删除或更新信息,请邮件至freekaoyan#163.com(#换成@)

Non-parametric reconstruction of dark energy and cosmic expansion from the Pantheon compilation of t

本站小编 Free考研考试/2022-01-01

闂傚倷娴囬褏鈧稈鏅犻、娆撳冀椤撶偟鐛ラ梺鍦劋椤ㄥ懐澹曟繝姘厵闁告挆鍛闂佹娊鏀遍崹鍫曞Φ閸曨垰绠涢柛鎾茬劍閸嬔冾渻閵堝繒鍒扮€殿喖澧庨幑銏犫攽鐎n亞鍔﹀銈嗗笒鐎氼剛绮婚妷锔轰簻闁哄啠鍋撻柛搴″暱閻g兘濡烽妷銏℃杸濡炪倖姊婚悺鏂库枔濡眹浜滈柨鏂垮⒔閵嗘姊婚崒姘偓鐑芥倿閿旈敮鍋撶粭娑樻噽閻瑩鏌熼悜姗嗘畷闁稿孩顨嗛妵鍕棘閸喒鎸冮梺鍛婎殕瀹€鎼佸箖濡も偓閳藉鈻庣€n剛绐楅梻浣哥-缁垰螞閸愵喖钃熸繛鎴欏灩鍞梺闈涚箚閸撴繈鎮甸敃鈧埞鎴︽倷閹绘帗鍊悗鍏夊亾闁归棿绀侀拑鐔兼煏閸繍妲哥紒鐙欏洦鐓曟い顓熷灥閺嬬喐绻涢崼婵堝煟婵﹨娅g槐鎺懳熼悡搴樻嫛闂備胶枪缁ㄦ椽宕愬Δ鍐ㄥ灊婵炲棙鍔曠欢鐐烘煙闁箑澧版い鏃€甯″娲嚃閳圭偓瀚涢梺鍛婃尰閻╊垶鐛繝鍌楁斀閻庯綆鍋嗛崢浠嬫⒑缂佹◤顏勵嚕閼搁潧绶為柛鏇ㄥ幐閸嬫挾鎲撮崟顒傤槰闂佹寧娲忛崹浠嬪极閹扮増鍊风痪鐗埫禍楣冩煥濠靛棝顎楀ù婊冨⒔缁辨帡骞夌€n剛袦闂佸搫鐬奸崰鎰缚韫囨柣鍋呴柛鎰ㄦ櫓閳ь剙绉撮—鍐Χ閸℃ê鏆楅梺纭呮珪閹瑰洦淇婇幘顔肩闁规惌鍘介崓鐢告⒑閹勭闁稿妫濇俊瀛樼節閸屾鏂€闂佺粯锕╅崑鍕妤e啯鈷戦柛娑橈功閳藉鏌f幊閸旀垵顕i弻銉晢闁告洦鍓欓埀顒€鐖奸弻锝夊箛椤撶偟绁烽梺鎶芥敱濮婅绌辨繝鍕勃闁稿本鑹鹃~鍥⒑閸濆嫮鐒跨紒缁樼箓閻i攱绺介崜鍙夋櫇闂侀潧绻掓慨瀵哥不閹殿喚纾介柛灞剧懅閸斿秵銇勯妸銉﹀殗閽樻繈姊婚崼鐔恒€掗柡鍡檮閹便劌顫滈崱妤€浼庣紓浣瑰敾缁蹭粙婀侀梺鎸庣箓鐎氼垶顢楅悢璁垮綊鎮℃惔銏犳灎濠殿喖锕ュ钘夌暦閵婏妇绡€闁稿本绮庨幊鍡樼節绾版ɑ顫婇柛瀣噽閹广垽宕奸妷褍绁﹂梺鍦濠㈡﹢鏌嬮崶顒佺厸闁搞儮鏅涢弸鎴炵箾閸涱喚澧紒缁樼⊕濞煎繘宕滆琚f繝鐢靛仜閹锋垹绱炴担鍝ユ殾闁炽儲鏋奸崼顏堟煕椤愩倕鏋庨柍褜鍓涢弫濠氬蓟閿濆顫呴柣妯哄悁缁敻姊洪幖鐐测偓鎰板磻閹剧粯鈷掑ù锝堫潐閸嬬娀鏌涢弬璺ㄐら柟骞垮灲瀹曠喖顢橀悙鑼喊闂佽崵濮村ú銈咁嚕椤掑嫬绫嶉柛灞绢殔娴滈箖鏌ㄥ┑鍡涱€楀褌鍗抽弻銊モ槈閾忣偄顏�
547闂傚倸鍊搁崐椋庣矆娴i潻鑰块梺顒€绉查埀顒€鍊圭粋鎺斺偓锝庝簽閿涙盯姊洪悷鏉库挃缂侇噮鍨堕崺娑㈠箳濡や胶鍘遍梺鍝勬处椤ㄥ棗鈻嶉崨瀛樼厽闊浄绲奸柇顖炴煛瀹€瀣埌閾绘牠鏌嶈閸撶喖寮绘繝鍥ㄦ櫜濠㈣泛锕﹂悿鍥⒑鐟欏嫬绀冩い鏇嗗懐鐭嗛柛鎰ㄦ杺娴滄粓鐓崶銊﹀鞍妞ゃ儲绮撻弻锝夊箻鐎靛憡鍒涘┑顔硷攻濡炶棄鐣峰Δ鍛闁兼祴鏅涢崵鎺楁⒒娴e憡鎲搁柛锝冨劦瀹曟垿宕熼娑樹患闂佺粯鍨兼慨銈夊疾閹间焦鐓ラ柣鏇炲€圭€氾拷1130缂傚倸鍊搁崐鎼佸磹妞嬪海鐭嗗〒姘e亾閽樻繃銇勯弽銊х煂闁活厽鎹囬弻锝夊閵忊晜姣岄梺绋款儐閹瑰洤鐣疯ぐ鎺濇晝闁挎繂娲﹂濠氭⒒娓氣偓閳ь剛鍋涢懟顖涙櫠閸欏浜滄い鎰╁焺濡叉椽鏌涢悩璇у伐妞ゆ挸鍚嬪鍕節閸愵厾鍙戦梻鍌欑窔閳ь剛鍋涢懟顖涙櫠閹绢喗鐓涢悘鐐登规晶鑼偓鍨緲鐎氼噣鍩€椤掑﹦绉靛ù婊勭矒閿濈偞鎯旈埦鈧弨浠嬫煟閹邦垰鐨哄褎鐩弻娑㈠Ω閵壯傝檸闂佷紮绲块崗姗€寮幘缁樺亹闁肩⒈鍓﹀Σ浼存煟閻斿摜鐭婄紒缁樺笧閸掓帒鈻庨幘宕囧€為梺鍐叉惈閸熶即鏁嶅⿰鍕瘈闁靛骏绲剧涵楣冩煥閺囶亪妾柡鍛劦濮婄粯鎷呴崨濠傛殘闁煎灕鍥ㄧ厱濠电姴鍟版晶杈╃磽閸屾稒宕岄柟绋匡攻缁旂喖鍩¢崒娑辨閻庤娲︽禍婵嬪箯閸涱垱鍠嗛柛鏇ㄥ幗琚欓梻鍌氬€风粈浣革耿闁秴鍌ㄧ憸鏃堝箖濞差亜惟闁宠桨鑳堕鍥⒑閸撴彃浜濇繛鍙夌墵閹偤宕归鐘辩盎闂佺懓顕崑娑㈩敋濠婂懐纾煎ù锝呮惈椤eジ鏌曢崶褍顏い銏℃礋婵偓闁宠桨绀佹竟澶愭⒒娴g懓顕滅紒瀣浮瀹曟繂鈻庨幘璺虹ウ闁诲函缍嗛崳顕€寮鍡欑瘈濠电姴鍊规刊鍏间繆閺屻儲鏁辩紒缁樼箞閹粙妫冨☉妤佸媰闂備焦鎮堕崝宀€绱炴繝鍌ゅ殨妞ゆ劑鍊楅惌娆愪繆椤愩倖鏆╅柛搴涘€楅幑銏犫攽鐎n亞鍊為梺闈浨归崕鏌ヮ敇濞差亝鈷戦柛婵嗗濡叉悂鏌eΔ浣虹煉鐎规洘鍨块獮鎺懳旈埀顒勫触瑜版帗鐓涢柛鎰╁妿婢ф盯鏌i幘宕囩闁哄本鐩崺鍕礃閳哄喚妲烽梻浣呵圭换鎰版儔閼测晜顫曢柟鐑橆殢閺佸﹪鏌涜箛鎿冩Ц濞存粓绠栭幃娲箳瀹ュ棛銈板銈庡亜椤︾敻鐛崱娑樻閹煎瓨鎸婚~宥夋⒑閸︻厼鍔嬮柛銊ㄦ珪缁旂喖寮撮悢铏诡啎闁哄鐗嗘晶浠嬪箖婵傚憡鐓涢柛婊€绀佹禍婊堝础闁秵鐓曟い鎰Т閸旀粓鏌i幘瀛樼闁哄瞼鍠栭幃婊兾熺拠鏌ョ€洪梻浣呵归鍥ㄧ箾閳ь剟鏌$仦鐣屝ユい褌绶氶弻娑滅疀閺冨倶鈧帗绻涢崱鎰仼妞ゎ偅绻勯幑鍕洪鍜冪船婵犲痉鏉库偓褏寰婃禒瀣柈妞ゆ牜鍋涚粻鐘虫叏濡顣抽柛瀣崌閻涱噣宕归鐓庮潛闂備礁鎽滈崰鎾寸箾閳ь剛鈧娲橀崹鍧楃嵁濡皷鍋撳☉娅亪顢撻幘缁樷拺缂備焦锚閻忥箓鏌ㄥ鑸电厓鐟滄粓宕滃☉銏犵;闁绘梻鍘ч悞鍨亜閹烘垵鏋ゆ繛鍏煎姍閺岀喖顢欓懖鈺佺厽閻庤娲樺ú鐔笺€佸☉銏″€烽柤纰卞墮婵附淇婇悙顏勨偓鏍垂婵傜ǹ纾垮┑鐘宠壘缁€鍌炴倶閻愭澘瀚庡ù婊勭矒閺岀喖骞嗚閹界娀鏌涙繝鍐ㄥ闁哄瞼鍠栭、娆撴嚃閳轰胶鍘介柣搴ゎ潐濞茬喐绂嶉崼鏇犲祦闁搞儺鍓欐儫闂侀潧顦崐鏇⑺夊顑芥斀闁绘劘鍩栬ぐ褏绱掗懠顒€浜剧紒鍌氱Ч閹崇偤濡疯濞村嫰姊洪幐搴㈢5闁稿鎹囧Λ浣瑰緞閹邦厾鍘遍棅顐㈡处濞叉牜鏁崼鏇熺厵闁稿繐鍚嬮崐鎰版煛鐏炵晫啸妞ぱ傜窔閺屾稖绠涢弮鍌楁闂傚洤顦甸弻娑㈠Ψ椤旂厧顫╃紒鐐劤閵堟悂寮婚弴鐔虹瘈闊洦娲滈弳鐘差渻閵堝棙绀夊瀛樻倐楠炲牓濡搁妷搴e枔缁