊ̈“iÃʈ˜ÊVœÃ“ˆVÊ ˆÃ̜ÀÞ°Ê ˆœ˜ˆâi`ÊÌ iÊÀi}ˆœ˜Ãʈ““i`ˆ>ÌiÞÊ>ÀœÕ˜`ÊÌ i“]ÊVÀi>̈˜}Ê
ÃÌÀœ˜œ“iÀÃÊ܈ÊLiÊ>LiÊ̜ÊLՈ`Ê>ÊÌ Àii‡`ˆ“i˜Ãˆœ˜>Ê“>«ÊœvÊ LÕLLiÃÊ iÀiÊ>˜`ÊÌ iÀi°Ê Ûi˜ÌÕ>ÞÊÌ iÃiÊLÕLLiÃʓiÀ}i`Ê Ì iʘiÕÌÀ>Ê Þ`Àœ}i˜Ê`ˆÃÌÀˆLṎœ˜°Ê/ iÞÊ܈ÊLiÊ>LiÊ̜ÊÜ>ÌV Ê Ìœ}iÌ iÀ]Ê>˜`ʈ˜ÌiÀ}>>V̈VÊ}>ÃÊLiV>“iÊi˜ÌˆÀiÞʈœ˜ˆâi`° `i˜ÃˆÌÞÊyÕVÌÕ>̈œ˜Ãʜvʜ˜iÊ«>ÀÌʈ˜Ê£ää]äääÊ­>Ãʈ˜ÊÌ iʓˆVÀœ‡ Ü>ÛiÊL>VŽ}ÀœÕ˜`®ÊLiVœ“iʜÀ`iÀÃʜvʓ>}˜ˆÌÕ`iÊ}Ài>ÌiÀ°Ê
ÌÊÌ iÊ œV>̈œ˜ÃʜvÊ}Ài>ÌiÃÌÊ`i˜ÃˆÌÞ]Ê}>>݈iÃÊà œÕ`ÊÌ>ŽiÊà >«iÊ>˜`Ê VÀi>ÌiÊLÕLLiÃʜvʈœ˜ˆâi`Ê Þ`Àœ}i˜°Ê/ iÊLÕLLiÃÊ܈Ê«ÀœˆviÀ‡ >ÌiÊ>˜`ʓiÀ}i]ÊiÛi˜ÌÕ>ÞÊVi>Àˆ˜}ʈ˜ÌiÀ}>>V̈VÊë>ViʜvʘiՇ ÌÀ>Ê Þ`Àœ}i˜ÊQÃiiÊLœÝÊ>ÌÊÀˆ} ÌR°Ê/ iÊà >À«˜iÃÃʜvÊÌ iÊLÕLLiÃ½Ê LœÕ˜`>ÀˆiÃÊ܈Ê>˜ÃÜiÀÊÌ iʵÕiÃ̈œ˜ÊœvÊÜ iÌ iÀÊÀiˆœ˜ˆâ>̈œ˜Ê Ü>ÃÊV>ÕÃi`ÊLÞʓ>ÃÈÛiÊÃÌ>ÀÃʜÀÊLÞÊL>VŽÊ œiÃ°Ê >ÃÈÛiÊÃÌ>ÀÃÊ «œÕÀʜÕÌʓœÃÌʜvÊÌ iˆÀÊi˜iÀ}Þʈ˜ÊՏÌÀ>ۈœiÌʏˆ} Ì]ÊÜ ˆV ʈÃÊÀi>`‡ ˆÞÊLœVŽi`ÊLÞʈ˜ÌiÀ}>>V̈VÊ Þ`Àœ}i˜]ÊÜ iÀi>ÃÊL>VŽÊ œiÃÊ }i˜iÀ>ÌiʓœÃ̏ÞÊ݇À>ÞÃ]ÊÜ ˆV Ê«i˜iÌÀ>ÌiÊ`ii«Þʈ˜ÌœÊÌ iÊ}>ðÊ-œÊ L>VŽÊ œiÃÊ«Àœ`ÕViÊvÕââˆiÀÊLœÕ˜`>ÀˆiÃ°Ê œÀÊÃiÛiÀ>ÊÀi>ܘÃ]ÊÌ iÊÓ£‡Vi˜Ìˆ“iÌiÀʓ>«Ê“>ÞÊV>ÀÀÞʓœÀiÊ LˆÌÃʜvʈ˜vœÀ“>̈œ˜ÊÌ >˜Ê>˜ÞÊœÌ iÀÊÃÕÀÛiÞʈ˜ÊVœÃ“œœ}Þp“œÀiÊ Ì >˜ÊiÛi˜ÊÌ iÊVœÃ“ˆVʓˆVÀœÜ>ÛiÊL>VŽ}ÀœÕ˜`°ÊˆÀÃÌ]ÊÜ iÀi>ÃÊ >˜Êˆ“>}iʜvÊÌ iʓˆVÀœÜ>ÛiÊL>VŽ}ÀœÕ˜`ʈÃÊÌܜ‡`ˆ“i˜Ãˆœ˜>]Ê LiV>ÕÃiʈÌʜÀˆ}ˆ˜>Ìi`Ê>ÌÊ>ÊȘ}iʓœ“i˜Ìʈ˜Ê̈“iÊ­Ü i˜ÊÌ iÊ Õ˜ˆÛiÀÃiÊVœœi`ÊLiœÜÊÎ]äääʎiÛˆ˜Ã®]ÊÌ iÊÓ£‡Vi˜Ìˆ“iÌiÀʓ>«]Ê >Ãʓi˜Ìˆœ˜i`Ê>LœÛi]Ê܈ÊLiÊvՏÞÊÌ Àii‡`ˆ“i˜Ãˆœ˜>°Ê-iVœ˜`]Ê Epoch of Reionization/Cosmic Dawn Science with SKA Ì iʓˆVÀœÜ>ÛiÊL>VŽ}ÀœÕ˜`ʈÃÊܓiÜ >ÌÊLÕÀÀÞÊLiV>ÕÃiʈÌÃÊ Àii>ÃiÊ`ˆ`ʘœÌʜVVÕÀÊ>ÌÊÌ iÊÃ>“iÊ̈“iÊiÛiÀÞÜ iÀi°Ê/ iÊ՘ˆ‡ ÛiÀÃiÊÜi˜ÌÊÌ ÀœÕ} Ê>Ê«iÀˆœ`ÊÜ i˜ÊˆÌÊÜ>ÃʘiˆÌ iÀÊvՏÞʜ«>µÕiÊ /ˆ“i\Ê Ó£äʓˆˆœ˜ÊÞi>Àà әäʓˆˆœ˜ÊÞi>Àà ÎÇäʓˆˆœ˜ÊÞi>Àà {Èäʓˆˆœ˜ÊÞi>Àà x{äʓˆˆœ˜ÊÞi>Àà ÈÓäʓˆˆœ˜ÊÞi>Àà ǣäʓˆˆœ˜ÊÞi>Àà 7ˆ`Ì ÊœvÊvÀ>“i\Ê Ó°{ʓˆˆœ˜Êˆ} ̇Þi>Àà ΰäʓˆˆœ˜Êˆ} ̇Þi>Àà ΰÈʓˆˆœ˜Êˆ} ̇Þi>Àà {°£Ê“ˆˆœ˜Êˆ} ̇Þi>Àà {°Èʓˆˆœ˜Êˆ} ̇Þi>Àà x°äʓˆˆœ˜Êˆ} ̇Þi>Àà x°xʓˆˆœ˜Êˆ} ̇Þi>Àà ˜œÀÊvՏÞÊÌÀ>˜Ã«>Ài˜Ì]ʏˆŽiÊ>Êvœ}ÊÌ >ÌÊ`ˆÃÈ«>Ìi`Ê}À>`Õ>Þ°Ê ÕÀ‡ "LÃiÀÛi`ÊÜ>Ûii˜}Ì \Ê {°£Ê“iÌiÀà JonathanΰÎʓiÌiÀà PritchardÓ°nʓiÌiÀà ! Ó°{ʓiÌiÀà Ӱ£Ê“iÌiÀà ӰäʓiÌiÀà £°nʓiÌiÀà ˆ˜}ÊÌ >ÌÊ̈“i]ÊÌ iÊÀ>`ˆ>̈œ˜Ê`ˆvvÕÃi`Ê>VÀœÃÃÊà œÀÌÊ`ˆÃÌ>˜ViÊÃV>iÃ]Ê
