Multimessenger search for sources of gravitational waves and high-energy
neutrinos: Initial results for LIGO-Virgo and IceCube

M. G. Aartsen,2,* M. Ackermann,45,* J. Adams,15,* J. A. Aguilar,23,* M. Ahlers,28,* M. Ahrens,36,* D. Altmann,22,*

T. Anderson,42,* C. Arguelles,28,* T. C. Arlen,42,* J. Auffenberg,1,* X. Bai,34,* S. W. Barwick,25,* V. Baum,29,*

J. J. Beatty,17,18,* J. Becker Tjus,10,* K.-H. Becker,44,* S. BenZvi,28,* P. Berghaus,45,* D. Berley,16,* E. Bernardini,45,*

A. Bernhard,31,* D. Z. Besson,26,* G. Binder,8,7,* D. Bindig,44,* M. Bissok,1,* E. Blaufuss,16,* J. Blumenthal,1,*

D. J. Boersma,43,* C. Bohm,36,* F. Bos,10,* D. Bose,38,* S. Böser,11,* O. Botner,43,* L. Brayeur,13,* H.-P. Bretz,45,*

A. M. Brown,15,* J. Casey,5,* M. Casier,13,* D. Chirkin,28,* A. Christov,23,* B. Christy,16,* K. Clark,39,* L. Classen,22,*

F. Clevermann,20,* S. Coenders,31,* D. F. Cowen,42,41,* A. H. Cruz Silva,45,* M. Danninger,36,* J. Daughhetee,5,*

J. C. Davis,17,* M. Day,28,* J. P. A. M. de André,42,* C. De Clercq,13,* S. De Ridder,24,* P. Desiati,28,* K. D. de Vries,13,*

M. de With,9,* T. DeYoung,42,* J. C. Díaz-Vélez,28,* M. Dunkman,42,* R. Eagan,42,* B. Eberhardt,29,* B. Eichmann,10,*

J. Eisch,28,* S. Euler,43,* P. A. Evenson,32,* O. Fadiran,28,* A. R. Fazely,6,* A. Fedynitch,10,* J. Feintzeig,28,* J. Felde,16,*

T. Feusels,24,* K. Filimonov,7,* C. Finley,36,* T. Fischer-Wasels,44,* S. Flis,36,* A. Franckowiak,11,* K. Frantzen,20,*

T. Fuchs,20,* T. K. Gaisser,32,* J. Gallagher,27,* L. Gerhardt,8,7,* D. Gier,1,* L. Gladstone,28,* T. Glüsenkamp,45,*

A. Goldschmidt,8,* G. Golup,13,* J. G. Gonzalez,32,* J. A. Goodman,16,* D. Góra,45,* D. T. Grandmont,21,* D. Grant,21,*

P. Gretskov,1,* J. C. Groh,42,* A. Groß,31,* C. Ha,8,7,* C. Haack,1,* A. Haj Ismail,24,* P. Hallen,1,* A. Hallgren,43,*

F. Halzen,28,* K. Hanson,12,* D. Hebecker,11,* D. Heereman,12,* D. Heinen,1,* K. Helbing,44,* R. Hellauer,16,* D. Hellwig,1,*

S. Hickford,15,* G. C. Hill,2,* K. D. Hoffman,16,* R. Hoffmann,44,* A. Homeier,11,* K. Hoshina,28,a,* F. Huang,42,*

W. Huelsnitz,16,* P. O. Hulth,36,* K. Hultqvist,36,* S. Hussain,32,* A. Ishihara,14,* E. Jacobi,45,* J. Jacobsen,28,*

K. Jagielski,1,* G. S. Japaridze,4,* K. Jero,28,* O. Jlelati,24,* M. Jurkovic,31,* B. Kaminsky,45,* A. Kappes,22,* T. Karg,45,*

A. Karle,28,* M. Kauer,28,* J. L. Kelley,28,* A. Kheirandish,28,* J. Kiryluk,37,* J. Kläs,44,* S. R. Klein,8,7,* J.-H. Köhne,20,*

G. Kohnen,30,* H. Kolanoski,9,* A. Koob,1,* L. Köpke,29,* C. Kopper,28,* S. Kopper,44,* D. J. Koskinen,19,* M. Kowalski,11,*

A. Kriesten,1,* K. Krings,1,* G. Kroll,29,* M. Kroll,10,* J. Kunnen,13,* N. Kurahashi,28,* T. Kuwabara,32,* M. Labare,24,*

D. T. Larsen,28,* M. J. Larson,19,* M. Lesiak-Bzdak,37,* M. Leuermann,1,* J. Leute,31,* J. Lünemann,29,* O. Macías,15,*

J. Madsen,35,* G. Maggi,13,* R. Maruyama,28,* K. Mase,14,* H. S. Matis,8,* F. McNally,28,* K. Meagher,16,* M. Medici,19,*

A. Meli,24,* T. Meures,12,* S. Miarecki,8,7,* E. Middell,45,* E. Middlemas,28,* N. Milke,20,* J. Miller,13,* L. Mohrmann,45,*

T. Montaruli,23,* R. Morse,28,* R. Nahnhauer,45,* U. Naumann,44,* H. Niederhausen,37,* S. C. Nowicki,21,* D. R. Nygren,8,*

A. Obertacke,44,* S. Odrowski,21,* A. Olivas,16,* A. Omairat,44,* A. O’Murchadha,12,* T. Palczewski,40,* L. Paul,1,*

Ö. Penek,1,* J. A. Pepper,40,* C. Pérez de los Heros,43,* C. Pfendner,17,* D. Pieloth,20,* E. Pinat,12,* J. Posselt,44,*

P. B. Price,7,* G. T. Przybylski,8,* J. Pütz,1,* M. Quinnan,42,* L. Rädel,1,* M. Rameez,23,* K. Rawlins,3,* P. Redl,16,*

I. Rees,28,* R. Reimann,1,* E. Resconi,31,* W. Rhode,20,* M. Richman,16,* B. Riedel,28,* S. Robertson,2,* J. P. Rodrigues,28,*

M. Rongen,1,* C. Rott,38,* T. Ruhe,20,* B. Ruzybayev,32,* D. Ryckbosch,24,* S. M. Saba,10,* H.-G. Sander,29,*

J. Sandroos,19,* M. Santander,28,* S. Sarkar,19,33,* K. Schatto,29,* F. Scheriau,20,* T. Schmidt,16,* M. Schmitz,20,*

S. Schoenen,1,* S. Schöneberg,10,* A. Schönwald,45,* A. Schukraft,1,* L. Schulte,11,* O. Schulz,31,* D. Seckel,32,*

Y. Sestayo,31,* S. Seunarine,35,* R. Shanidze,45,* C. Sheremata,21,* M.W. E. Smith,42,* D. Soldin,44,* G. M. Spiczak,35,*

C. Spiering,45,* M. Stamatikos,17,b,* T. Stanev,32,* N. A. Stanisha,42,* A. Stasik,11,* T. Stezelberger,8,* R. G. Stokstad,8,*

A. Stößl,45,* E. A. Strahler,13,* R. Ström,43,* N. L. Strotjohann,11,* G.W. Sullivan,16,* H. Taavola,43,* I. Taboada,5,*

A. Tamburro,32,* A. Tepe,44,* S. Ter-Antonyan,6,* A. Terliuk,45,* G. Tešić,42,* S. Tilav,32,* P. A. Toale,40,* M. N. Tobin,28,*

D. Tosi,28,* M. Tselengidou,22,* E. Unger,10,* M. Usner,11,* S. Vallecorsa,23,* N. van Eijndhoven,13,* J. Vandenbroucke,28,*