瑩宕归纰辨綍闂傚倷鑳舵灙妞ゆ垵妫濋獮鎰節濮橆剛顔嗛梺鍛婁緱閸ㄩ亶宕伴崱娑欑厱闁哄洢鍔屾晶浼存煛閸℃ê鍝烘慨濠勭帛閹峰懘宕崟顐$帛闁诲孩顔栭崰妤呭磿婵傜ǹ桅闁圭増婢樼粈鍐┿亜韫囨挻顥犲璺哄娣囧﹪濡惰箛鏇炲煂闂佸摜鍣ラ崹璺虹暦閹达附鍋愮紓浣贯缚閸橀亶姊洪弬銉︽珔闁哥噥鍋呴幈銊╁焵椤掑嫭鈷戠紒瀣儥閸庢劙鏌熺粙娆剧吋妤犵偛绻樺畷銊р偓娑櫭禒鎯ь渻閵堝棛澧柤鐟板⒔缁骞嬮敂瑙f嫽婵炶揪绲介幉锟犲箚閸儲鐓曞┑鐘插閸︻厼寮查梻渚€娼х换鍫ュ磹閺囥垺鍊块柛顭戝亖娴滄粓鏌熺€电ǹ浠滄い鏇熺矌缁辨帗鎷呯憴鍕嚒濡炪値鍙€濞夋洟骞夐幘顔肩妞ゆ巻鍋撶痪鐐▕閹鈻撻崹顔界亾闂佽桨绀侀…鐑藉Υ娴g硶妲堟俊顖涚矌閸犲酣鎮鹃埄鍐跨矗濞达絽澹婂Λ婊勭節閻㈤潧浠╅柟娲讳簽缁辩偤鍩€椤掍降浜滄い鎰╁焺濡偓闂佽鍣换婵嬪春閳ь剚銇勯幒鎴濐仾闁抽攱甯¢弻娑氫沪閹规劕顥濋梺閫炲苯鍘哥紒顔界懇閵嗕礁鈻庨幇顔剧槇闂佸憡娲﹂崜锕€岣块悢鍏尖拺闁告挻褰冩禍婵囩箾閸欏澧辩紒顔垮吹缁辨帒螣闂€鎰泿闂備礁婀遍崑鎾翅缚濞嗘拲澶婎潩閼哥數鍘遍柣搴秵閸嬪懐浜告导瀛樼厵鐎瑰嫮澧楅崵鍥┾偓瑙勬礈閸忔﹢銆佸Ο琛℃敠闁诡垎鍌氼棜濠电姷鏁告慨鏉懨洪敃鍌氱9闁割煈鍋嗙粻楣冩煙鐎涙ḿ绠橀柡瀣暟缁辨帡鍩€椤掑倵鍋撻敐搴℃灍闁绘挸鍟伴幉绋库堪閸繄顦у┑鐐村灦濮樸劑鎯岄崱妞曞綊鏁愰崼鐔粹偓鍐煟閹烘埊韬柡宀€鍠庨埢鎾诲垂椤旂晫浜愰梻浣呵归鍡涘箰閹间礁鐓″璺哄閸嬫捇宕烽鐐愩儲銇勯敂鍨祮婵﹥妞介弻鍛存倷閼艰泛顏梺鍛娒幉锛勬崲濞戙垹绾ч柟瀵稿仜閺嬬姴顪冮妶鍐ㄧ仾闁挎洏鍨归悾鐑筋敃閿曗偓鍞悷婊冪灱缁厽寰勬繛鐐杸闁圭儤濞婂畷鎰板箻缂佹ê鈧潡鏌ㄩ弮鈧畷妯绘叏閾忣偅鍙忔俊顖氱仢閻撴劙鏌i幘宕囩闁哄本鐩崺鍕礃閳哄喚妲舵俊鐐€х拋锝嗕繆閸ヮ剙鐒垫い鎺嗗亾婵犫偓鏉堛劎浠氭俊鐐€ら崢濂稿床閺屻儲鍋╅柣鎴eГ閺呮煡鏌涢妷顖炴闁告洖鍟村铏圭矙閹稿孩鎷卞銈冨妼閹冲繒绮嬪澶婄畾妞ゎ兘鈧磭绉洪柡浣瑰姍瀹曘劑顢欓崗鍏肩暭闂傚倷绀侀幉鈥趁洪悢铏逛笉闁哄稁鍘奸拑鐔兼煥濠靛棭妲归柛濠勫厴閺屾稑鈻庤箛锝嗏枔濠碘槅鍋呴崹鍨潖濞差亝鐒婚柣鎰蔼鐎氫即鏌涘Ο缁樺€愰柡宀嬬秮楠炴帡鎮欓悽鍨闁诲孩顔栭崳顕€宕滈悢椋庢殾闁圭儤鍩堝ḿ鈺呮煥濠靛棙顥犻柛娆忓暞缁绘繂鈻撻崹顔界亾闂佺娅曢幐鍝ュ弲闂佺粯枪椤曆呭婵犳碍鐓欓柟顖嗗懏鎲兼繝娈垮灡閹告娊寮诲☉妯锋婵鐗婇弫楣冩⒑闂堚晝绋婚柟顔煎€垮濠氭晲閸℃ê鍔呴梺闈涚箳婵挳寮稿▎鎾寸厽闁绘ê鍟挎慨澶愭煕閻樺磭澧电€规洘妞介崺鈧い鎺嶉檷娴滄粓鏌熺€电ǹ浠滄い鏇熺矋閵囧嫰鏁冮崒銈嗩棖缂備浇椴搁幐鎼侇敇婵傜ǹ妞藉ù锝嚽规竟搴ㄦ⒒娴d警鏀版繛鍛礋閹囨偐鐠囪尙鐤囬梺缁樕戝鍧楀极閸℃稒鐓曢柟閭﹀枛娴滈箖鏌﹂幋婵愭Ш缂佽鲸鎹囧畷鎺戔枎閹存繂顬夐梻浣告啞閸旀洟鈥﹂悜鐣屽祦闊洦绋掗弲鎼佹煥閻曞倹瀚�28缂傚倸鍊搁崐鎼佸磹妞嬪孩顐介柨鐔哄Т绾捐顭块懜闈涘Е闁轰礁顑囬幉鎼佸籍閸稈鍋撴担鑲濇棃宕ㄩ闂寸盎闂備焦鍎崇换鎰耿闁秵鍋傞悗锝庡枟閳锋垿鎮峰▎蹇擃仾闁稿孩顨婇弻娑㈠Ω閵壯嶇礊婵犮垼顫夊ú鐔煎极閹剧粯鏅搁柨鐕傛嫹
Hai-Nan Lin
, Xin Li
, Li Tang
, Department of Physics, Chongqing University, Chongqing 401331, China
Received Date:2019-03-20
Available Online:2019-07-01
Abstract:The equation of state (EoS) of dark energy plays an important role in the evolution of the universe and has attracted considerable interest in the recent years. With the progress in observational technique, a precise constraint on the EoS of dark energy can be obtained. In this study, we reconstruct the EoS of dark energy and cosmic expansion using Gaussian processes (GP) from the most up-to-date Pantheon compilation of type Ia supernovae (SNe Ia), which consists of 1048 finely calibrated SNe Ia. The reconstructed EoS of dark energy has a large uncertainty owing to its dependence on the second-order derivative of the construction. Adding the direct measurements of Hubble parameters $H(z)$ as an additional constraint on the first-order derivative can partially reduce the uncertainty; however, it is still not sufficiently