ÊÌ iÊ}>ÃʈÃʘiÕÌÀ>° >ˆ˜ÌÊÀi`Ê«>ÌV ià / iÃiÊLÕLLiÃʜv iÜÊÃÌ>ÀÃÊ>˜` / iÊLÕLLiÃÊ>Ài / iÊLÕLLiÃÊ >ÛiÊ / iʜ˜ÞÊÀi“>ˆ˜ˆ˜}Ê Ã“i>Àˆ˜}ÊÌ iÊw˜iÊ«Àˆ˜Ìʈ˜ÊÌ iʓˆVÀœÜ>ÛiÊL>VŽ}ÀœÕ˜`ÊÎްÊ˜Ê / iÊÜ ˆÌiÊ>Ài>ÃÊ>ÀiÊ Ã œÜÊÌ >ÌÊÌ iÊÃÌ>ÀÃÊ ˆœ˜ˆâi`Ê}>ÃÊ}ÀœÜ° µÕ>Ã>ÀÃÊvœÀ“Ê>˜`Ê Li}ˆ˜˜ˆ˜}ÊÌœÊ “iÀ}i`Ê>˜`ʘi>ÀÞÊ ˜iÕÌÀ>Ê Þ`Àœ}i˜ÊÊ Ì iÊ`i˜ÃiÃÌÊ>˜`Ê܈Ê >˜`ʵÕ>Ã>ÀÃÊ >ÛiÊ VÀi>ÌiÊÌ iˆÀÊœÜ˜Ê ˆ˜ÌiÀVœ˜˜iVÌ° Ì>Ži˜ÊœÛiÀÊ>ÊœvÊë>Vi° ˆÃÊVœ˜Vi˜ÌÀ>Ìi`ÊÊ Vœ˜ÌÀ>ÃÌ]ÊÜ i˜ÊÓ£‡Vi˜Ìˆ“iÌiÀÊÀ>`ˆ>̈œ˜Êi“iÀ}i`ÊvÀœ“Ê Þ`Àœ‡ -ˆ“Տ>Ìi`ʈ“>}iÃʜvÊÓ£‡Vi˜Ìˆ“iÌiÀÊÀ>`ˆ>̈œ˜Êà œÜÊ œÜÊ Þ`Àœ}i˜Ê }ˆÛiÊÀˆÃiÊ̜ÊCo-chairÌ iÊwÀÃÌÊ Li}՘ÊÌœÊ ˆœEoR-SWG˜ˆâiÊÌ iÊ LÕLLi𠈘Ê}>>݈ið }i˜Ê>̜“Ã]ʘœÌ ˆ˜}ÊLœVŽi`ʈÌÃÊ«Àœ«>}>̈œ˜ÊÌ ÀœÕ} Êë>Vi]ÊÃœÊ }>ÃÊÌÕÀ˜Ãʈ˜ÌœÊ>Ê}>>ÝÞÊVÕÃÌiÀ°Ê/ iÊ>“œÕ˜ÌʜvÊÀ>`ˆ>̈œ˜Ê­Ü ˆÌiʈÃÊ ÃÌ>ÀÃÊ>˜`ʵÕ>Ã>Àð }>ÃÊ>ÀœÕ˜`ÊÌ i“° ˆ} iÃÌÆʜÀ>˜}iÊ>˜`ÊÀi`Ê>Àiʈ˜ÌiÀ“i`ˆ>ÌiÆÊL>VŽÊˆÃʏi>ÃÌ®ÊÀiyiVÌÃÊ ˆÌÊÌÀ>ViÃÊÌ iÊ}>ÃÊ`ˆÃÌÀˆLṎœ˜ÊÜˆÌ œÕÌÊÃÕV ÊLÕÀÀˆ˜}°Ê/ ˆÀ`]ÊÌ iÊ LœÌ ÊÌ iÊ`i˜ÃˆÌÞʜvÊÌ iÊ}>ÃÊ>˜`ʈÌÃÊ`i}Àiiʜvʈœ˜ˆâ>̈œ˜\Ê`i˜Ãi]Ê “ˆVÀœÜ>ÛiÊL>VŽ}ÀœÕ˜`ÊV>ÀÀˆiÃʈ˜vœÀ“>̈œ˜Ê>LœÕÌÊÌ iʓ>ÌÌiÀÊ iiVÌÀˆV>ÞʘiÕÌÀ>Ê}>ÃÊ>««i>ÀÃÊÜ ˆÌiÆÊ`i˜Ãi]ʈœ˜ˆâi`Ê}>ÃÊ>««i>ÀÃÊ `i˜ÃˆÌÞÊyÕVÌÕ>̈œ˜ÃÊÌ >ÌÊÃii`i`Ê}>>݈iÃ]ÊÜ iÀi>ÃÊÌ iÊÓ£‡Vi˜‡ L>VŽ°Ê/ iʈ“>}iÃÊ >ÛiÊLii˜ÊÀiÃV>i`Ê̜ÊÀi“œÛiÊÌ iÊivviVÌʜvÊVœÃ“ˆVÊ iÝ«>˜Ãˆœ˜Ê>˜`ÊÌ ÕÃÊ ˆ} ˆ} ÌÊÌ iÊVÕÃÌiÀ‡vœÀ“ˆ˜}Ê«ÀœViÃÃiÃ°Ê Ìˆ“iÌiÀʓ>«Ê܈Ê`i«ˆVÌÊLœÌ ÊÌ iÊÃii`ÃʜvÊ}>>݈iÃÊ>˜`ÊÌ iÊivviVÌÊ iV>ÕÃiʜvÊiÝ«>˜Ãˆœ˜]ÊÌ iÊÓ£‡Vi˜Ìˆ“iÌiÀÊÀ>`ˆ>̈œ˜ÊˆÃÊ>VÌÕ>ÞÊ Ì >ÌÊÌ iÊ}>>݈iÃ]ʜ˜ViÊvœÀ“i`]Ê >`ʜ˜ÊÌ iˆÀÊÃÕÀÀœÕ˜`ˆ˜}ð œLÃiÀÛi`Ê>ÌÊ>ʏœ˜}iÀÊÜ>Ûii˜}Ì ÆÊÌ iÊi>ÀˆiÀÊÌ iʈ“>}i]ÊÌ iʏœ˜}iÀÊ UK EoR-SWG members:! /œÊ`iÌiVÌÊÌ iÊÓ£‡Vi˜Ìˆ“iÌiÀÊÈ}˜>]ʜLÃiÀÛiÀÃÊ܈Ê >ÛiÊÌœÊ Ì iÊÜ>Ûii˜}Ì ° œÛiÀVœ“iʘՓiÀœÕÃÊV >i˜}iðʜ܇vÀiµÕi˜VÞÊÀ>`ˆœÊLÀœ>`‡ Filipe Abdalla (UCL)! V>ÃÌÃʜ˜Êi>ÀÌ Ê >ÛiÊ̜ÊLiÊwÌiÀi`ʜÕÌ°Ê Ûi˜Ê“œÀiÊ`ˆvwVՏÌÊ܈Ê LiÊ`i>ˆ˜}ÊÜˆÌ ÊvœÀi}ÀœÕ˜`ÊÀ>`ˆœÊi“ˆÃȜ˜ÊvÀœ“ÊœÕÀÊ}>>ÝÞ]Ê ÀiÃEpochi˜Ì>̈ÛiÊÃÌ>Ì ˆÃof̈V> ÊReionizationÃ>“«iʜvʜÕÀÊ՘ˆÛiÀÃiÊ >Science˜`ÊÜˆÌ Ê ˆ} Ê Working Group}ÀœÜÌ Êë Õ ÀÌÃ]ÊÌ iÊ>VVÀï˜}Ê}>ÃÊà ˆ˜iÃʓÕV ʓœÀiÊLÀˆ} ̏ÞÊ Anna " , ÊBonaldi/ " Ê 8 *  " , (Manchester)!