J. van Santen,28,* M. Vehring,1,* M. Voge,11,* M. Vraeghe,24,* C. Walck,36,* M. Wallraff,1,* Ch. Weaver,28,* M. Wellons,28,*

C. Wendt,28,* S. Westerhoff,28,* B. J. Whelan,2,* N. Whitehorn,28,* C. Wichary,1,* K. Wiebe,29,* C. H. Wiebusch,1,*

D. R. Williams,40,* H. Wissing,16,* M. Wolf,36,* T. R. Wood,21,* K. Woschnagg,7,* D. L. Xu,40,* X. W. Xu,6,* J. P. Yanez,45,*

G. Yodh,25,* S. Yoshida,14,* P. Zarzhitsky,40,* J. Ziemann,20,* S. Zierke,1,* M. Zoll,36,* J. Aasi,46,† B. P. Abbott,46,†

R. Abbott,46,† T. Abbott,47,† M. R. Abernathy,46,† F. Acernese,48,49,† K. Ackley,50,† C. Adams,51,† T. Adams,52,†

P. Addesso,53,† R. X. Adhikari,46,† C. Affeldt,54,† M. Agathos,55,† N. Aggarwal,56,† O. D. Aguiar,57,† P. Ajith,58,†

A. Alemic,59,† B. Allen,54,60,61,† A. Allocca,62,63,† D. Amariutei,50,† M. Andersen,64,† R. A. Anderson,46,† S. B. Anderson,46,†

W. G. Anderson,60,† K. Arai,46,† M. C. Araya,46,† C. Arceneaux,65,† J. S. Areeda,66,† S. Ast,61,† S. M. Aston,51,† P. Astone,67,†

P. Aufmuth,61,† H. Augustus,68,† C. Aulbert,54,† B. E. Aylott,68,† S. Babak,69,† P. T. Baker,70,† G. Ballardin,71,†

S.W. Ballmer,59,† J. C. Barayoga,46,† M. Barbet,50,† B. C. Barish,46,† D. Barker,72,† F. Barone,48,49,† B. Barr,73,†

L. Barsotti,56,† M. Barsuglia,74,† M. A. Barton,72,† I. Bartos,75,† R. Bassiri,64,† A. Basti,76,63,† J. C. Batch,72,†

J. Bauchrowitz,54,† Th. S. Bauer,55,† C. Baune,54,† V. Bavigadda,71,† B. Behnke,69,† M. Bejger,77,† M. G. Beker,55,†

C. Belczynski,78,† A. S. Bell,73,† C. Bell,73,† G. Bergmann,54,† D. Bersanetti,79,80,† A. Bertolini,55,† J. Betzwieser,51,†

I. A. Bilenko,81,† G. Billingsley,46,† J. Birch,51,† S. Biscans,56,† M. Bitossi,63,† C. Biwer,59,† M. A. Bizouard,82,† E. Black,46,†

J. K. Blackburn,46,† L. Blackburn,83,† D. Blair,84,† S. Bloemen,55,85,† O. Bock,54,† T. P. Bodiya,56,† M. Boer,86,†

PHYSICAL REVIEW D 90, 102002 (2014)

1550-7998=2014=90(10)=102002(22) 102002-1 © 2014 American Physical Society



G. Bogaert,86,† C. Bogan,54,† P. Bojtos,87,† C. Bond,68,† F. Bondu,88,† L. Bonelli,76,63,† R. Bonnand,89,† R. Bork,46,†

M. Born,54,† V. Boschi,63,† Sukanta Bose,90,91,† L. Bosi,92,† C. Bradaschia,63,† P. R. Brady,60,93,† V. B. Braginsky,81,†

M. Branchesi,94,95,† J. E. Brau,96,† T. Briant,97,† D. O. Bridges,51,† A. Brillet,86,† M. Brinkmann,54,† V. Brisson,82,†

A. F. Brooks,46,† D. A. Brown,59,† D. D. Brown,68,† F. Brückner,68,† S. Buchman,64,† A. Buikema,56,† T. Bulik,78,†

H. J. Bulten,98,55,† A. Buonanno,99,† R. Burman,84,† D. Buskulic,89,† C. Buy,74,† L. Cadonati,100,52,† G. Cagnoli,101,†

J. Calderón Bustillo,102,† E. Calloni,103,49,† J. B. Camp,83,† P. Campsie,73,† K. C. Cannon,104,† B. Canuel,71,† J. Cao,105,†

C. D. Capano,99,† F. Carbognani,71,† L. Carbone,68,† S. Caride,106,† G. Castaldi,53,† S. Caudill,60,† M. Cavaglià,65,†

F. Cavalier,82,† R. Cavalieri,71,† C. Celerier,64,† G. Cella,63,† C. Cepeda,46,† E. Cesarini,107,† R. Chakraborty,46,†

T. Chalermsongsak,46,† S. J. Chamberlin,60,† S. Chao,108,† P. Charlton,109,† E. Chassande-Mottin,74,† X. Chen,84,†

Y. Chen,110,† A. Chincarini,80,† A. Chiummo,71,† H. S. Cho,111,† M. Cho,99,† J. H. Chow,112,† N. Christensen,113,† Q. Chu,84,†

S. S. Y. Chua,112,† S. Chung,84,† G. Ciani,50,† F. Clara,72,† D. E. Clark,64,† J. A. Clark,100,† J. H. Clayton,60,† F. Cleva,86,†

E. Coccia,114,115,† P.-F. Cohadon,97,† A. Colla,116,67,† C. Collette,117,† M. Colombini,92,† L. Cominsky,118,†

M. Constancio, Jr.,57,† A. Conte,116,67,† D. Cook,72,† T. R. Corbitt,47,† N. Cornish,70,† A. Corsi,119,† C. A. Costa,57,†

M.W. Coughlin,120,† J.-P. Coulon,86,† S. Countryman,75,† P. Couvares,59,† D.M. Coward,84,† M. J. Cowart,51,†

D. C. Coyne,46,† R. Coyne,119,† K. Craig,73,† J. D. E. Creighton,60,† R. P. Croce,53,† S. G. Crowder,121,† A. Cumming,73,†

L. Cunningham,73,† E. Cuoco,71,† C. Cutler,110,† K. Dahl,54,† T. Dal Canton,54,† M. Damjanic,54,† S. L. Danilishin,84,†

S. D’Antonio,107,† K. Danzmann,61,54,† V. Dattilo,71,† H. Daveloza,122,† M. Davier,82,† G. S. Davies,73,† E. J. Daw,123,†

R. Day,71,† T. Dayanga,90,† D. DeBra,64,† G. Debreczeni,124,† J. Degallaix,101,† S. Deléglise,97,† W. Del Pozzo,68,†

W. Del Pozzo,55,† T. Denker,54,† T. Dent,54,† H. Dereli,86,† V. Dergachev,46,† R. De Rosa,103,49,† R. T. DeRosa,47,†

R. DeSalvo,53,† S. Dhurandhar,91,† M. Díaz,122,† J. Dickson,112,† L. Di Fiore,49,† A. Di Lieto,76,63,† I. Di Palma,54,†

A. Di Virgilio,63,† V. Dolique,101,† E. Dominguez,125,† F. Donovan,56,† K. L. Dooley,54,† S. Doravari,51,† R. Douglas,73,†

T. P. Downes,60,† M. Drago,126,127,† R.W. P. Drever,46,† J. C. Driggers,46,† Z. Du,105,† M. Ducrot,89,† S. Dwyer,72,†

T. Eberle,54,† T. Edo,123,† M. Edwards,52,† A. Effler,47,† H.-B. Eggenstein,54,† P. Ehrens,46,† J. Eichholz,50,†

S. S. Eikenberry,50,† G. Endrőczi,124,† R. Essick,56,† T. Etzel,46,† M. Evans,56,† T. Evans,51,† M. Factourovich,75,†

V. Fafone,114,107,† S. Fairhurst,52,† X. Fan,73,† Q. Fang,84,† S. Farinon,80,† B. Farr,128,† W.M. Farr,68,† M. Favata,129,†