precise to distinguish between the evolving and the constant dark energy. Moreover, the results heavily rely on the prior of the Hubble constant $H_0$. The $H_0$ value inferred from SNe+$H(z)$ without prior is $H_0=70.5\pm 0.5~{\rm km~s^{-1}~Mpc^{-1}}$. Moreover, the matter density $\Omega_M$ has a non-negligible effect on the reconstruction of dark energy. Therefore, more accurate determinations on $H_0$ and $\Omega_M$ are required to tightly constrain the EoS of dark energy.

HTML

--> --> -->
1.Introduction
The late time cosmic acceleration is one of the most important discoveries in modern cosmology and it revives Einstein's cosmological constant hypothesis. Since the first discovery of cosmic acceleration from type Ia supernovae (SNe Ia) in the late 1990s [1, 2], it has now been confirmed by various other independent observations, such as the large scale structure [3], growth function [4] and cosmic microwave background radiations [5, 6]. This leads to the final foundation of the standard model of cosmology, i.e. the cold dark matter plus a cosmological constant ($ \Lambda {\rm{CDM}}$) model. Here, $ \Lambda $ denotes the cosmological constant (or an alternative name dark energy), which provides a negative pressure and is responsible for the acceleration of the universe. According to the $ \Lambda {\rm{CDM}}$ model, the equation of state (EoS) of dark energy is a constant and does not evolve with cosmos, i.e., $ w = p/\rho\equiv -1 $. Although the $ \Lambda {\rm{CDM}}$ model has achieved great success, it still confronts some problems, among which the most important ones are the coincidence problem and the fine-turning problem [7, 8]. The EoS of dark energy plays an essential role in the evolution of the universe. In terms of different EoS of dark energy, several alternative models have been proposed, such as the model with a constant w but does not necessarily equate to ?1, the evolving dark energy models, e.g., the Chevallier-Polarski-Linder parametrization [9, 10] and various other parameterizations [11, 12]. In some models the dark energy is replaced by a scalar field, such as the quintessence field [13], phantom field [14] and tachyon field [15]. The effective EoS of these scalar fields is also evolving with the cosmos.