Ü ˆV ʈÃÊ£ä]äääÊ̈“iÃʓœÀiʈ˜Ìi˜ÃiÊÌ >˜ÊÌ iÊÈ}˜>ÊvÀœ“ÊÌ iÊ i˜œÕ} ÊÀi܏Ṏœ˜Ê̜ÊV>«ÌÕÀiÊ`Ü>ÀvÊ}>>݈iðÊ/ iÊȓՏ>̈œ˜Ê ® Ì >˜ÊÌ iÊi˜ÌˆÀiÊÀiÃÌʜvÊÌ iÊ}>>ÝÞ]Ê«Àœ`ÕVˆ˜}Ê>ʵÕ>Ã>À°Ê/ iÊ i>ÃÕÀˆ˜}ÊÌ iÊ-“>‡-V>iÊ*œÜiÀÊ-«iVÌÀՓʜvÊ œÃ“ˆVÊ i˜ÃˆÌÞÊ Ã

Chair: Leon Koopmans (Groningen)˜ ! Emma Chapman (UCL)! œ ÕVÌÕ>̈œ˜ÃÊÌ ÀœÕ} ÊÓ£ÊV“Ê/œ“œ}À>« ÞÊ*ÀˆœÀÊ̜ÊÌ iÊ «œV ÊÊ i«œV ʜvÊÀiˆœ˜ˆâ>̈œ˜°ÊœÀÌ՘>ÌiÞ]ÊÌ iÊ}>>V̈VʘœˆÃiʈÃÊÀœÕ} ‡ >ÃœÊ˜ii`ÃÊ̜ÊÌÀ>ViÊÌ iÊ«Àœ«>}>̈œ˜ÊœvÊÌ iʈœ˜ˆâˆ˜}ÊÀ>`ˆ>̈œ˜Ê ˆ -œ>˜Ê ˆ}ˆÌ>Ê-ŽÞÊ-ÕÀÛiÞÊ >ÃÊÀiÛi>i`ÊÌ >ÌʵÕ>Ã>ÀÃÊÜˆÌ ÊL>VŽÊ Ì Ê Ê >

 œvÊ-ÌÀÕVÌÕÀiÊœÀ“>̈œ˜°Ê
LÀ> >“ÊœiLÊ>˜`Ê >̈>ÃÊ<>`>ÀÀˆ>}>ʈ˜Ê

ÞÊÌ iÊÃ>“iÊ>ÌÊψ} ̏ÞÊ`ˆvviÀi˜ÌÊÜ>Ûii˜}Ì Ã]ÊÜ iÀi>ÃÊÌ iÊÈ}‡ vÀœ“ÊÌ iÊ}>>݈iÃÊÌ ÀœÕ} ÊÌ iÊÃÕÀÀœÕ˜`ˆ˜}Ê}>Ã]Ê>Ê«ÀœViÃÃʓœ`‡ Õ œiÃʜvʓœÀiÊÌ >˜Ê>ÊLˆˆœ˜Ê܏>Àʓ>ÃÃiÃÊ>Ài>`ÞÊi݈ÃÌi`Ê>ÌÊ>Ê

Ê Ê “
Æ Ê Co-chairs: Garrelt Mellema (Stockholm)ˆ ! Ilian* ÞÈ V>IlievÊ,iۈiÜÊ i(Sussex)ÌÌiÀÃ]Ê6œ°Ê™Ó]Ê œ°ÊÓ£!]Ê*>«iÀÊ œ°ÊÓ££Îä£ÆÊ >ÞÊÓx]ÊÓää{°Ê ® à ˜>ÊyÕVÌÕ>ÌiÃÊÜˆÌ ÊÜ>Ûii˜}Ì ]ÊÀiyiV̈˜}ÊÌ iÊë>̈>ÊÃÌÀÕVÌÕÀiÊ ii`ʜ˜ÞÊÛiÀÞÊVÀÕ`iÞÊÜÊv>À°Ê"LÃiÀÛiÀÃʓ>ÞÊÜiÊÃiiÊÀiˆœ˜ˆâ>‡ VœÃ“ˆVÊ>}iʜvʜ˜iÊLˆˆœ˜ÊÞi>ÀðÊœÜÊ`ˆ`ÊÃÕV ʓ>ÃÈÛiÊL>VŽÊ ­ ˜
Ê  œ *Ài«Àˆ˜ÌÊ>Û>ˆ>LiÊ>ÌÊ>À݈۰œÀ}É>LÃÉ>ÃÌÀœ‡« ÉäΣӣÎ{ Þ ˆ - Ì Ì ˆ

œvÊÌ iʈœ˜ˆâi`ÊLÕLLiðÊ/ ˆÃÊ`ˆvviÀi˜Viʓ>ŽiÃʈÌÊ«œÃÈLiÊÌœÊ Ìˆœ˜ÊLivœÀiÊÌ iœÀˆÃÌÃÊ>ÀiÊ>LiÊ̜ÊvœÀiV>ÃÌÊÜ >ÌÊÌ iÞÊà œÕ`ÊÃii°
œiÃÊVœ“iÊ̜Êi݈ÃÌÊÜÊi>ÀÞ¶Ê7 ÞÊ`ˆ`ÊÌ iÞÊÃ̜«Ê}ÀœÜˆ˜}¶ > à À  À

!! Jonathan Pritchard (Imperial)Ì Mike/ iÊ-Ì> ÌiJonesʜvÊÌ iÊ1˜ˆÛi À(Oxford)Ãi°Ê*iÌiÀÊ œiÃʈ˜Ê >ÌÕÀi]Ê6œ°Ê{ÎÎ]ÊÊ " i à - iÝÌÀ>VÌÊÌ iÊÈ}˜>°Ê
ÃÌÀœ˜œ“iÀÃÊà œÕ`ÊLiÊ>LiÊ̜ÊVœ“«>ÀiÊÌ iÊ / ˆÃÊVœ“Lˆ˜i`ʜLÃiÀÛ>̈œ˜>Ê>˜`ÊÌ iœÀïV>ÊivvœÀÌÊà œÕ`Ê Û
˜œÌ iÀÊÃiÌÊVœ˜ViÀ˜ÃÊÌ iÊÈâiÊ`ˆÃÌÀˆLṎœ˜ÊœvÊ}>>݈iðÊ/ i‡ Ê Õ ˆ 

«>}iÃÊÓ{nqÓxÈÆÊ>˜Õ>ÀÞÊÓx]ÊÓääx° ˜  ˆ " 1 ­ Ê Ê

Ó£‡Vi˜Ìˆ“iÌiÀʓ>«ÃÊÜˆÌ Êˆ“>}iÃÊvÀœ“ʈ˜ÃÌÀՓi˜ÌÃÊÃÕV Ê>ÃÊ Ã i`ʏˆ} Ìʜ˜ÊÛ>ÀˆœÕÃʓÞÃÌiÀˆiÃÊÌ >ÌʘœÜÊ«>}ÕiÊÌ iÊÌ iœÀÞʜvÊ , œÀˆÃÌÃÊLiˆiÛiÊÌ >ÌÊÌ iÊՏÌÀ>ۈœiÌÊÀ>`ˆ>̈œ˜Ê«Àœ`ÕVi`ÊLÞÊ`Ü>ÀvÊ `

À

 ˆÀÃÌʈ} Ì°Ê
LÀ> >“ÊœiL°ÊiVÌÕÀiʘœÌiÃÊvœÀÊÌ iÊ-

-‡iiÊ7ˆ˜ÌiÀÊ
> 6 Ê ] 7-/°Ê/ iÊ}>>݈iÃÊÃii˜Êˆ˜Êˆ˜vÀ>Ài`ʏˆ} ÌÊà œÕ`ÊVœÀÀi>ÌiÊÜˆÌ Ê }>>ÝÞÊvœÀ“>̈œ˜°Ê"˜iÊÃiÌʜvʵÕiÃ̈œ˜ÃÊVœ˜ViÀ˜ÃÊÌ iʓ>ÃÈÛiÊ Û }>>݈iÃÊ`ÕÀˆ˜}ÊÌ iÊi«œV ʜvÊÀiˆœ˜ˆâ>̈œ˜Ê i>Ìi`ÊÌ iÊVœÃ“ˆVÊ

À " -V œœ]Ê
«ÀˆÊÓääÈ°Ê>À݈۰œÀ}É>LÃÉ>ÃÌÀœ‡« ÉäÈäÎÎÈä

> / " /  Ê

ˆœ˜ˆâi`ÊLÕLLiÃʈ˜ÊÌ iʘiÕÌÀ>Ê Þ`Àœ}i˜° L>VŽÊ œiÃʈ˜ÊÌ iÊVi˜ÌiÀÃʜvÊ}>>݈iðÊ"ÛiÀÊÌ iÊ«>ÃÌÊ`iV>`iÊ>Ç * }>ÃÊ>˜`ÊÃÕ««ÀiÃÃi`ÊÌ iÊvœÀ“>̈œ˜ÊœvʘiÜʏœÜ‡“>ÃÃÊ}>>݈iðÊ

Ê / -

>Ș}ÊÕLLi½ÃÊ- >`œÜÃ\Ê/ iÊ-i>ÀV ÊvœÀÊ>>݈iÃÊ>ÌÊÌ iÊ `}iÊÊ -   See white
paper: Mellema+ 2013 (arXiv:1210.0197) " 1