D. Fazi,128,† H. Fehrmann,54,† M.M. Fejer,64,† D. Feldbaum,50,51,† F. Feroz,120,† I. Ferrante,76,63,† E. C. Ferreira,57,†

F. Ferrini,71,† F. Fidecaro,76,63,† L. S. Finn,130,† I. Fiori,71,† R. P. Fisher,59,† R. Flaminio,101,† J.-D. Fournier,86,† S. Franco,82,†

S. Frasca,116,67,† F. Frasconi,63,† M. Frede,54,† Z. Frei,131,† A. Freise,68,† R. Frey,96,† T. T. Fricke,54,† P. Fritschel,56,†

V. V. Frolov,51,† P. Fulda,50,† M. Fyffe,51,† J. R. Gair,120,† L. Gammaitoni,132,92,† S. Gaonkar,91,† F. Garufi,103,49,†

N. Gehrels,83,† G. Gemme,80,† B. Gendre,86,† E. Genin,71,† A. Gennai,63,† S. Ghosh,55,85,† J. A. Giaime,51,47,†

K. D. Giardina,51,† A. Giazotto,63,† J. Gleason,50,† E. Goetz,54,† R. Goetz,50,† L. Gondan,131,† G. González,47,† N. Gordon,73,†

M. L. Gorodetsky,81,† S. Gossan,110,† S. Goßler,54,† R. Gouaty,89,† C. Gräf,73,† P. B. Graff,83,† M. Granata,101,† A. Grant,73,†

S. Gras,56,† C. Gray,72,† R. J. S. Greenhalgh,133,† A.M. Gretarsson,134,† P. Groot,85,† H. Grote,54,† K. Grover,68,†

S. Grunewald,69,† G.M. Guidi,94,95,† C. J. Guido,51,† K. Gushwa,46,† E. K. Gustafson,46,† R. Gustafson,106,† J. Ha,135,†

E. D. Hall,46,† W. Hamilton,47,† D. Hammer,60,† G. Hammond,73,† M. Hanke,54,† J. Hanks,72,† C. Hanna,136,130,†

M. D. Hannam,52,† J. Hanson,51,† J. Harms,46,† G.M. Harry,137,† I. W. Harry,59,† E. D. Harstad,96,† M. Hart,73,†

M. T. Hartman,50,† C.-J. Haster,68,† K. Haughian,73,† A. Heidmann,97,† M. Heintze,50,51,† H. Heitmann,86,† P. Hello,82,†

G. Hemming,71,† M. Hendry,73,† I. S. Heng,73,† A.W. Heptonstall,46,† M. Heurs,54,† M. Hewitson,54,† S. Hild,73,†

D. Hoak,100,† K. A. Hodge,46,† D. Hofman,101,† K. Holt,51,† P. Hopkins,52,† T. Horrom,138,† D. Hoske,139,† D. J. Hosken,139,†

J. Hough,73,† E. J. Howell,84,† Y. Hu,73,† E. Huerta,59,† B. Hughey,134,† S. Husa,102,† S. H. Huttner,73,† M. Huynh,60,†

T. Huynh-Dinh,51,† A. Idrisy,130,† D. R. Ingram,72,† R. Inta,130,† G. Islas,66,† T. Isogai,56,† A. Ivanov,46,† B. R. Iyer,140,†

K. Izumi,72,† M. Jacobson,46,† H. Jang,141,† P. Jaranowski,142,† Y. Ji,105,† F. Jiménez-Forteza,102,† W.W. Johnson,47,†

D. I. Jones,143,† R. Jones,73,† R. J. G. Jonker,55,† L. Ju,84,† Haris K,144,† P. Kalmus,46,† V. Kalogera,128,† S. Kandhasamy,65,†

G. Kang,141,† J. B. Kanner,46,† J. Karlen,100,† M. Kasprzack,82,71,† E. Katsavounidis,56,† W. Katzman,51,† H. Kaufer,61,†

S. Kaufer,61,† T. Kaur,84,† K. Kawabe,72,† F. Kawazoe,54,† F. Kéfélian,86,† G.M. Keiser,64,† D. Keitel,54,† D. B. Kelley,59,†

W. Kells,46,† D. G. Keppel,54,† A. Khalaidovski,54,† F. Y. Khalili,81,† E. A. Khazanov,145,† C. Kim,135,141,† K. Kim,146,†

N. G. Kim,141,† N. Kim,64,† S. Kim,141,† Y.-M. Kim,111,† E. J. King,139,† P. J. King,46,† D. L. Kinzel,51,† J. S. Kissel,72,†

S. Klimenko,50,† J. Kline,60,† S. Koehlenbeck,54,† K. Kokeyama,47,† V. Kondrashov,46,† S. Koranda,60,† W. Z. Korth,46,†

I. Kowalska,78,† D. B. Kozak,46,† V. Kringel,54,† A. Królak,147,148,† G. Kuehn,54,† A. Kumar,149,† D. Nanda Kumar,50,†

P. Kumar,59,† R. Kumar,73,† L. Kuo,108,† A. Kutynia,147,† P. K. Lam,112,† M. Landry,72,† B. Lantz,64,† S. Larson,128,†

P. D. Lasky,150,† A. Lazzarini,46,† C. Lazzaro,151,† P. Leaci,69,† S. Leavey,73,† E. O. Lebigot,105,† C. H. Lee,111,† H. K. Lee,146,†

H.M. Lee,135,† J. Lee,146,† P. J. Lee,56,† M. Leonardi,126,127,† J. R. Leong,54,† A. Le Roux,51,† N. Leroy,82,† N. Letendre,89,†

Y. Levin,152,† B. Levine,72,† J. Lewis,46,† T. G. F. Li,46,† K. Libbrecht,46,† A. Libson,56,† A. C. Lin,64,† T. B. Littenberg,128,†

N. A. Lockerbie,153,† V. Lockett,66,† D. Lodhia,68,† K. Loew,134,† J. Logue,73,† A. L. Lombardi,100,† E. Lopez,154,†

M. Lorenzini,114,107,† V. Loriette,155,† M. Lormand,51,† G. Losurdo,95,† J. Lough,59,† M. J. Lubinski,72,† H. Lück,61,54,†

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-2



A. P. Lundgren,54,† Y. Ma,84,† E. P. Macdonald,52,† T. MacDonald,64,† B. Machenschalk,54,† M. MacInnis,56,†

D.M. Macleod,47,† F. Magaña-Sandoval,59,† R. Magee,90,† M. Mageswaran,46,† C. Maglione,125,† K. Mailand,46,†

E. Majorana,67,† I. Maksimovic,155,† V. Malvezzi,114,107,† N. Man,86,† G.M. Manca,54,† I. Mandel,68,† V. Mandic,121,†

V. Mangano,116,67,† N.M. Mangini,100,† G. Mansell,112,† M. Mantovani,63,† F. Marchesoni,156,92,† F. Marion,89,† S. Márka,75,†

Z. Márka,75,† A. Markosyan,64,† E. Maros,46,† J. Marque,71,† F. Martelli,94,95,† I. W. Martin,73,† R.M. Martin,50,†

L. Martinelli,86,† D. Martynov,46,† J. N. Marx,46,† K. Mason,56,† A. Masserot,89,† T. J. Massinger,59,† F. Matichard,56,†