The above models depend on the specific parametrization of dark energy or the scalar fields; thus, they are model dependent. Moreover, most parameterizations lack physical interpretation; hence, they are just phenomenological. Because we have no prior knowledge on the explicit form of dark energy, reconstructing it in a non-parametric manner is of great importance. Therefore, some model-independent methods have been proposed, among which the Gaussian processes (GP) is one of the most widely used methods. Unlike the best-fitting method that must have a concrete model to fit the data, the GP method can reconstruct a theoretical curve from the discrete data points without evolving any specific model. Since Ref. [16] first applied the GP method to investigate the dark energy, it has been widely used and has shown its applicability in cosmology [1726]. The advantage of the GP method is that it does not need the concrete form of the model; the only assumption is that the observational data points are drawn from the multivariate Gaussian distribution.
In this study, we attempt to reconstruct the EoS of dark energy using the GP method from the latest dataset of SNe Ia, i.e., the Pantheon compilation [27], which consist of 1048 finely calibrated SNe Ia. The EoS of dark energy has strong influence on the Hubble expansion rate $ H(z) $ and the deceleration parameter $ q(z) $, which will be simultaneously obtained in the reconstruction. The reconstructed dark energy depends on the second-order derivative of the GP (see the next section for details); hence, it has a large uncertainty. To improve the significance, we use the direct measurement of Hubble parameters $ H(z) $ as an additional constraint in the GP reconstruction. Because tensin is greater than $ 3\sigma $ is the between the Hubble constant $ H_0 $ from the local distance ladders [28] and from the global CMB radiation [6], we will also investigate the impact of different values of $ H_0 $ on the reconstruction.
The remainder of this paper is organized as follows. In section 2, we introduce the methodology and the relevant data used in our analysis. The results along with some discussions are presented in section 3. Finally, a short summary is given in section 4.