˜Ê>``ˆÌˆœ˜Ê̜ÊÌ iÊ>LœÛiʜLÃiÀÛ>̈œ˜>ÊV >i˜}iÃ]Ê>ʘՓ‡ ÌÀœ˜œ“iÀÃÊ >ÛiÊÀi>ˆâi`ÊÌ >ÌÊ>“œÃÌÊiÛiÀÞÊ}>>ÝÞʈ˜ÊÌ iÊ«ÀiÇ  œÜÊ`ˆ`ÊÌ ˆÃÊÃÕ««ÀiÃȜ˜Ê՘vœ`ʜÛiÀÊ̈“i¶Ê7 ˆV ʜvÊÌ iÊ

œvÊ/ˆ“i°ÊivvÊ>˜ˆ«i°ÊˆÊ>˜`Ê7>˜}]ÊÓääÈ° + ,

1
, LiÀʜvÊÌ>ÎÃÊÀi“>ˆ˜ÊvœÀÊÌ iœÀˆÃÌÃ°Ê œÃÌʈ“«œÀÌ>˜Ì]ÊÌ iÞʘii`Ê i˜Ì‡`>ÞÊ՘ˆÛiÀÃi]ʈ˜VÕ`ˆ˜}ʜÕÀÊœÜ˜Ê ˆŽÞÊ7>Þ]Ê œÃÌÃÊ>ʓ>Ç  `Ü>ÀvÊ}>>݈iÃÊÜiÊw˜`Ê̜`>ÞÊÜiÀiÊ>Ài>`Þʈ˜Êi݈ÃÌi˜ViÊ>ÌÊÌ iÊ Ê , œÃ“œœ}ÞÊ>ÌÊœÜÊÀiµÕi˜VˆiÃ\Ê/ iÊÓ£ÊV“Ê/À>˜ÃˆÌˆœ˜Ê>˜`ÊÌ iʈ} ‡

  ‡ Ê

-

̜ÊÀ՘ÊLˆ}}iÀÊVœ“«ÕÌiÀÊȓՏ>̈œ˜ÃÊ̜ÊÌÀ>VŽÊiÛi˜ÌÃʈ˜Ê>ÊۜÕ“iÊ ÃˆÛiÊL>VŽÊ œi°Ê/ iÃiÊ œiÃÊ>ÀiÊLiˆiÛi`Ê̜ÊLiÊvi`ÊÜˆÌ Ê}>Ãʈ˜Ê 6 Li}ˆ˜˜ˆ˜}¶Ê/ iÃiÊ>Àiʜ˜ÞÊ>ÊviÜʜvÊÌ iʓ>˜ÞʵÕiÃ̈œ˜ÃÊÜ œÃiÊ ,i`à ˆvÌÊ1˜ˆÛiÀÃi°Ê-ÌiÛi˜ÊÕÀ>˜iÌ̜]Ê-°Ê*i˜}Ê" Ê>˜`ÊÀ>˜ŽÊ Àˆ}}Ãʈ˜Ê ,

/

-  œvÊë>Viʏ>À}iÊi˜œÕ} Ê­>ÊLˆˆœ˜Êˆ} ̇Þi>ÀÃÊ>VÀœÃîÊ̜ÊLiÊ>ÊÀi«‡ i«ˆÃœ`ˆVÊiÛi˜ÌÃ]ÊÌÀˆ}}iÀi`ÊLÞʓiÀ}iÀÃʜvÊ}>>݈iÃ°Ê ÕÀˆ˜}ÊÌ iÃiÊ  >˜ÃÜiÀÃʏˆiʈ˜ÊÌ iÊ >ÀŽÊ
}iÃ°Ê * ÞÈVÃÊ,i«œÀÌÃÊ­vœÀÌ Vœ“ˆ˜}®°Ê>À݈۰œÀ}É>LÃÉ>ÃÌÀœ‡« ÉäÈänäÎÓ xÓÊ -  / Ê
,
Ê "6 ,ÊÓääÈ Ü Ü Ü°ÃVˆ>“°Vœ“ÊÊ -  / Ê
,
Ê xÎ SKA-LOW Western Australia targets EoR signal at 50MHz-250MHz

UK-SKA Birmingham 2014 Jonathan Pritchard The first billion years

Reionization marks the limits of current CMB observations Dark ages

Cosmic Dawn

Reionization

Galaxy ! formation

UK-SKA Birmingham 2014 Jonathan Pritchard The first billion years

Reionization marks the limits of current CMB observations

Dark ages When did the first galaxies form? Cosmic Dawn When did the first ! Reionization black holes form? How did reionization Galaxy ! proceed? formation How do galaxies form and evolve?

UK-SKA Birmingham 2014 Jonathan Pritchard More needed...

Age of Universe= QSO

CMB ∫

HUDF

redshift= Existing observations leaves much unanswered Possible hints of neutral hydrogen at z~7, e.g. z=7 QSO, LAE/LBG ratio

By 2020: possible advances... 1) Planck polarisation could constrain redshift and duration of reionization! 2) HST+JWST will have observed bright end of luminosity function to z~12 (faint end will still be incomplete; connection to ionizing photons may still be unclear)! 3) Little advance in QSO (more at z~7) - wait for Euclid in 2020 to push to z~8! 4) LAE surveys into EoR will be more advanced (HSC) - maybe clustering => patchy reionization? SKA will map out details of reionization and cosmic dawn

UK-SKA Birmingham 2014 Jonathan Pritchard More than reionization ! • 21 cm fluctuations contain wealth of information! - Lyman alpha fluctuations => star formation rate and first galaxies! - Temperature fluctuations => X-ray sources and first black holes! - Neutral fraction fluctuations => topology of reionization! - Density fluctuations => cosmology

First galaxies Reionization! Reionization! Dark Ages begins ends

Brightness [mK] Brightness Heating begins 50 100 150 200

PritchardUK-SKA &Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) More than reionization ! • 21 cm fluctuations contain wealth of information! - Lyman alpha fluctuations => star formation rate and first galaxies! - Temperature fluctuations => X-ray sources and first black holes! - Neutral fraction fluctuations => topology of reionization! - Density fluctuations => cosmology

Dark! Ages

First galaxies Reionization! Reionization! Dark Ages begins ends

Brightness [mK] Brightness Heating begins 50 100 150 200

PritchardUK-SKA &Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) More than reionization ! • 21 cm fluctuations contain wealth of information! - Lyman alpha fluctuations => star formation rate and first galaxies! - Temperature fluctuations => X-ray sources and first black holes! - Neutral fraction fluctuations => topology of reionization! - Density fluctuations => cosmology

Dark! Cosmic! Ages Dawn

First galaxies Reionization! Reionization! Dark Ages begins ends

Brightness [mK] Brightness Heating begins 50 100 150 200

PritchardUK-SKA &Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) More than reionization ! • 21 cm fluctuations contain wealth of information! - Lyman alpha fluctuations => star formation rate and first galaxies! - Temperature fluctuations => X-ray sources and first black holes! - Neutral fraction fluctuations => topology of reionization! - Density fluctuations => cosmology

Dark! Cosmic! Reionization Ages Dawn

First galaxies Reionization! Reionization! Dark Ages begins ends

Brightness [mK] Brightness Heating begins 50 100 150 200

PritchardUK-SKA &Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) Time evolution of signal 1 cMpc ~ 1deg SKA FOV~ 20deg2

Santos+ 2008

UK-SKA Birmingham 2014 Jonathan Pritchard Time evolution of signal 1 cMpc ~ 1deg SKA FOV~ 20deg2

Santos+ 2008

UK-SKA Birmingham 2014 Jonathan Pritchard STScI MAR SKA will image reionization 25 2009 Simulation + Lya +X-rays SKA will be first instrument with sensitivity for imaging! => map topology of reionization Mellema+ 2013 ~1deg Numerical simulations of baryons + dark matter with radiative 10’ 1’ transfer needed for reionization Iliev+ 2014

Fast approximate schemes being developed:! - Santos+ 2009 “Fast21CM”! - Baek+ 2008, 2010 - Mesinger+ 2010 “21cmFast”! - Thomas+ 2010 “BEARS”

Detailed numerical simulation including spin temperature needed:Lya &! T fluctuations -can Baek+ be 2008, important 2010 Santos, Amblard, Pritchard+ 2008 UK-SKA Birmingham 2014 Jonathan Pritchard Santos, Amblard, JRP, Trac, Cen, Cooray 2008 Evolution of the power spectrum Mesinger+ 2010

10 mK

1 mK

~10’

Measure power spectrum from z=27 to z~6 ! =>traces onset of star formation and IGM heating

UK-SKA Birmingham 2014 Jonathan Pritchard Evolution of the power spectrum Mesinger+ 2010

10 mK

1 mK

~10’

Measure power spectrum from z=27 to z~6 ! =>traces onset of star formation and IGM heating

UK-SKA Birmingham 2014 Jonathan Pritchard First galaxies

Pop II vs Pop III stars - onset of metal enrichment! ! Sterilisation of H2 channel with Lyman-Werner photons! ! Sinks of ionising radiation! ! Star formation rate - Lyman alpha production! ! Black holes - X-ray emission

Complementarity with other observations e.g. HST, JWST, LAEs, ALMA + SKA => galaxies and IGM

UK-SKA Birmingham 2014 Jonathan Pritchard The Astrophysical Journal,736:147(5pp),2011August1 Greif et al.