L. Matone,75,† N. Mavalvala,56,† G. May,47,† N. Mazumder,144,† G. Mazzolo,54,† R. McCarthy,72,† D. E. McClelland,112,†

S. C. McGuire,157,† G. McIntyre,46,† J. McIver,100,† K. McLin,118,† D. Meacher,86,† G. D. Meadors,106,† M. Mehmet,54,†

J. Meidam,55,† M. Meinders,61,† A. Melatos,150,† G. Mendell,72,† R. A. Mercer,60,† S. Meshkov,46,† C. Messenger,73,†

M. S. Meyer,51,† P. M. Meyers,121,† F. Mezzani,67,116,† H. Miao,110,† C. Michel,101,† E. E. Mikhailov,138,† L. Milano,103,49,†

J. Miller,56,† Y. Minenkov,107,† C.M. F. Mingarelli,68,† C. Mishra,144,† S. Mitra,91,† V. P. Mitrofanov,81,† G. Mitselmakher,50,†

R. Mittleman,56,† B. Moe,60,† A. Moggi,63,† M. Mohan,71,† S. R. P. Mohapatra,59,† D. Moraru,72,† G. Moreno,72,†

N. Morgado,101,† S. R. Morriss,122,† K. Mossavi,54,† B. Mours,89,† C.M. Mow-Lowry,54,† C. L. Mueller,50,† G. Mueller,50,†

S. Mukherjee,122,† A. Mullavey,47,† J. Munch,139,† D. Murphy,75,† P. G. Murray,73,† A. Mytidis,50,† M. F. Nagy,124,†

I. Nardecchia,114,107,† L. Naticchioni,116,67,† R. K. Nayak,158,† V. Necula,50,† G. Nelemans,55,85,† I. Neri,132,92,† M. Neri,79,80,†

G. Newton,73,† T. Nguyen,112,† A. B. Nielsen,54,† S. Nissanke,110,† A. H. Nitz,59,† F. Nocera,71,† D. Nolting,51,†

M. E. N. Normandin,122,† L. K. Nuttall,60,† E. Ochsner,60,† J. O’Dell,133,† E. Oelker,56,† J. J. Oh,159,† S. H. Oh,159,†

F. Ohme,52,† S. Omar,64,† P. Oppermann,54,† R. Oram,51,† B. O’Reilly,51,† W. Ortega,125,† R. O’Shaughnessy,60,†

C. Osthelder,46,† D. J. Ottaway,139,† R. S. Ottens,50,† H. Overmier,51,† B. J. Owen,130,† C. Padilla,66,† A. Pai,144,†

O. Palashov,145,† C. Palomba,67,† H. Pan,108,† Y. Pan,99,† C. Pankow,60,† F. Paoletti,71,63,† M. A. Papa,60,69,† H. Paris,64,†

A. Pasqualetti,71,† R. Passaquieti,76,63,† D. Passuello,63,† M. Pedraza,46,† A. Pele,72,† S. Penn,160,† A. Perreca,59,†

M. Phelps,46,† M. Pichot,86,† M. Pickenpack,54,† F. Piergiovanni,94,95,† V. Pierro,53,† L. Pinard,101,† I. M. Pinto,53,†

M. Pitkin,73,† J. Poeld,54,† R. Poggiani,76,63,† A. Poteomkin,145,† J. Powell,73,† J. Prasad,91,† V. Predoi,52,†

S. Premachandra,152,† T. Prestegard,121,† L. R. Price,46,† M. Prijatelj,71,† S. Privitera,46,† G. A. Prodi,126,127,† L. Prokhorov,81,†

O. Puncken,122,† M. Punturo,92,† P. Puppo,67,† M. Pürrer,52,† J. Qin,84,† V. Quetschke,122,† E. Quintero,46,†

R. Quitzow-James,96,† F. J. Raab,72,† D. S. Rabeling,98,55,† I. Rácz,124,† H. Radkins,72,† P. Raffai,131,† S. Raja,161,†

G. Rajalakshmi,162,† M. Rakhmanov,122,† C. Ramet,51,† K. Ramirez,122,† P. Rapagnani,116,67,† V. Raymond,46,†

M. Razzano,76,63,† V. Re,114,107,† S. Recchia,114,115,† C.M. Reed,72,† T. Regimbau,86,† S. Reid,163,† D. H. Reitze,46,50,†

O. Reula,125,† E. Rhoades,134,† F. Ricci,116,67,† R. Riesen,51,† K. Riles,106,† N. A. Robertson,46,73,† F. Robinet,82,†

A. Rocchi,107,† S. B. Roddy,51,† L. Rolland,89,† J. G. Rollins,46,† R. Romano,48,49,† G. Romanov,138,† J. H. Romie,51,†

D. Rosińska,164,77,† S. Rowan,73,† A. Rüdiger,54,† P. Ruggi,71,† K. Ryan,72,† F. Salemi,54,† L. Sammut,150,† V. Sandberg,72,†

J. R. Sanders,106,† S. Sankar,56,† V. Sannibale,46,† I. Santiago-Prieto,73,† E. Saracco,101,† B. Sassolas,101,†

B. S. Sathyaprakash,52,† P. R. Saulson,59,† R. Savage,72,† J. Scheuer,128,† R. Schilling,54,† M. Schilman,125,† P. Schmidt,52,†

R. Schnabel,54,61,† R.M. S. Schofield,96,† E. Schreiber,54,† D. Schuette,54,† B. F. Schutz,52,69,† J. Scott,73,† S. M. Scott,112,†

D. Sellers,51,† A. S. Sengupta,165,† D. Sentenac,71,† V. Sequino,114,107,† A. Sergeev,145,† D. A. Shaddock,112,† S. Shah,55,85,†

M. S. Shahriar,128,† M. Shaltev,54,† Z. Shao,46,† B. Shapiro,64,† P. Shawhan,99,† D. H. Shoemaker,56,† T. L. Sidery,68,†

K. Siellez,86,† X. Siemens,60,† D. Sigg,72,† D. Simakov,54,† A. Singer,46,† L. Singer,46,† R. Singh,47,† A.M. Sintes,102,†

B. J. J. Slagmolen,112,† J. Slutsky,83,† J. R. Smith,66,† M. R. Smith,46,† R. J. E. Smith,46,† N. D. Smith-Lefebvre,46,†

E. J. Son,159,† B. Sorazu,73,† T. Souradeep,91,† A. Staley,75,† J. Stebbins,64,† M. Steinke,54,† J. Steinlechner,54,73,†

S. Steinlechner,54,73,† B. C. Stephens,60,† S. Steplewski,90,† S. Stevenson,68,† R. Stone,122,† D. Stops,68,† K. A. Strain,73,†

N. Straniero,101,† S. Strigin,81,† R. Sturani,166,† A. L. Stuver,51,† T. Z. Summerscales,167,† S. Susmithan,84,† P. J. Sutton,52,†

B. Swinkels,71,† M. Tacca,74,† D. Talukder,96,† D. B. Tanner,50,† J. Tao,47,† S. P. Tarabrin,54,† R. Taylor,46,† G. Tellez,122,†

M. P. Thirugnanasambandam,46,† M. Thomas,51,† P. Thomas,72,† K. A. Thorne,51,† K. S. Thorne,110,† E. Thrane,46,†

V. Tiwari,50,† K. V. Tokmakov,153,† C. Tomlinson,123,† M. Tonelli,76,63,† C. V. Torres,122,† C. I. Torrie,46,73,† F. Travasso,132,92,†

G. Traylor,51,† M. Tse,75,† D. Tshilumba,117,† H. Tuennermann,54,† D. Ugolini,168,† C. S. Unnikrishnan,162,† A. L. Urban,60,†