2.Data and methodology
The Hubble expansion rate $ H(z) $ strongly depends on the contents of the universe and the EoS of dark energy. In a spatially flat Friedmann-Robertson-Walker universe dominated by non-relativistic matter (including baryons and dark matter) and dark energy, the evolution of $ H(z) $ is governed by the Friedmann equation [29]
$ H^2(z) = H_0^2\left\{\Omega_M(1+z)^3+\Omega_{\Lambda}\exp\left[3\int_0^z\frac{1+w(z)}{1+z}{\rm d}z\right]\right\}, $
(1)
where $ H_0 $ is the Hubble constant, $ \Omega_M $ and $ \Omega_{\Lambda} $ are the normalized density of non-relativistic matter and dark energy at current epoch, respectively, and $ w(z) = p(z)/\rho(z) $ is the EoS of dark energy. The normalized co-moving distance is related to the Hubble expansion rate by [30]
$ d_c(z) = \int_0^z\frac{{\rm d}z}{E(z)}, $
(2)
where $ E(z)\equiv H(z)/H_0 $ is the normalized Hubble parameter. From Eq. (2) we have
$ E(z) = \frac{1}{d_c'(z)}, $
(3)
where the prime denotes the derivative with respect to redshift z. Combining Eqs. (1) and (3), we can solve for $ w(z) $,
$ w(z) = \frac{-2(1+z)d_c''-3d_c'}{3[d_c'-\Omega_M(1+z)^3d_c'^3]}. $
(4)
The acceleration of the universe is often represented by the so-called "deceleration parameter", which is defined by $ q(z) = -\ddot{a}a/\dot{a}^2 $, where $ a = 1/(1+z) $ is the scale factor of the universe, and the dot denotes the derivative with respect to cosmic time. Using $ H = \dot{a}/a $, the deceleration parameter can be rewritten as
$ q(z) = (1+z)\frac{H'}{H}-1 = -(1+z)\frac{d_c''}{d_c'}-1. $
(5)
A positive or negative q implies a decelerating or accelerating universe, respectively.
If we know $ d_c(z) $ as a function of z, $ E(z) $ can be obtained from Eq. (3). Similarly $ w(z) $ and $ q(z) $ can be obtained from Eqs. (4) and (5), respectively. In a spatially flat universe, the normalized co-movig distance $ d_c(z) $ is related to the luminosity distance $ D_{\rm L}(z) $ by
$ d_c(z) = \frac{1}{1+z}\frac{H_0}{c}D_{\rm L}(z). $
(6)
The luminosity distance can be measured from the brightness of SNe Ia. SNe Ia are often assumed to have an approximately constant absolute magnitude after the color and stretch corrections; therefore, they are widely regarded as the standard candles. The distance modulus of SNe Ia can be derived from the observation of light curves through the empirical relation [3133]
$ \mu_{\rm sn} = m_B^*+\alpha X_1-\beta {\cal C}-M_B, $
(7)
where $ m_B^* $ is the B-band apparent magnitude, $ X_1 $ and $ {\cal C} $ are the stretch and color parameters respectively, and $ M_{\rm B} $ is the absolute magnitude. $ \alpha $ and $ \beta $ are two nuisance parameters. The luminosity distance of SNe Ia can be calculated from the distance modulus through the following relation
$ \mu = 5\log_{10}\frac{D_{\rm L}}{\rm Mpc}+25. $
(8)
Several SNe Ia samples have been released, among which the most up-to-date is the Pantheon compilation [27]. The Pantheon sample is at present the largest sample that consists of different supernovae surveys, including SDSS, SNLS, various low-z samples, and some high-z samples from the HST. The total number of SNe in the Pantheon dataset is 1048, which is about twice that of the Union2.1 sample [34] and is approximately 40% more than that of the JLA sample [35]. Moreover, the furthest SNe reaches approximately $ z\sim 2.3 $, and the systematic uncertainty is further reduced compared with the previous samples. Usually, the nuisance parameters $ \alpha $ and $ \beta $ are optimized simultaneously with the cosmological parameters or are marginalized over. However, this method is model dependent; thus, the distance calibrated in one cosmological model could not be directly used to constrain the other models. The Pantheon sample applies a new method called the BEAMS with Bias Corrections (BBC) to calibrated the SNe. According to the BBC method, the nuisance parameters $ \alpha $ and $ \beta $ are determined by fitting to an randomly chosen reference cosmology. There is no special requirement on the reference cosmology; however, it should not deviate too far from the data. Once $ \alpha $ and $ \beta $ are determined, we can fix them in other cosmology fits. In the Pantheon sample, the corrected apparent magnitude $ m_{B, {\rm corr}}^* = m_B^*+\alpha X_1-\beta {\cal C} $ are reported. Therefore, we do not need to do the color and stretch corrections any more; hence, we fix $ \alpha = \beta = 0 $ in Eq. (7). The statistical uncertainty $ { D}_{\rm stat} $ and systematic uncertainty $ { C}_{\rm sys} $ are also given in Ref. [27]. The total uncertainty matrix of distance modulus is given by
$ {\bf \Sigma}_\mu = { D}_{\rm stat}+{ C}_{\rm sys}. $
(9)
We convert the distance modulus of SNe to the normalized co-moving distance through the relation
$ d_c = \frac{1}{1+z}\frac{H_0}{c}10^{\frac{\mu-25}{5}}. $
(10)
The uncertainty of $ d_c $ is propagated from the uncertainties of $ \mu $ and $ H_0 $ using the standard error propagation formula,
$ {\bf \Sigma}_{d_c} = { D}_1{\bf \Sigma}_\mu { D}_1^{\rm T}+{ \sigma}_{H_0}^2{ D}_2{ D}_2^{\rm T}, $
(11)
where $ \sigma_{H_0} $ is the uncertainty of Hubble constant, the superscript "T" denotes the transpose of a matrix, $ { D}_1 $ and $ { D}_2 $ are the Jacobian matrices,
$ { D}_1 = {\rm diag}\left(\frac{\ln10}{5}{ d}_c\right), $
(12)
$ { D}_2 = {\rm diag}\left(\frac{1}{H_0}{ d}_c\right), $
(13)
where $ { d}_c $ is a vector whose components are the normalized co-moving distances of all the SNe Ia in Pantheon, and $ {\rm diag}({ v}) $ is the square diagonal matrix with the elements of vector v on the main diagonal.