MH-1-NOREL MH-2-NOREL MH-3-NOREL

The Astrophysical Journal,736:147(5pp),2011August1 Greif et al. z = 19.58 z = 22.67 z = 31.73 MH-1-NORELHow theMH-2-NOREL windMH-1-REL blows?MH-3-NORELMH-2-REL MH-3-REL

Recombination leads to sudden drop in sound speed Tseliakhovich ! => coherent supersonic relative motion of baryons and dark matter & Hirata 2010

The Astrophysical Journal,736:147(5pp),2011August1 Greif et al. No-rel: galaxy forms at z~20 z Rel:= 19.58 snapshot at z~20 z Rel:= z22.67 =gal 19.58 formation delayed to z~16 z = z31.73Galaxy = 22.67 formation in lowz = 31.73 mass 8 MH-1-NOREL MH-2-NORELMH-1-REL MH-3-NORELMH-2-RELMH-1-REL MH-3-RELMH-2-REL<10 Msol halos delayedMH-3-REL

Little effect on high mass halos => importance of effect decreases at late times Maio+2010, Greif+2011, z = 19.58 z = z22.67 = 19.58 z = z31.73 = z22.67 = 15.65 z = z31.73 = 19.48 Stacey+2011 z = 26.06 MH-1-REL MH-2-RELMH-1-REL MH-3-RELMH-2-REL MH-3-REL flow Greif+ 2011 T [K] Side Length: 10 kpc (comoving) 10 100 1000

1 360 360Figure 1. Comparison of three statistically independent minihalos with no streaming velocity (top panels), and with an initial streaming velocity of 3 km s− applied 40 at z 99 from left to right (middle and bottom40 panels). WeCoherence show the density-squared of velocity weighted gas temperaturefield leads projected along the line of sight when the hydrogen 300 300 = 9 3 density in the center has just exceeded nH 10 cm− (top and bottom panels), and when the streaming case has evolved to the same redshift as the no-streaming case 20 (middle panels). In the presence of streaming= 20 velocities, theto effective boost Jeans in mass 21cm of the gas fluctuations is increased. The underlying! DM halo therefore becomes more massive 240 240before the gas can cool, which delays the onset of collapse. We also find that virial shocks are more pronounced in the direction of the incoming streamingflowthan 0 0 => much more detectable signal Tb in other directions. Nonlinear effects of this sort mayTb result in a higher velocity dispersion of the gas (see also Figure 4). 180 180 −20 (A color version of this figure is available in− the20 online journal.)+ enhanced BAO signature z = 19.58 z = z22.67 = 15.65 z = z31.73 = 19.48 z = 26.06 120 120 −40 −40 MH-1-REL MH-2-REL MH-3-REL1 8 60 −60 60greater than 1.5kms− at z 99,− which60 we considerT [K] a lower forms a Pop III star. We set Mmax 10 M ,butnotethatour = = ⊙ Side Length:limit 10 for kpc the (comoving) above delay to be significant, may be found by results are not sensititive to this parameter, since massive halos

60 120 180 240 300 360 integrating60 120 the180 above240 function300 360 from10σ/2 σ1d√ 1003/2toinfinity, 1000 are rare. As shown in Figure 5,thenumberofminihalosthat = IF star formation in low mass which yields approximately 0.86. This shows that our results cool and form stars is1 reduced by up to an order of magnitude in WithFigure 1. vComparison of three statisticallyz=20 independentWithout minihalos with nov streaming velocity (top panels), and with an initial streaming velocity of 3 km s− applied at z 99 from left to right (middle and bottommay panels). be considered We show the representative density-squared for weighted most gas of the temperaturehalos volume projected ofimportant the along the line presence of sight ofwhen streaming the hydrogen velocities. Such a large effect implies Visbal+ 2012= Figure 3: The 21-cm brightness9 temperature.3 density in the center has just exceeded nH universe.10 cm (top and bottom panels), and when the streaming case has evolved to the samethat redshift streaming as the no-streaming velocities shouldcase be taken into account when the = − UK-SKA Birmingham(middle panels). 2014 In the presence of streamingThe velocities, cosmological the effective Jeans number mass of density the gas is of increased. minihalos The underlying hosting DM haloinfluence thereforeJonathan becomesof the first Pritchard more stars massive on observables is investigated. before the gas can cool, which delays the onsetPop of III collapse. stars We may also then find that be virialestimated shocks usingare more the pronounced Sheth–Tormen in the direction of the incoming streamingflowthan in other directions. Nonlinear effects of this sort may result in a higher velocity dispersion of the gas (see also Figure 4). (Sheth et al. 2001)massfunction: (A color version of this figure is available in the online journal.) 4. DISCUSSION z = 15.65 z = 19.48 z = 26.06 Mmax We have found that supersonic streaming velocities between 1 nmh(z) nst(M,z) dM, (4) 8 greater than 1.5kms− at z 99, which we considerT [K] a lower= M forms a Pop III star. We set Mmax the10 DMM and,butnotethatour gas substantially delay the onset of gravitational = ! min = ⊙ Side Length:limit 10 for kpc the (comoving) above delay to be significant, may be found by results are not sensititive to this parameter,collapse since in minihalos. massive halos The virial mass required for efficient 5 integrating the above function from10whereσ/2 weσ set√ 100M3/min2toinfinity,1.5 10 1000M arefor rare. the case As shown of no streaming in Figure 5,thenumberofminihalosthatcooling is increased by a factor of 3, which results in 1d = × 5 ⊙ ≃ velocities= and Mmin 5 10 Mcoolfor and the form case stars of a isuniversal reduced by upan to average an order of delay magnitude of Pop in III star formation by ∆z 4. Figure 1. Comparison of three statisticallywhich independent yields minihalos approximately with no streaming 0.86. velocity This shows (top panels), that and our with= results an× initial streaming⊙ velocity of 3 km s 1 applied = 1σ streaming velocity, representingthe the presence factor of of streaming3increasein− velocities.Streaming Such a large velocities effect implies also enhance the buildup of turbulence at z 99 from left to right (middle and bottommay panels). be considered We show the representative density-squared for weighted most gas of the temperature volume projected of the along the line of sight when≃ the hydrogen = 9 3 minimum halo mass. The resulting number densities should be during runaway collapse, which could affect the fragmentation density in the center has just exceeded nH universe.10 cm− (top and bottom panels), and when the streaming case has evolved to the samethat redshift streaming as the no-streaming velocities shouldcase be taken into account when the (middle panels). In the presence of streaming= The velocities, cosmological the effective Jeans number mass of density theconsidered gas is of increased. minihalos upper The limits, underlying hosting since DM not halo everyinfluence therefore halo at becomesof the the low-mass first more stars massive end on observablesof the gas is investigated. and hence the mass function of the first stars. before the gas can cool, which delays the onsetPop of III collapse. stars We may also then find that be virialestimated shocks usingare more the pronounced Sheth–Tormen in the direction of the incoming streamingflowthan in other directions. Nonlinear effects of this sort may result in a higher velocity dispersion of the gas (see also Figure 4). (Sheth et al. 2001)massfunction: 3 (A color version of this figure is available in the online journal.) 4. DISCUSSION Mmax We have found that supersonic streaming velocities between nmh(z) nst(M,z) dM, (4) greater than 1.5kms 1 at z 99, which we consider a lower= forms a Pop III star. We set M the10 DM8 M and,butnotethatour gas substantially delay the onset of gravitational − Mmin13 max limit for the above delay to= be significant, may be found! by results are not sensititive to this parameter,=collapse since⊙ in minihalos. massive halos The virial mass required for efficient 5 integrating the above function fromwhereσ/2 weσ set√M3/min2toinfinity,1.5 10 M arefor rare. the case As shown of no streaming in Figure 5,thenumberofminihalosthatcooling is increased by a factor of 3, which results in 1d = × 5 ⊙ ≃ velocities= and Mmin 5 10 Mcoolfor and the form case stars of a isuniversal reduced by upan to average an order of delay magnitude of Pop in III star formation by ∆z 4. which yields approximately 0.86. This shows that our= results× ⊙ = 1σ streaming velocity, representingthe the presence factor of of streaming3increasein velocities.Streaming Such a large velocities effect implies also enhance the buildup of turbulence may be considered representative for most of the volume of the ≃ universe. minimum halo mass. The resultingthat number streaming densities velocities should should be beduring taken into runaway account collapse, when the which could affect the fragmentation The cosmological number densityconsidered of minihalos upper limits, hosting since not everyinfluence halo at of the the low-mass first stars end on observablesof the gas is investigated. and hence the mass function of the first stars. Pop III stars may then be estimated using the Sheth–Tormen 3 (Sheth et al. 2001)massfunction: 4. DISCUSSION Mmax We have found that supersonic streaming velocities between nmh(z) nst(M,z) dM, (4) = M the DM and gas substantially delay the onset of gravitational ! min collapse in minihalos. The virial mass required for efficient 5 where we set Mmin 1.5 10 M for the case of no streaming cooling is increased by a factor of 3, which results in = × 5 ⊙ ≃ velocities and Mmin 5 10 M for the case of a universal an average delay of Pop III star formation by ∆z 4. 1σ streaming velocity,= representing× ⊙ the factor of 3increasein Streaming velocities also enhance the buildup of turbulence= minimum halo mass. The resulting number densities≃ should be during runaway collapse, which could affect the fragmentation considered upper limits, since not every halo at the low-mass end of the gas and hence the mass function of the first stars.