S. A. Usman,59,† H. Vahlbruch,61,† G. Vajente,76,63,† G. Valdes,122,† M. Vallisneri,110,† M. van Beuzekom,55,†

J. F. J. van den Brand,98,55,† C. Van Den Broeck,55,† M. V. van der Sluys,55,85,† J. van Heijningen,55,† A. A. van Veggel,73,†

S. Vass,46,† M. Vasúth,124,† R. Vaulin,56,† A. Vecchio,68,† G. Vedovato,151,† J. Veitch,55,† P. J. Veitch,139,†

K. Venkateswara,169,† D. Verkindt,89,† F. Vetrano,94,95,† A. Viceré,94,95,† R. Vincent-Finley,157,† J.-Y. Vinet,86,† S. Vitale,56,†

T. Vo,72,† H. Vocca,132,92,† C. Vorvick,72,† W. D. Vousden,68,† S. P. Vyachanin,81,† A. R. Wade,112,† L. Wade,60,† M. Wade,60,†

M. Walker,47,† L. Wallace,46,† S. Walsh,60,† M. Wang,68,† X. Wang,105,† R. L. Ward,112,† M. Was,54,† B. Weaver,72,†

L.-W. Wei,86,† M. Weinert,54,† A. J. Weinstein,46,† R. Weiss,56,† T. Welborn,51,† L. Wen,84,† P. Wessels,54,† M. West,59,†

T. Westphal,54,† K. Wette,54,† J. T. Whelan,170,† D. J. White,123,† B. F. Whiting,50,† K. Wiesner,54,† C. Wilkinson,72,†

K. Williams,157,† L. Williams,50,† R. Williams,46,† T. D. Williams,171,† A. R. Williamson,52,† J. L. Willis,172,† B. Willke,61,54,†

M. Wimmer,54,† W. Winkler,54,† C. C. Wipf,56,† A. G. Wiseman,60,† H. Wittel,54,† G. Woan,73,† N. Wolovick,125,†

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-3



J. Worden,72,† Y. Wu,50,† J. Yablon,128,† I. Yakushin,51,† W. Yam,56,† H. Yamamoto,46,† C. C. Yancey,99,† H. Yang,110,†

S. Yoshida,171,† M. Yvert,89,† A. Zadrożny,147,† M. Zanolin,134,† J.-P. Zendri,151,† Fan Zhang,56,105,† L. Zhang,46,† C. Zhao,84,†

H. Zhu,130,† X. J. Zhu,84,† M. E. Zucker,56,† S. Zuraw,100,† and J. Zweizig46,†

(IceCube Collaboration)*

(LIGO Scientific Collaboration and the Virgo Collaboration)†

1III. Physikalisches Institut, RWTH Aachen University, D-52056 Aachen, Germany
2School of Chemistry & Physics, University of Adelaide, Adelaide, South Australia 5005, Australia

3Department of Physics and Astronomy, University of Alaska Anchorage,
3211 Providence Drive, Anchorage, Alaska 99508, USA

4CTSPS, Clark-Atlanta University, Atlanta, Georgia 30314, USA
5School of Physics and Center for Relativistic Astrophysics, Georgia Institute of Technology,

Atlanta, Georgia 30332, USA
6Department of Physics, Southern University, Baton Rouge, Louisiana 70813, USA
7Department of Physics, University of California, Berkeley, California 94720, USA

8Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA
9Institut für Physik, Humboldt-Universität zu Berlin, D-12489 Berlin, Germany

10Fakultät für Physik & Astronomie, Ruhr-Universität Bochum, D-44780 Bochum, Germany
11Physikalisches Institut, Universität Bonn, Nussallee 12, D-53115 Bonn, Germany
12Université Libre de Bruxelles, Science Faculty CP230, B-1050 Brussels, Belgium

13Vrije Universiteit Brussel, Dienst ELEM, B-1050 Brussels, Belgium
14Department of Physics, Chiba University, Chiba 263-8522, Japan
15Department of Physics and Astronomy, University of Canterbury,

Private Bag 4800, Christchurch, New Zealand
16Department of Physics, University of Maryland, College Park, Maryland 20742, USA

17Department of Physics and Center for Cosmology and Astro-Particle Physics, Ohio State University,
Columbus, Ohio 43210, USA

18Department of Astronomy, Ohio State University, Columbus, Ohio 43210, USA
19Niels Bohr Institute, University of Copenhagen, DK-2100 Copenhagen, Denmark
20Department of Physics, TU Dortmund University, D-44221 Dortmund, Germany

21Department of Physics, University of Alberta, Edmonton, Alberta T6G 2E1, Canada
22Erlangen Centre for Astroparticle Physics, Friedrich-Alexander-Universität Erlangen-Nürnberg,

D-91058 Erlangen, Germany
23Département de physique nucléaire et corpusculaire, Université de Genève,

CH-1211 Genève, Switzerland
24Department of Physics and Astronomy, University of Gent, B-9000 Gent, Belgium

25Department of Physics and Astronomy, University of California, Irvine, California 92697, USA
26Department of Physics and Astronomy, University of Kansas, Lawrence, Kansas 66045, USA

27Department of Astronomy, University of Wisconsin, Madison, Wisconsin 53706, USA
28Department of Physics and Wisconsin IceCube Particle Astrophysics Center, University of Wisconsin,

Madison, Wisconsin 53706, USA
29Institute of Physics, University of Mainz, Staudinger Weg 7, D-55099 Mainz, Germany

30Université de Mons, 7000 Mons, Belgium
31Technische Universität München, D-85748 Garching, Germany

32Bartol Research Institute and Department of Physics and Astronomy, University of Delaware,
Newark, Delaware 19716, USA

33Department of Physics, University of Oxford, 1 Keble Road, Oxford OX1 3NP, United Kingdom
34Physics Department, South Dakota School of Mines and Technology,

Rapid City, South Dakota 57701, USA
35Department of Physics, University of Wisconsin, River Falls, Wisconsin 54022, USA

36Oskar Klein Centre and Department of Physics, Stockholm University, SE-10691 Stockholm, Sweden
37Department of Physics and Astronomy, Stony Brook University,

Stony Brook, New York 11794-3800, USA
38Department of Physics, Sungkyunkwan University, Suwon 440-746, Korea

39Department of Physics, University of Toronto, Toronto, Ontario M5S 1A7, Canada
40Department of Physics and Astronomy, University of Alabama, Tuscaloosa, Alabama 35487, USA

41Department of Astronomy and Astrophysics, Pennsylvania State University,
University Park, Pennsylvania 16802, USA

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-4



42Department of Physics, Pennsylvania State University, University Park, Pennsylvania 16802, USA
43Department of Physics and Astronomy, Uppsala University, Box 516, S-75120 Uppsala, Sweden

44Department of Physics, University of Wuppertal, D-42119 Wuppertal, Germany
45DESY, D-15735 Zeuthen, Germany

46LIGO, California Institute of Technology, Pasadena, California 91125, USA
47Louisiana State University, Baton Rouge, Louisiana 70803, USA

48Università di Salerno, Fisciano, I-84084 Salerno, Italy
49INFN, Sezione di Napoli, Complesso Universitario di Monte S.Angelo, I-80126 Napoli, Italy

50University of Florida, Gainesville, Florida 32611, USA
51LIGO Livingston Observatory, Livingston, Louisiana 70754, USA

52Cardiff University, Cardiff, CF24 3AA, United Kingdom
53University of Sannio at Benevento, I-82100 Benevento, Italy,

and INFN, Sezione di Napoli, I-80100 Napoli, Italy
54Albert-Einstein-Institut, Max-Planck-Institut für Gravitationsphysik, D-30167 Hannover, Germany