To obtain the co-moving distance - redshift relation from the discrete data points, we use the GP method [19] to reconstruct the $ d_c(z) $ function, and the derivatives $ d_c'(z) $ and $ d_c''(z) $ can be obtained simultaneously in the reconstruction procedure. The GP can reconstruct a function $ y = f(x) $ from the discrete data points $ (x_i,y_i $) without assuming a particular parametrization of the function $ f(x) $. It assumes that the data points are drawn from the multivariate Gaussian distribution,
$ { y}\sim {\cal N}({ \mu},{ K}({ x},{ x})+{ C}), $
(14)
where $ { x} = \{x_i\} $, $ { y} = \{y_i\} $, $ { \mu} $ is the mean of the Gaussian distribution, C is the covariance matrix of the data, and $ [{ K}({ x},{ x})]_{ij} = k(x_i,x_j) $ is another covariance matrix that controls the behavior of the reconstructed function. All the freedoms of GP originate from the choice of the covariance function $ k(x_i,x_j) $. There are several covariance functions available; however, any covariance function should be symmetric, positive definite, and monotonously decreasing with $ |x_i-x_j| $. In this study, we use the simplest and most widely used squared-exponential covariance function defined by
$ k(x_i,x_j) = \sigma_f^2\exp\left[-\frac{(x_i-x_j)^2}{2l^2}\right]. $
(15)
The hyperparameters $ \sigma_f $ and l are optimized by maximizing the marginalized likelihood. For more details on the GP, please refer to Ref. [19].
It has been noted that although the GP method can reconstruct the function $ f(x) $ with a relatively high precision, the reconstructed derivatives, especially the higher order derivatives of $ f(x) $ have large uncertainty. If there are observational constraints on the derivatives of the function, then the uncertainty can be reduced. From Eq. (3) we can know that $ d'_c(z) = 1/E(z) = H_0/H(z) $; thus, the direct measurement of Hubble parameters can be used as an additional constraint on the first-order derivative of $ d_c(z) $. Here, we use the 51 $ H(z) $ data points (except for $ H_0 $) complied in Ref. [36], which is, to the best our knowledge, the largest data sample available at present. These $ H(z) $ data points are measured from two different methods, i.e., the differential age of galaxies (DAG) method and baryon acoustic oscillations (BAO) method. Because the BAO method relies on the cosmological model, to avoid the model dependence we only use the remaining 31 data points measured from the DAG method.
To normalize $ H(z) $, the precise measurement of $ H_0 $ is necessary. It is well known that the is greater than $ 3\sigma $ between the values measured from the local distance ladders and that from the global CMB radiation, where the former gives $ H_0 = 73.24\pm 1.74\; {\rm km\; s^{-1}\; Mpc^{-1}} $ [28] and the latter gives $ H_0 = 67.4\pm 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}} $ [6]. To investigate the influence of Hubble constant on the reconstruction, we consider these two different values as the prior on $ H_0 $.
3.Results and discussions
The publicly available python package $ {GaPP}$ [19] is used to do the GP reconstructions. First, we reconstruct from the Pantheon dataset. Then, we add the $ H(z) $ data to make a combined reconstruction. Because $ w(z) $ depends on the matter density parameter $ \Omega_M $, we fix it to the value of Planck 2018 results, i.e., $ \Omega_M = 0.315 $ [6]. The impact of different $ \Omega_M $ values on the reconstruction of $ w(z) $ will be discussed later. Note that the other quantities ($ d_c(z) $, $ E(z) $ and $ q(z) $) are independent of $ \Omega_M $. The absolute magnitude of SNe Ia is degenerated with the Hubble constant, and we fix it to $ M_{\rm B} = -19.35 $, the best-fitting value of $ \Lambda {\rm{CDM}}$ model.
The GP reconstructions of $ d_c(z) $, $ E(z) $, $ q(z) $, and $ w(z) $ from SNe data with prior $ H_0 = 67.4\pm 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}} $ plotted in Fig. 1. The blue curves are the reconstructed central values, and the shaded regions are the $ 1\sigma $ and $ 2\sigma $ uncertainties. For comparison, we also plot the best-fitting curves of $ \Lambda {\rm{CDM}}$ model, with the best-fitting parameters $ \Omega_M = 0.3 $, $ H_0 = 68.9\; {\rm km\; s^{-1}\; Mpc^{-1}} $. Figure 1 shows that the reconstructions of all the four quantities are well matched to the $ \Lambda {\rm{CDM}}$ model in low redshift (z < 0.5) region. However, in the redshift region higher than 0.5, the reconstructed curves do not agree with the $ \Lambda {\rm{CDM}}$ model. This disagreement is especially obvious for $ w(z) $, which is more than $ 2\sigma $ deviation from ?1 in the intermediate redshift region. Owing to the large uncertainties at z > 0.5, the remaining three quantities ($ d_c(z) $, $ E(z) $, and $ q(z) $) are still consistent with the $ \Lambda {\rm{CDM}}$ model within $ 2\sigma $ confidence level.
Figure1. (color online) GP reconstructions of $d_c(z)$, $E(z)$, $q(z)$, and $w(z)$ from SNe, with prior $H_0 = 67.4\pm 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}}$. The black dashed curves are the predictions of flat $\Lambda$CDM model with parameters $\Omega_M = 0.3$, $H_0 = 68.9\; {\rm km\; s^{-1}\; Mpc^{-1}}$.