3 EoR/CD Cosmology Cosmology Pritchard SKA-LOW probes new ! To get cosmology must first volume of universe disentangle astrophysics

Cosmology requires some degree of inventiveness:! 1) Infer density field directly (avoid + model astro, RSD)! 2) Heating driven by exotic sources (DM annihilation,primordial BH, ...)! 3) Impact of cosmology on sources ! (non-Gaussianity, WDM, ...)! 4) Weak lensing (map DM)! 5) Other …!

UK-SKA Birmingham 2014 Jonathan Pritchard Figure 1: Illustration of the volume probed by SKA temperature fluctuations may be sourced by variation in the spin temperature and neutral fraction in addition to the density field.

1/2 1 T T 1 + z ∂ v dT = 27x (1 + d ) S CMB r r mK (2.1) B HI b T 10 (1 + z)H(z) ✓ S ◆✓ ◆  Equation 2.1 shows how these different terms come into play. In a regime where T T and S CMB xH = 1 then dTB will be an unbiased tracer of the density field. At all other times the effects of astrophysics must be modelled and removed or somehow avoided. We will return to a discussion of this point in §7 as this is a critical point. In this section, we take the optimistic view that there will a regime in which dTb µ (1 + d) so that the 21cm signal provides a clean measurement of the density field. This approach enables us to evaluate the best case scenario for SKA in measuring cosmological parameters. By comparing this to galaxy surveys we get a sense of how competitive SKA could be, if astrophysics could be overcome. The sensitivity of a radio interferometer to the 21cm power spectrum has been well studied [?, ?, ?] and we follow the same approach here. The variance of a 21 cm power spectrum estimate for a single k-mode with line of sight component k = µk is given by [?]: || 2 2 2 2 2 1 ¯ 2 2 1 D DD l sP(k, µ)= Tb P21(k, µ)+Tsys . (2.2) Nfield Btint n(k ) Ae " ? ✓ ◆ # The first term on the right-hand-side of the above expression provides the contribution from sample variance, while the second describes the thermal noise of the radio telescope. The thermal

3 26 CARILLI, GNEDIN, & OWEN Vol. 577

3.3. Simulated Spectra for Sources at z 8 and 10 ¼ Figure 5 shows a simulated spectrum at 1 kHz resolution of a z 10 radio source with a flux density of 20 mJy at an ¼ observing frequency of 120 MHz (S120). The implied lumi- nosity density at a rest-frame frequency of 151 MHz is then 35 1 1 P151 2:5 10 ergs sÀ HzÀ .Figure5a shows a spectrum covering¼ a large frequency range (100–200 MHz, or H i 21 cm redshifts of 13–6). Figure 5b shows an expanded view of the frequency range corresponding to the H i 21 cm line at the source redshift (129 MHz). The onset of H i 21 cm absorption by the neutral IGM is clearly seen at 129 MHz. The general continuum level drops by about 1% at this frequency because of the diffuse neutral IGM. Deeper narrow lines are also visible to frequencies as high as 170 MHz. Again, the narrow lines decrease in red- shift density with increasing frequency. At around 130 MHz there are roughly five narrow lines with  0:02 per unit MHz, while at 160 MHz the redshift density has decreased by a factor of 10 or so. Figure 6 shows a simulated spectrum at 1 kHz resolution of a z 8radiosourcewithS120 35 mJy, again corre- sponding¼ to P 2:5 1035 ergs s¼1 Hz 1.Thedepression 151 ¼  À À Fig. 4.—Radio spectrum of the powerful radio galaxy Cygnus A at in the continuum due to absorption by the diffuse IGM is 36 1 1 z 0:057 (P151 1:1 10 ergs sÀ HzÀ ;Baarsetal.1977).Thedashed much less evident than at higher redshift, with a mean value line¼ is a first-order¼ polynomial High fit to the (log) data,resolution corresponding to a of studies0:1%. The deep narrowof IGM lines are still easily seen but, power law of index 1:05 0:03. The solid line is a second-order polyno- again, at lower redshift density than is found at higher mial fit. À Æ 21 cm forest: HI absorption to radioredshifts. bright source spondsTraces to a first-order cosmic polynomial web fit toand the datastructure in the log of halos with3.4. Limits~kpc to Detectionresolution plane, corresponding to a power law of index 1:05 0:03. We next consider the detection limit of the absorption sig- The dashed line corresponds to a second-orderÀ polynomial.Æ nal using statistical tests. The challenge is greater at lower WeRadio use this second-order bright polynomialQSO at in the z>7? analysis below. redshifts becausez=7 of QSO the decreasingMortlock+2011 strength and redshift z>7 QSO with Euclid

never resonant reionization HI absorption

cosmic web

simulated! 20 mJy QSO! at z=10 halos

Carilli+2002 Fig. 5a Fig. 5b

Fig.UK-SKA5.—(a)Simulatedspectrumfrom100to200MHzofasourcewith Birmingham 2014 S120 20 mJy at z 10 using the Cygnus A spectral model and assumingJonathan H i 21 cmPritchard absorption by the IGM. Thermal noise has been added using the specifications¼ of the SKA and¼ assuming 10 days integration with 1 kHz wide spectral chan- nels. (b)Sameas(a)butshowinganexpandedviewofthespectralregionaroundthefrequencycorrespondingtotheredshiftHi 21 cm line at the source red- shift (129 MHz). The solid line is the Cygnus A model spectrum without noise or absorption.
  

   Foreground removal Foregrounds ~ 103 signal

&+D

Foreground removal challenging, but exploiting spectral smoothness of foregrounds seems effective ! Various techniques e.g. ICA, GMCA, … Chapman+ 2013 UK-SKA Birmingham 2014 Jonathan Pritchard Instrumental path to detection PAPER MWA LOFAR SKA-LOW

HERA (2015-2020)

Fig. 5.— The MWA (top left) and PAPER (bottom left) arrays, each currently deployed with 128 elements. The 14-m HERA2 element (right) dramatically improves sensitivity while still delivering the spectral smooth- 2 PAPER/MWA setting upper limits <(50mK)ness and stability ofcompared response that are required forwith managing foregrounds.signal The ~(few core of HERA 568mK) consists ! of a redundant hexagonal array with outrigger antennas (not shown) for imaging and foreground mitigation.