55Nikhef, Science Park, 1098 XG Amsterdam, The Netherlands
56LIGO, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA

57Instituto Nacional de Pesquisas Espaciais, 12227-010 São José dos Campos, SP, Brazil
58International Centre for Theoretical Sciences, Tata Institute of Fundamental Research,

Bangalore 560012, India
59Syracuse University, Syracuse, New York 13244, USA

60University of Wisconsin–Milwaukee, Milwaukee, Wisconsin 53201, USA
61Leibniz Universität Hannover, D-30167 Hannover, Germany

62Università di Siena, I-53100 Siena, Italy
63INFN, Sezione di Pisa, I-56127 Pisa, Italy

64Stanford University, Stanford, California 94305, USA
65The University of Mississippi, University, Mississippi 38677, USA

66California State University Fullerton, Fullerton, California 92831, USA
67INFN, Sezione di Roma, I-00185 Roma, Italy

68University of Birmingham, Birmingham B15 2TT, United Kingdom
69Albert-Einstein-Institut, Max-Planck-Institut für Gravitationsphysik, D-14476 Golm, Germany

70Montana State University, Bozeman, Montana 59717, USA
71European Gravitational Observatory (EGO), I-56021 Cascina, Pisa, Italy

72LIGO Hanford Observatory, Richland, Washington 99352, USA
73SUPA, University of Glasgow, Glasgow G12 8QQ, United Kingdom

74APC, AstroParticule et Cosmologie, Université Paris Diderot, CNRS/IN2P3, CEA/Irfu, Observatoire de
Paris, Sorbonne Paris Cité, 10, rue Alice Domon et Léonie Duquet, F-75205 Paris Cedex 13, France

75Columbia University, New York, New York 10027, USA
76Università di Pisa, I-56127 Pisa, Italy
77CAMK-PAN, 00-716 Warsaw, Poland

78Astronomical Observatory Warsaw University, 00-478 Warsaw, Poland
79Università degli Studi di Genova, I-16146 Genova, Italy

80INFN, Sezione di Genova, I-16146 Genova, Italy
81Faculty of Physics, Lomonosov Moscow State University, Moscow 119991, Russia

82LAL, Université Paris-Sud, IN2P3/CNRS, F-91898 Orsay, France
83NASA/Goddard Space Flight Center, Greenbelt, Maryland 20771, USA

84University of Western Australia, Crawley, Western Australia 6009, Australia
85Department of Astrophysics/IMAPP, Radboud University Nijmegen,

P.O. Box 9010, 6500 GL Nijmegen, The Netherlands
86Université Nice-Sophia-Antipolis, CNRS, Observatoire de la Côte d’Azur, F-06304 Nice, France

87MTA Atomki, Debrecen, Hungary
88Institut de Physique de Rennes, CNRS, Université de Rennes 1, F-35042 Rennes, France
89Laboratoire d’Annecy-le-Vieux de Physique des Particules (LAPP), Université de Savoie,

CNRS/IN2P3, F-74941 Annecy-le-Vieux, France
90Washington State University, Pullman, Washington 99164, USA

91Inter-University Centre for Astronomy and Astrophysics, Pune 411007, India
92INFN, Sezione di Perugia, I-06123 Perugia, Italy

93Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-8502, Japan
94Università degli Studi di Urbino “Carlo Bo,” I-61029 Urbino, Italy
95INFN, Sezione di Firenze, I-50019 Sesto Fiorentino, Firenze, Italy

96University of Oregon, Eugene, Oregon 97403, USA

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-5



97Laboratoire Kastler Brossel, ENS, CNRS, UPMC, Université Pierre et Marie Curie,
F-75005 Paris, France

98VU University Amsterdam, 1081 HV Amsterdam, The Netherlands
99University of Maryland, College Park, Maryland 20742, USA

100University of Massachusetts Amherst, Amherst, Massachusetts 01003, USA
101Laboratoire des Matériaux Avancés (LMA), IN2P3/CNRS, Université de Lyon,

F-69622 Villeurbanne, Lyon, France
102Universitat de les Illes Balears, E-07122 Palma de Mallorca, Spain

103Università di Napoli ’Federico II’, Complesso Universitario di Monte S. Angelo, I-80126 Napoli, Italy
104Canadian Institute for Theoretical Astrophysics, University of Toronto,

Toronto, Ontario M5S 3H8, Canada
105Tsinghua University, Beijing 100084, China

106University of Michigan, Ann Arbor, Michigan 48109, USA
107INFN, Sezione di Roma Tor Vergata, I-00133 Roma, Italy

108National Tsing Hua University, Hsinchu 300, Taiwan
109Charles Sturt University, Wagga Wagga, NSW 2678, Australia

110Caltech-CaRT, Pasadena, California 91125, USA
111Pusan National University, Busan 609-735, Korea

112Australian National University, Canberra, ACT 0200, Australia
113Carleton College, Northfield, Minnesota 55057, USA
114Università di Roma Tor Vergata, I-00133 Roma, Italy

115INFN, Gran Sasso Science Institute, I-67100L’Aquila, Italy
116Università di Roma ’La Sapienza’, I-00185 Roma, Italy

117University of Brussels, Brussels 1050 Belgium
118Sonoma State University, Rohnert Park, California 94928, USA

119The George Washington University, Washington, DC 20052, USA
120University of Cambridge, Cambridge CB2 1TN, United Kingdom
121University of Minnesota, Minneapolis, Minnesota 55455, USA

122The University of Texas at Brownsville, Brownsville, Texas 78520, USA
123The University of Sheffield, Sheffield S10 2TN, United Kingdom

124Wigner RCP, RMKI, H-1121 Budapest, Konkoly Thege Miklós út 29-33, Hungary
125Argentinian Gravitational Wave Group, Cordoba Cordoba 5000, Argentina

126Università di Trento, I-38050 Povo, Trento, Italy
127INFN, Gruppo Collegato di Trento, I-38050 Povo, Trento, Italy

128Northwestern University, Evanston, Illinois 60208, USA
129Montclair State University, Montclair, New Jersey 07043, USA

130The Pennsylvania State University, University Park, Pennsylvania 16802, USA
131MTA Eötvös University, ‘Lendulet’ A. R. G., Budapest 1117, Hungary

132Università di Perugia, I-06123 Perugia, Italy
133Rutherford Appleton Laboratory, HSIC, Chilton, Didcot, Oxon OX11 0QX, United Kingdom

134Embry-Riddle Aeronautical University, Prescott, AZ 86301, USA
135Seoul National University, Seoul 151-742, Korea

136Perimeter Institute for Theoretical Physics, Waterloo, Ontario N2L 2Y5, Canada
137American University, Washington, DC 20016, USA

138College of William and Mary, Williamsburg, Virginia 23187, USA
139University of Adelaide, Adelaide, South Australia 5005, Australia
140Raman Research Institute, Bangalore, Karnataka 560080, India

141Korea Institute of Science and Technology Information, Daejeon 305-806, Korea
142Biał ystok University, 15-424 Biał ystok, Poland

143University of Southampton, Southampton SO17 1BJ, United Kingdom
144IISER-TVM, CET Campus, Trivandrum 695016, Kerala, India
145Institute of Applied Physics, Nizhny Novgorod 603950, Russia

146Hanyang University, Seoul 133-791, Korea
147NCBJ, 05-400 Świerk-Otwock, Poland

148IM-PAN, 00-956 Warsaw, Poland
149Institute for Plasma Research, Bhat, Gandhinagar 382428, India
150The University of Melbourne, Parkville, Victoria 3010, Australia

151INFN, Sezione di Padova, I-35131 Padova, Italy
152Monash University, Victoria 3800, Australia