The GP reconstructions from the same data but with prior $ H_0 = 73.24\pm 1.74\; {\rm km\; s^{-1}\; Mpc^{-1}} $ are plotted in Fig. 2. Similar to Fig. 1, the reconstructions of $ d_c(z) $, $ E(z) $, and $ q(z) $ are consistent with the $ \Lambda {\rm{CDM}}$ model within $ 2\sigma $ confidence level, especially in the low redshift region; they are excellently in agreement with $ \Lambda {\rm{CDM}}$. However, the reconstruction of $ w(z) $ shows a discrepancy greater than $ 2\sigma $ from ?1 in the low redshift region; however, it is consistent with ?1 in the intermediate and high redshift regions. This is contrary to Fig. 1, which shows the discrepancy from $ \Lambda {\rm{CDM}}$ in the intermediate redshift region. This implies that the $ H_0 $ value has a significant impact on the reconstruction of $ w(z) $.
Figure2. (color online) Same as Fig.1 but with prior $H_0 = 73.24\pm $$ 1.74\; {\rm km\; s^{-1}\; Mpc^{-1}}$.

The tension is greater than $ 3\sigma $ tension between the two $ H_0 $ priors we used here. A wrong $ H_0 $ prior may lead to a wrong result on the reconstruction of $ w(z) $. It is interesting to see which of the two $ H_0 $ priors is more consistent with the SNe Ia data. To this end, instead of reconstructing the normalized comoving distance $ d_c(z) $, we first directly reconstruct the un-normalized co-moving distance $ D_c(z) $ (and its derivatives) as the function of redshift, where
$ D_c(z) = \frac{c}{H_0}d_c(z) = \frac{1}{1+z}D_{\rm L}(z). $
(16)
Because $ D_{\rm L}(z) $ is directly measured from SNe Ia, $ D_c(z) $ is independent of $ H_0 $. Because $ d'_c(0) = 1/E(0)\equiv 1 $, we can infer the Hubble constant by
$ H_0 = \frac{c}{D'_c(0)}, \quad \sigma_{H_0} = \frac{c\sigma_{D'_c(0)}}{[D'_c(0)]^2}, $
(17)
where $ D'_c(0) $ is the derivative of $ D_c(z) $ at z = 0. Using this method, we obtain $ H_0 = 70.6\pm 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}} $, which is approximately the mean value of the local and global measurements of $ H_0 $. Then, we normalize $ D_c(z) $ (and its derivatives) with the inferred $ H_0 $ and calculate $ E(z) $, $ q(z) $, and $ w(z) $ in the same manner as the previous cases. The results are plotted in Fig. 3. Now, the reconstructed $ w(z) $ is consistent with ?1 within $ 2\sigma $ confidence level in the entire redshift region, except for a small region near $ z\sim 1.2 $. In addition, from the $ q(z) $ subfigure we see a turn point at $ z = 0.59_{-0.06}^{+0.08} $, where the universe changes from accelerating to decelerating. The location of the turn point is in good agreement with the prediction of $ \Lambda {\rm{CDM}}$ model.
Figure3. (color online) Same as Fig.1 but with no prior on $ H_0$.

The reconstructed quantities, especially $ E(z) $, $ q(z) $, and $ w(z) $, which depend on the derivatives of $ d_c(z) $, have a large uncertainty in the high redshift region. To reduce the uncertainty, we combine the SNe Ia with the 31 DAG $ H(z) $ data in the reconstruction, where the inverse of the normalized $ H(z) $ data are treated as an additional constraint on the first-order derivative of $ d_c(z) $. The reconstruction from the SNe+$ H(z) $ data with prior $ H_0 = 67.4\pm$$ 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}} $ and prior $ H_0 = 73.24\pm1.74\; {\rm km\; s^{-1}\; Mpc^{-1}} $ are plotted in Fig. 4 and Fig. 5, respectively. Compared with Fig. 1 and Fig. 2, we may see that adding the $ H(z) $ data can partially reduce the uncertainty. However, now the constructed $ d_c(z) $ is not consistent with $ \Lambda {\rm{CDM}}$ within $ 2\sigma $ confidence level in the intermediate redshift region. The $ H_0 $ prior directly affects $ E(z) $; thus, it affects the slope of $ d_c(z) $. With the small $ H_0 $ prior, the slope of $ d_c(z) $ is small; therefore, the reconstructed $ d_c(z) $ increases slower than the prediction of $ \Lambda {\rm{CDM}}$. On the contrary, with the large $ H_0 $ prior, the reconstructed $ d_c(z) $ increases faster than the prediction of $ \Lambda {\rm{CDM}}$. Similar to the SNe only case, with a small $ H_0 $ prior, $ w(z) $ deviates from ?1 in the intermediate redshift region, while with large $ H_0 $ prior, $ w(z) $ deviates from ?1 in the low redshift region.
Figure4. (color online) GP reconstruction of $d_c(z)$, $E(z)$, $q(z)$, and $w(z)$ from SNe+$H(z)$, with prior $H_0 = 67.4\pm 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}}$. The black dashed curves are the predictions of flat $\Lambda$CDM model with parameters $\Omega_M = 0.3$, $H_0 = 68.9\; {\rm km\; s^{-1}\; Mpc^{-1}}$.

Figure5. (color online) Same to Fig.4 but with prior $ H_0 = 73.24\pm $$1.74\; {\rm km\; s^{-1}\; Mpc^{-1}}$.