Sensitivity required for detection almostimaging achieved (MWA). Together, these (LOFAR advances enable HERA ~100hrs to achieve the science analysed goals envisioned in ~600hrs collected)! the decadal survey at a fraction of the anticipated cost. HERA follows a staged deployment in both physical construction and scientific processing. In Each step in sensitivity pushes ability toeach deploymentremove stage, improvements foregrounds are incorporated into the& system control and new science instrumental capabilities systematics are unlocked. This approach o↵ers two advantages: providing early access to science, and reducing the project risk by testing systems early and changing them incrementally. As shown in Figure 2, each stage of HERA brings an associated improvement in sensitivity that allows key aspects of current observations21-cm reionization science to be addressed. The timeline of HERA development and its associated science products is outlined below. Year 1–Infrastructure and First 37 Antennas (FY 2015). Install basic infrastructure (ground leveling, power, network connectivity) at a new site

] • PAPER10 km from the current PAPER site in the Karoo. 2 ⇠ Move existing PAPER-128 antennas, correlator, and EMC container to new site. • Install first 37 HERA antennas with existing PAPER feeds, electronics, and correlator. 21cm signal • Start developingMWA improved HERA baluns, receivers,LOFAR feeds, nodes, and in-situ antenna calibra- • tion system. Continue delay-spectrum, FHD, and optimal estimator software development. [mK

2 Year 2–Hardware Commissioning and Deep Foreground Survey (FY 2016). Commissioning observations using a hybrid array of 37 HERA antennas in a close-packed

π • hexagon surrounded by 91 PAPER antennas in an imaging configuration. Perform a polarized foreground survey using hybrid-antenna capability of FHD. Determine • on-sky beam response of HERA antennas to facilitate future source subtraction e↵orts.SKA Finalize site infrastructure (high-bandwidth optical network, surveying, trenching). • Commission new feeds, receivers, nodes, and calibration systems in Green Bank and SA. P(k)/2 •

3 Begin HERA 127 construction. •

k Year 3–HERA 127 and Detecting the Rise and Fall of Reionization (FY 2017). Complete HERA 127 construction. Science observations begin using the PAPER correlator. tint=1000hours • Apply proven delay-spectrum analysis techniques to HERA 127 observations to constrain the • timing and duration of reionization. Mellema+ 2013 k[Mpc-1] MWA-32T: Dillon+ 2013; GMRT: Paciga+ 2013; PAPER-32: Parsons+ 2013; LOFAR: Yatawatta+ 2013 UK-SKA Birmingham 2014 Jonathan Pritchard

Figure 21: Comparison of current arrays, PAPER, MWA and LOFAR, with SKA, assuming B=10 MHz, tint = 1000 hrs and k = k. For the existing arrays we assumed the latest published (or inferred) specifications, see Table 2. The black line indicates the expected power spectrum of the 21cm power signal. the same assumptions and the same scaling relations. To properly compare the different arrays, 1 we take k =0.1 cMpc as the reference point where to compare sensitivities.

PAPER and MWA: We find that the current array-configurations of PAPER and MWA perform equally well, even though PAPER has a smaller collecting area (Acoll) than MWA and a similar number of stations. The lower collecting area of PAPER is compensated by making the array even more compact than MWA, hence lowering Acore. Equation 12 shows that this improves the power spectrum sensitivity of the array. In addition, PAPER gains sensitivity by having a somewhat smaller Ae↵ , since only single dipoles are used rather than tiles. Overall this results in PAPER and MWA having similar sensitivities to the power spectrum. Both PAPER and MWA are able to probe only the smallest k modes, because of their compact configurations. We note

62 EoR/CD Cosmology Pritchard

noise depends upon the system temperature Tsys, the survey bandwidth B, the total observing time tint, the conformal distance D(z) to the center of the survey at redshift z, the depth of the survey DD, the observed wavelength l, and the effective collecting area of each antennae tile Ae. The effect of the configuration of the antennae is encoded in the number density of baselines n (k) that observe ? a mode with transverse wavenumber k [?]. Observing a number of fields Nfield further reduces the ? variance. Estimates of the error on a power spectrum measurement are calculated using the Fisher matrix 1 1/2 formalism, so that the 1 s errors on the model parameter l are (F ) , where i ij ek3V 1 ∂P ∂P = survey Tb Tb . Fij  2 2 (2.3) µ 4p sP(k, µ) ∂li ∂lj

2 2 In this equation, Vsurvey = D DD(l /Ae) denotes the effective survey volume of our radio tele- scopes and we assume wavenumber bins of width Dk = ek. We will be interested in the cases where l = P¯ and l = P 0 , P 2 , P 4 . i { Tb } i { µ µ µ }

Table 1: Low-frequency radio telescopes and their parameters. We specify the number of antennae Na, total collecting area Atot, bandwidth B, and total integration time tint for each instrument. These values are fixed at z =?? and extrapolated to other frequencies using Atot = NantNdipAdip with the number of antennae per 2 2 station Ndip = 289 and AdipStory= min(l /3,3.2m of). Universe is story of hydrogen

3 2 Array Na Atot(10 m ) B (MHz) tint (hr) Rmin(m) Rmax(km) SKA-LOW Phase 1 MWA 112 1.6 8 1000 4 0.75 LOFAR Core 48 38.6 8 1000 100 1.5 ~10 more collecting area! HERA 331 50.0 8 1000 14.3 0.3 than LOFAR SKA0 850 0.5 290 0.5 8 1000 35 2 ⇥ ⇥ SKA1 850 290 8 1000 35 2 ~10 number of stations SKA2 850 4 290 4 8 1000 35 2 ⇥ ⇥

We first illustrate the sensitivity of different iterations of SKA in Figure 2, where we take the parameters in Table 1 for SKA0 - with 50% of the SKA1 baseline collecting area, SKA1, and SKA2 - with x4 the collecting area of SKA1. For each of these we assume a filled core followed 2 by r distribution out to a maximum radius Rmax. HERA is assumed to have a uniform antennae distribution. SKA1 has 911 stations total with 899 in the core and 650 stations within a radius of 1km accounting for 75% of the total number of stations and collecting area. We limit to the innermost 850 within 2km, as the outer stations add little to the sensitivity. Physical station size is 2 2 35m. Stations have 289 antennae with antennae area Ae = l /3 giving 3.2m at 110MHz. At lower frequencies the array is densely packedFirst andgalaxies has constant collecting area, at higher frequencies the array becomes sparse. Figure 2 illustrates a few key points governing parameter constraints.Reionization Here we have! eliminated Reionization! modes whose wavelengthDark Ages exceeds the instrument bandwidth removingbegins sensitivity to the largest ends physical scales (smallest k modes). At z = 8, SKA0 is directly comparable in sensitivity to the proposed HERA experiment [2], which is more centrally concentrated to compensate for its small Brightness [mK] Brightness Heating begins 50 4 100 150 200

PritchardUK-SKA & Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) EoR/CD Cosmology Pritchard

noise depends upon the system temperature Tsys, the survey bandwidth B, the total observing time tint, the conformal distance D(z) to the center of the survey at redshift z, the depth of the survey DD, the observed wavelength l, and the effective collecting area of each antennae tile Ae. The effect of the configuration of the antennae is encoded in the number density of baselines n (k) that observe ? a mode with transverse wavenumber k [?]. Observing a number of fields Nfield further reduces the ? variance. Estimates of the error on a power spectrum measurement are calculated using the Fisher matrix 1 1/2 formalism, so that the 1 s errors on the model parameter l are (F ) , where i ij ek3V 1 ∂P ∂P = survey Tb Tb . Fij  2 2 (2.3) µ 4p sP(k, µ) ∂li ∂lj

2 2 In this equation, Vsurvey = D DD(l /Ae) denotes the effective survey volume of our radio tele- scopes and we assume wavenumber bins of width Dk = ek. We will be interested in the cases where l = P¯ and l = P 0 , P 2 , P 4 . i { Tb } i { µ µ µ }

Table 1: Low-frequency radio telescopes and their parameters. We specify the number of antennae Na, total collecting area Atot, bandwidth B, and total integration time tint for each instrument. These values are fixed at z =?? and extrapolated to other frequencies using Atot = NantNdipAdip with the number of antennae per 2 2 station Ndip = 289 and AdipStory= min(l /3,3.2m of). Universe is story of hydrogen

3 2 Array Na Atot(10 m ) B (MHz) tint (hr) Rmin(m) Rmax(km) SKA-LOW Phase 1 MWA 112 1.6 8 1000 4 0.75 LOFAR Core 48 38.6 8 1000 100 1.5 ~10 more collecting area! HERA 331 50.0 8 1000 14.3 0.3 than LOFAR SKA0 850 0.5 290 0.5 8 1000 35 2 ⇥ ⇥ SKA1 850 290 8 1000 35 2 ~10 number of stations SKA2 850 4 290 4 8 1000 35 2 ⇥ ⇥