153SUPA, University of Strathclyde, Glasgow G1 1XQ, United Kingdom

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-6



154Louisiana Tech University, Ruston, Louisiana 71272, USA
155ESPCI, CNRS, F-75005 Paris, France

156Università di Camerino, Dipartimento di Fisica, I-62032 Camerino, Italy
157Southern University and A&M College, Baton Rouge, Louisiana 70813, USA

158IISER-Kolkata, Mohanpur 741252, West Bengal, India
159National Institute for Mathematical Sciences, Daejeon 305-390, Korea
160Hobart and William Smith Colleges, Geneva, New York 14456, USA

161RRCAT, Indore MP 452013, India
162Tata Institute for Fundamental Research, Mumbai 400005, India

163SUPA, University of the West of Scotland, Paisley PA1 2BE, United Kingdom
164Institute of Astronomy, 65-265 Zielona Góra, Poland

165Indian Institute of Technology, Gandhinagar Ahmedabad 382424, Gujarat, India
166Instituto de Física Teórica, Universidade Estadual Paulista/ICTP South American Institute for

Fundamental Research, São Paulo SP 01140-070, Brazil
167Andrews University, Berrien Springs, Michigan 49104, USA

168Trinity University, San Antonio, Texas 78212, USA
169University of Washington, Seattle, Washington 98195, USA

170Rochester Institute of Technology, Rochester, New York 14623, USA
171Southeastern Louisiana University, Hammond, Louisiana 70402, USA

172Abilene Christian University, Abilene, Texas 79699, USA
(Received 14 July 2014; published 17 November 2014)

We report the results of a multimessenger search for coincident signals from the LIGO and Virgo
gravitational-wave observatories and the partially completed IceCube high-energy neutrino detector,
including periods of joint operation between 2007–2010. These include parts of the 2005–2007 run and the
2009–2010 run for LIGO-Virgo, and IceCube’s observation periods with 22, 59 and 79 strings. We find no
significant coincident events, and use the search results to derive upper limits on the rate of joint sources
for a range of source emission parameters. For the optimistic assumption of gravitational-wave emission
energy of 10−2 M⊙c2 at ∼150 Hz with ∼60 ms duration, and high-energy neutrino emission of 1051 erg
comparable to the isotropic gamma-ray energy of gamma-ray bursts, we limit the source rate below
1.6 × 10−2 Mpc−3 yr−1. We also examine how combining information from gravitational waves and
neutrinos will aid discovery in the advanced gravitational-wave detector era.

DOI: 10.1103/PhysRevD.90.102002 PACS numbers: 95.85.Sz, 95.85.Ry, 98.70.Rz

I. INTRODUCTION

Gravitational-wave (GW) and high-energy-neutrino
observations represent important novel opportunities in
studying a variety of cosmic processes, providing comple-
mentary information to electromagnetic searches. They
carry information on the origin of high-energy emission
as well as the dynamics and structure of the process driving
the emission [1–7], which may be difficult to decipher via
electromagnetic observations ([8,9] and references therein).
GWs are produced by the bulk motion of the progenitor,

typically carrying information on the dynamics of the
source’s central region. High-energy neutrinos, on the other
hand, require hadron acceleration in, e.g., relativistic out-
flows from a central engine. Astrophysical processes that
produce GWs may also drive relativistic outflows, which
can emit high-energy radiation, such as GeV-PeV neutrinos

or gamma rays. Detecting GWs along with other forms of
radiation from a common source would shed light on the
internal processes within the source, and could increase
detection confidence.
The search for common sources of GWs and high-

energy-neutrino has recently become possible with the
construction and upgrade of large-scale observatories. GW
detectors include LIGO [1] and Virgo [10], which are being
upgraded to second-generation detectors [11,12], and GEO
[13], which is currently acquiring data. Another advanced
GW detector, KAGRA [14], is being constructed in Japan.
The possibility to construct a third advanced LIGO obser-
vational facility in India is also currently being explored
[15]. With the construction of the advanced LIGO detectors
finishing in 2015, multiple detections of GWs from
compact binary mergers are expected after 2016–2018
[16]. High-energy-neutrino observatories currently in
operation include IceCube [17–19]—a cubic-kilometer
detector at the geographic South Pole—and ANTARES [20]
in the Mediterranean sea. ANTARES is planned to be
followed by a multi-cubic-kilometer detector in the
Mediterranean sea called KM3NeT in the following years

aVisitor from Earthquake Research Institute, University of
Tokyo, Bunkyo, Tokyo 113-0032, Japan.

bVisitor from NASA Goddard Space Flight Center, Greenbelt,
Maryland 20771, USA.

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-7

http://dx.doi.org/10.1103/PhysRevD.90.102002
http://dx.doi.org/10.1103/PhysRevD.90.102002
http://dx.doi.org/10.1103/PhysRevD.90.102002
http://dx.doi.org/10.1103/PhysRevD.90.102002


[21]. The Baikal Neutrino Telescope, operating at Lake
Baikal, is also planned to be upgraded to a km3 volume
[22]. IceCube is also sensitive to low energy (MeV)
thermal neutrinos from nearby supernovae [23], and
contributes to the Supernova Early Warning System
(SNEWS) network along with several other neutrino
detectors, including Super-Kamiokande [24], Borexino
[25], the Large Volume Detector [26] and the Sudbury
Neutrino Observatory [27].
High-energy neutrinos of cosmic origin have recently

been observed, for the first time, by IceCube [28,29]. Their
detection represents a major step towards multimessenger
astronomy. In an all-sky search for the most energetic
(> 30 TeV) neutrinos, a total of 28 astrophysical neutrino
candidates were identified over a period of 2 years,
significantly above the number of expected atmospheric
background (≈10 events). The neutrinos have no clear
directional or temporal clustering. While many sources
of astrophysical origin have been suggested (see, e.g.,
[30–39]), so far there is no compelling evidence for a
specific source population.
Several classes of astrophysical transients that are detected

electromagnetically are also associated with significant
GW and high-energy-neutrino emission. These processes
include the gamma-ray bursts (GRBs) [8,40–49], core-
collapse supernovae with rapidly rotating cores [50–52],
flares from soft-gamma repeaters [53–55], or even cosmic
string cusps [56–59].
GRBs are one of the most promising sources for joint

GWþ neutrino observations. GRBs are intense flashes of
∼ MeV photons of typically extragalactic origin, occurring
a few times a day in the observable universe [60]. They are
thought to be produced in relativistic outflows likely driven
by accretion onto a compact object [61,62], or by a rapidly
rotating neutron star [63]. Hadrons accelerated in the
relativistic outflows will result in the production of high-
energy neutrinos through interaction with photons or other
hadrons [42,43]. The central engines that drive the outflow
can be formed by (i) the collapse of the rapidly rotating core
of a massive star, which can also produce supernovae [64],
or by (ii) the merger of two neutron stars or a neutron star
and a black hole [65]. Stellar core collapse with rapidly
rotating cores are thought to be the origin of long (≳2s)
GRBs. They can lead to GW emission via, e.g., rotational
instabilities in the newly formed protoneutron stars, non-
axisymmetric instabilities of the accretion disk, or the
fragmentation of the collapsing core [66–68]. Compact
binary neutron star or neutron-star-black-hole mergers are
thought to be the progenitors of short (≲2s) GRBs [69,70].
They emit strong GWs as the two compact objects
inspiral and merge, in the sensitive frequency band of
LIGO/Virgo [71–73].
Another promising source type, soft-gamma repeaters