Finally, we reconstruct $ d_c(z) $, $ E(z) $, $ q(z) $, and $ w(z) $ without $ H_0 $ prior using a method similar to the one mentioned above. We directly reconstruct the un-normalized $ D_c(z) $, treating the $ c/H(z) $ data as an additional constraint on the first derivative of $ D_c(z) $. Then, we infer $ H_0 $ and its uncertainty from the reconstructed $ D_c(z) $ curve using Eq. (17). The results are plotted in Fig. 6. The inferred Hubble constant is $ H_0 = 70.5\pm 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}} $, which agrees with the value inferred from SNe Ia only. The $ d_c(z) $ is excellently in agreement with $ \Lambda {\rm{CDM}}$ model in the entire redshift region, and $ E(z) $ is consistent with the $ \Lambda {\rm{CDM}}$ model within $ 1\sigma $ confidence level. Compared with Fig. 3, an obvious difference can be seen in the $ q(z) $ subfigure. Except for an unambiguous accelerating-to-decelerating turn point at $ z = 0.59_{-0.05}^{+0.05} $, there is another possible, but with large uncertainty, turn point near $ z\sim 1.8 $. In addition, $ q(z) $ and $ w(z) $ are consistent with $ \Lambda {\rm{CDM}}$ within $ 2\sigma $ confidence level. Therefore, we conclude that the combined data of SNe+$ H(z) $ shows no evidence for the deviation from the standard cosmological model.
Figure6. (color online) Same as Fig.4 but with no prior on $ H_0$.

To investigate the influence of $ \Omega_M $ on $ w(z) $, we reconstruct $ w(z) $ from SNe+$ H(z) $ data with different $ \Omega_M $ values and without $ H_0 $ prior. Define the significance of deviation of $ w(z) $ from ?1 as
$ \sigma(z) = \frac{w-(-1)}{\sigma_{w}}, $
(18)
where w and $ \sigma_w $ are the reconstructed central value and $ 1\sigma $ uncertainty, respectively. We plot $ \sigma(z) $ for different $ \Omega_M $ values ($ \Omega_M = [0.27,0.30,0.315,0.33] $) in Fig. 7. It is shown that the $ \Omega_M $ value has a significant effect on the reconstruction of $ w(z) $. For $ \Omega_M = 0.27 $, $ w(z) $ deviates from ?1 at more than $ 3\sigma $ confidence level in the intermediate redshift region. For $ \Omega_M = 0.3 $ and $ \Omega_M = 0.33 $, $ w(z) $ deviates from ?1 at more than $ 2\sigma $ confidence level near $ z\sim 1.0 $ and $ z\sim 0.3 $, respectively. However, for $ \Omega_M = 0.315 $,$ w(z) $ is consistent with ?1 within $ 2\sigma $ confidence level in the entire redshift region. It is interesting that the deviation of $ w(z) $ from ?1 reaches its peak value at redshift $ z \sim 1.0 $ for any $ \Omega_M $. Owing to the large uncertainty of $ w(z) $ in the high redshift region, $ w(z) $ is consistent with ?1 within $ 1\sigma $ uncertainty for all $ \Omega_M $ values. Therefore, a precise measurement of the matter density parameter $ \Omega_M $ is necessary to tightly constrain the dark energy.
Figure7. (color online) The deviation of $ w(z)$ from ?1 for different $ \Omega_M$ values.

Note that the uncertainty of $ E(z) $, $ q(z) $, and $ w(z) $ increases sharply beyond redshift $ z\sim 1.5 $ owing to the lack of data points. Adding the $ H(z) $ data can partially reduce the uncertainty; however, it is still unacceptably large. Especially, the $ w(z) $ reconstructed from SNe+$ H(z) $ has a sudden break and the uncertainty blows up near $ z\sim 2 $, which is unreasonable. This limiatation of the GP method has already been noted in Ref. [19]. To overcome this flaw, more data points in the high redshift region must be considered.
Recently, Zhang and Li [21] used the Union2.1 and JLA compilations of SNe Ia combined with the $ H(z) $ data to reconstruct the dark energy in the redshift region z < 1.5. They found that the Union2.1+$ H(z) $ and JLA+$ H(z) $ data give similar results, i.e., slight dynamical dark energy can be observed for both; however, the constant dark energy cannot be excluded. They also investigated the effect of $ H_0 $ and $ \Omega_M $ on the construction, and they showed that $ H_0 $ has a notable influence on the results; however, the influence of $ \Omega_M $ is slight. In our work, with the most recent SNe Ia data, we reconstructed the dark energy up to redshift $ z\sim 2.5 $ and obtained similar results to those in Ref. [21]. However, our results show that the effect of both $ H_0 $ and $ \Omega_M $ is non-negligible on the reconstruction of dark energy. This difference may be caused by the reduction of the uncertainty at z < 1.5. From Fig. 7, we see that $ \Omega_M $ value only affects the result bellow $ z\sim 1.5 $. Beyond $ z\sim 1.5 $, owing to the large uncertainty, the influence of $ \Omega_M $ is negligible.
4.Summary
In this study, we reconstructed the EoS of dark energy and the cosmic expansion from SNe Ia using the non-parametric method. To improve the significance, we also added the direct measurement of Hubble parameter, $ H(z) $ data, to construct a combined reconstruction. However, even with the $ H(z) $ data, the reconstruction still has a large uncertainty in the high redshift region. It was found that the $ H_0 $ value has a strong effect on the reconstruction. Without $ H_0 $ prior, the inferred Hubble constant from the combination of SNe+$ H(z) $ data is $H_0 = 70.5\pm $$ 0.5\; {\rm km\; s^{-1}\; Mpc^{-1}} $, thus alleviating the tension between the local and global measurements of $ H_0 $. We have also investigate if the matter density parameter, $ \Omega_M $, has some influence on the reconstruction. It was shown that the reconstruction of $ w(z) $ strongly depends on $ \Omega_M $. With the inferred Hubble constant and the Planck 2018 matter density parameter ($ \Omega_M = 0.315 $), the reconstructed $ w(z) $ is consistent with $ \Lambda {\rm{CDM}}$ model within $ 2\sigma $ confidence level. With the current observational accuracy, it is still premature to distinguish between the evolving and constant dark energy.
相关话题/parametric reconstruction energy