We first illustrate the sensitivity of different iterations of SKA in Figure 2, where we take the parameters in Table 1 for SKA0 - with 50% of the SKA1 baseline collecting area, SKA1, and MWA SKA2 - with x4 the collecting area of SKA1. For each of these we assume a filled core followed 2 by r distribution out to a maximum radius Rmax. HERA is assumed to haveHERA a uniform antennaeLOFAR GMRT distribution. SKA1 has 911 stations total with 899 in the core and 650 stations within a radius of 1km accounting for 75% of the total number of stations and collecting area. We limit to thePAPER innermost 850 within 2km, as the outer stations add little to the sensitivity. Physical station size is 2 2 35m. Stations have 289 antennae with antennae area Ae = l /3 giving 3.2m at 110MHz. At lower frequencies the array is densely packedFirst andgalaxies has constant collecting area, at higher frequencies the array becomes sparse. Figure 2 illustrates a few key points governing parameter constraints.Reionization Here we have! eliminated Reionization! modes whose wavelengthDark Ages exceeds the instrument bandwidth removingbegins sensitivity to the largest ends physical scales (smallest k modes). At z = 8, SKA0 is directly comparable in sensitivity to the proposed HERA experiment [2], which is more centrally concentrated to compensate for its small Brightness [mK] Brightness Heating begins 50 4 100 150 200

PritchardUK-SKA & Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) EoR/CD Cosmology Pritchard

noise depends upon the system temperature Tsys, the survey bandwidth B, the total observing time tint, the conformal distance D(z) to the center of the survey at redshift z, the depth of the survey DD, the observed wavelength l, and the effective collecting area of each antennae tile Ae. The effect of the configuration of the antennae is encoded in the number density of baselines n (k) that observe ? a mode with transverse wavenumber k [?]. Observing a number of fields Nfield further reduces the ? variance. Estimates of the error on a power spectrum measurement are calculated using the Fisher matrix 1 1/2 formalism, so that the 1 s errors on the model parameter l are (F ) , where i ij ek3V 1 ∂P ∂P = survey Tb Tb . Fij  2 2 (2.3) µ 4p sP(k, µ) ∂li ∂lj

2 2 In this equation, Vsurvey = D DD(l /Ae) denotes the effective survey volume of our radio tele- scopes and we assume wavenumber bins of width Dk = ek. We will be interested in the cases where l = P¯ and l = P 0 , P 2 , P 4 . i { Tb } i { µ µ µ }

Table 1: Low-frequency radio telescopes and their parameters. We specify the number of antennae Na, total collecting area Atot, bandwidth B, and total integration time tint for each instrument. These values are fixed at z =?? and extrapolated to other frequencies using Atot = NantNdipAdip with the number of antennae per 2 2 station Ndip = 289 and AdipStory= min(l /3,3.2m of). Universe is story of hydrogen

3 2 Array Na Atot(10 m ) B (MHz) tint (hr) Rmin(m) Rmax(km) SKA-LOW Phase 1 MWA 112 1.6 8 1000 4 0.75 LOFAR Core 48 38.6 8 1000 100 1.5 ~10 more collecting area! HERA 331 50.0 8 1000 14.3 0.3 than LOFAR SKA0 850 0.5 290 0.5 8 1000 35 2 ⇥ ⇥ SKA1 850 290 8 1000 35 2 ~10 number of stations SKA2 850 4 290 4 8 1000 35 2 ⇥ ⇥

We first illustrate the sensitivity of different iterations of SKA in Figure 2, where we take the parameters in Table 1 for SKA0 - with 50% of the SKA1 baseline collecting area, SKA1, and MWA SKA2 - with x4 the collecting area of SKA1. For each of these we assume a filled core followed 2 by r distribution out to a maximum radius Rmax. HERASKA is assumed to haveHERA a uniform antennaeLOFAR GMRT distribution. SKA1 has 911 stations total with 899 in the core and 650 stations within a radius of 1km accounting for 75% of the total number of stations and collecting area. We limit to thePAPER innermost 850 within 2km, as the outer stations add little to the sensitivity. Physical station size is 2 2 35m. Stations have 289 antennae with antennae area Ae = l /3 giving 3.2m at 110MHz. At lower frequencies the array is densely packedFirst andgalaxies has constant collecting area, at higher frequencies the array becomes sparse. Figure 2 illustrates a few key points governing parameter constraints.Reionization Here we have! eliminated Reionization! modes whose wavelengthDark Ages exceeds the instrument bandwidth removingbegins sensitivity to the largest ends physical scales (smallest k modes). At z = 8, SKA0 is directly comparable in sensitivity to the proposed HERA experiment [2], which is more centrally concentrated to compensate for its small Brightness [mK] Brightness Heating begins 50 4 100 150 200

PritchardUK-SKA & Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) EoR/CD Cosmology Pritchard

noise depends upon the system temperature Tsys, the survey bandwidth B, the total observing time tint, the conformal distance D(z) to the center of the survey at redshift z, the depth of the survey DD, the observed wavelength l, and the effective collecting area of each antennae tile Ae. The effect of the configuration of the antennae is encoded in the number density of baselines n (k) that observe ? a mode with transverse wavenumber k [?]. Observing a number of fields Nfield further reduces the ? variance. Estimates of the error on a power spectrum measurement are calculated using the Fisher matrix 1 1/2 formalism, so that the 1 s errors on the model parameter l are (F ) , where i ij ek3V 1 ∂P ∂P = survey Tb Tb . Fij  2 2 (2.3) µ 4p sP(k, µ) ∂li ∂lj

2 2 In this equation, Vsurvey = D DD(l /Ae) denotes the effective survey volume of our radio tele- scopes and we assume wavenumber bins of width Dk = ek. We will be interested in the cases where l = P¯ and l = P 0 , P 2 , P 4 . i { Tb } i { µ µ µ }

Table 1: Low-frequency radio telescopes and their parameters. We specify the number of antennae Na, total collecting area Atot, bandwidth B, and total integration time tint for each instrument. These values are fixed at z =?? and extrapolated to other frequencies using Atot = NantNdipAdip with the number of antennae per 2 2 station Ndip = 289 and AdipStory= min(l /3,3.2m of). Universe is story of hydrogen

3 2 Array Na Atot(10 m ) B (MHz) tint (hr) Rmin(m) Rmax(km) SKA-LOW Phase 1 MWA 112 1.6 8 1000 4 0.75 LOFAR Core 48 38.6 8 1000 100 1.5 ~10 more collecting area! HERA 331 50.0 8 1000 14.3 0.3 than LOFAR SKA0 850 0.5 290 0.5 8 1000 35 2 ⇥ ⇥ SKA1 850 290 8 1000 35 2 ~10 number of stations SKA2 850 4 290 4 8 1000 35 2 ⇥ ⇥

We first illustrate the sensitivity of different iterations of SKA in Figure 2, where we take the parameters in Table 1 for SKA0 - with 50% of the SKA1 baseline collecting area, SKA1, and MWA SKA2 - with x4 theLUNAR collecting area of SKA1.! For each of these we assume a filled core followed 2 by r distribution out to a maximum radius Rmax. HERASKA is assumed to haveHERA a uniform antennaeLOFAR GMRT distribution. SKA1ARRAY has 911 stations total with 899 in the core and 650 stations within a radius of 1km accounting for 75% of the total number of stations and collecting area. We limit to thePAPER innermost 850 within 2km, as the outer stations add little to the sensitivity. Physical station size is 2 2 35m. Stations have 289 antennae with antennae area Ae = l /3 giving 3.2m at 110MHz. At lower frequencies the array is densely packedFirst andgalaxies has constant collecting area, at higher frequencies the array becomes sparse. Figure 2 illustrates a few key points governing parameter constraints.Reionization Here we have! eliminated Reionization! modes whose wavelengthDark Ages exceeds the instrument bandwidth removingbegins sensitivity to the largest ends physical scales (smallest k modes). At z = 8, SKA0 is directly comparable in sensitivity to the proposed HERA experiment [2], which is more centrally concentrated to compensate for its small Brightness [mK] Brightness Heating begins 50 4 100 150 200

PritchardUK-SKA & Birmingham Loeb 2010 2014 Frequency [MHz] ⌫ = 1420MHzJonathan/(1 Pritchard + z) Key science with SKA • SKA will help complete our understanding of cosmic history! • SKA will image ionised regions directly mapping out topology of reionization! • 21cm fluctuations from z=27 to z=6 traces evolution of first galaxies and x-ray sources! • Test cosmology with bulk flows at z>20! • Cosmology and fundamental physics from lensing, thermal history, and density fluctuations! • High resolution IGM studies with 21cm forest! • Complements UK strengths in high-z galaxy observations, galaxy formation simulation, CMB analysis, … See Mellema+ 2013 (arXiv:1210.0197) and upcoming SKA Science Case UK-SKA Birmingham 2014 Jonathan Pritchard