(SGR), are sources of short bursts of gamma rays emitted
at irregular intervals. They occasionally emit giant

gamma-ray/x-ray flares that can be detected from the
Milky Way or nearby galaxies within a few megaparsecs,
at a rate lower than the GRB rate (e.g., [74]). SGRs are
thought to be highly magnetized neutron stars. Their
gamma outbursts may be a consequence of the tectonic
activity of the neutron star crust (starquake). The seismic
vibration of a neutron star following a starquake gives rise
to GW emission (e.g., [55,75,76]). Tectonic activity in the
neutron star is followed by the reconfiguration of the
neutron star’s magnetic fields, resulting in x-ray and
gamma-ray radiation [77,78]. Sudden magnetic reconfig-
uration may also accelerate protons and other nuclei,
leading to the production of high-energy neutrinos [53,79].
For joint GWþ neutrino detection, a particularly inter-

esting subset of these sources are those that have faint
electromagnetic emission, such as choked GRBs [80–82]
for which the relativistic outflow stalls before it can break
out of the stelar envelope, and low-luminosity GRBs with
mildly relativistic ejecta [43,83].
The coincident detectability of GWs and high-energy

neutrinos, and the scientific potential of such searches, was
first recognized in [84], and later independently in [85].
The population of cosmic events producing both GWs and
high-energy-neutrino emission was first constrained obser-
vationally [86] with data from the initial LIGO-Virgo
detectors and the partially completed IceCube detector
(see Sec. III B). While the population of cosmic events
expected to produce detectable joint emission falls below
these early observational limits, the results were used to
estimate the projected upper limits achievable with
advanced GW detectors and the completed IceCube, which
can provide meaningful constraints or detection.
A search for coincident GWs and high-energy neutrinos

has recently been carried out using the initial LIGO-Virgo
observatories and the partially completed ANTARES detec-
tor [87]. The search found no significant coincident events.
ANTARES is mainly sensitive to TeV neutrino sources from
the southern hemisphere, making this a complementary
search to that performed with IceCube, which is mainly
sensitive to the northern hemisphere at TeV energies.
Recognizing the scientific potential of multimessenger

observations, various initiatives aim to use GWs and
neutrinos in combination with electromagnetic observa-
tions. For instance, multiple astrophysical event candidates
from both the initial LIGO-Virgo detectors [88,89] and
IceCube [90,91] have been followed up by EM telescopes:
QUEST, TAROT, ZADKO, Pi of the sky, ROTSE,
SkyMapper, the Palomar Transient Factory, and the
Swift satellite [92–95]. Sub-threshold GRB candidates
detected through gamma rays are also being used for
searches for coincident GW signals [96–98]. The
SNEWS network [99] aims to use subthreshold MeV
neutrino signals from neutrino observatories around the
world to trigger electromagnetic follow-up searches for
supernovae. The network utilizes the fact that neutrinos

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-8



arrive up to several hours earlier than the detectable
electromagnetic signal from supernovae, as was the case
for the first confirmed astronomical MeV neutrino source,
SN 1987A [100,101]. A current initiative for multimes-
senger searches in the weak-signal limit is the proposed
Astrophysical Multimessenger Observatory Network
(AMON; [102]). AMON is planned to perform a real-time
correlation analysis of subthreshold signals from available
astronomical messengers; photons, low and high-energy
neutrinos, cosmic rays, and GWs. These subthreshold
triggers will be used to (1) increase the detectable number
of sources and (2) trigger follow-up searches that can
further enhance the significance of a potential coincident
detection.
In this paper we present a multimessenger search for

joint sources of GWs and high-energy neutrinos using the
initial LIGO and Virgo GW detectors together with the
partially complete IceCube detector in its 22, 59 and
79-string configurations. The analysis is based on the
method described in [103]. In Sec. II we briefly review
the search method and data-analysis details. In Sec. III
we present the results. Observing no evidence of a
GWþ neutrino event we set upper limits on the rate of
GWþ neutrino sources as a function of the source energy
budget. We summarize our findings and discuss the outlook
for future observations in Sec. IV.

II. MULTIMESSENGER ANALYSIS

In this section we present the status of GW and neutrino
observations and the analysis we used to combine their
results; a more detailed description of the method can be
found in [103]. Hereafter, we use the word “neutrinos” to
refer exclusively to high-energy neutrinos.

A. Gravitational-wave data

LIGO [1] and Virgo [10] are km-scale interferometric
GW observatories. LIGO has two Michelson interferom-
eters in the U.S., both with 4-km arm length. One of the
detectors is in Hanford, Washington (which we refer to as
H1), while the other one is located in Livingston, Louisiana
(which we refer to as L1). During the 2005–2007 obser-
vation period, an additional detector with 2-km arm length
was operating at the Hanford site (which we refer to as H2).
H2’s location and orientation were identical to that of H1.
The LIGO detectors were sensitive to GWs in the frequency
band of 40–7000 Hz. Virgo is a Michelson interferometer
with 3-km arm length (which we refer to as V1), located
near Cascina, Italy. Virgo is sensitive to GWs in a frequency
band similar to that of LIGO. For the present search, data
from initial Virgo’s three observation periods (VSR1-3)
were used.
LIGO and Virgo detect GWs by monitoring the relative

displacement of test masses (mirrors). An incoming GW
changes the distance between test masses with an amount

proportional to their distance. The detectors measure the
fractional difference h ¼ ðL1 − L2Þ=L between the lengths
of the two interferometer arms L1 and L2 (L2 ≃ L1 ¼ L),
which is referred to as strain. The two GW polarizations,
denoted with hþ and h×, independently contribute to the
detector strain. The detector strain also depends on the
relative orientation of the detector and the GW, which is
described by so-called antenna response factors Fþ and F×:
hðtÞ ¼ FþhþðtÞ þ F×h×ðtÞ [104].
The astrophysical reach of GW detectors can be charac-

terized, e.g., by the so-called horizon distance: the maximum
distance at which the GW signal from an optimally oriented
and optimally located source can be detected, with detection
defined as a single-detector signal-to-noise ratio of 8. It is
easiest to establish this horizon distance for compact binary
mergers for which the gravitational waveform is known to
sufficient precision. This horizon distance reached tens of
megaparsecs for the initial LIGO-Virgo detectors [105,106].
The GW output, and therefore the horizon distances, of
stellar core collapse with rapidly rotating cores and SGR
starquakes are less certain [9]. Adopting an optimistic
standard-siren GW emission with energy Egw¼10−2M⊙c2
and frequency f0 ¼ 150 Hz (e.g., [107]), such a GW signal
could be detected out to ≳10 Mpc with initial LIGO-Virgo.
For SGRs, the GW counterpart of giant flares may be
detectable, in the most optimistic scenarios, from within the
Galaxy [108].
For the present search, data from the initial LIGO

detectors’ fifth (S5) and sixth (S6) observation periods
were used. During the S5/VSR1 observation period, a small
fraction of the data was taken with only the H1, L1 and V1
detectors, i.e. without H2. Given that this period was
significantly shorter than when {H1, H2, L1, V1} were
operational, and for simplicity, we exclude this period from
the present analysis.
The LIGO S5 observation period started on November 4,

2005 and ended on October 1, 2007, while S6 lasted from
July 7, 2009 until October 20, 2010. Virgo VSR1 started on
May 18, 2007 and ended on October 1, 2007. Virgo VSR2
ran from July 7, 2009 to January 11, 2010, while Virgo
VSR3 ran from August 11, 2010 to October 20, 2010. The
observational periods of LIGO and Virgo used in this
analysis are shown graphically in Fig. 1 in comparison to
IceCube. The three GW search periods have similar noise
spectra and have comparable sensitivities. The S6 period
having a nominal factor of ∼2 lower background noise
level than the S5 period.
In the present analysis, we only use data when all

available detectors were operational. This means that for
the S5 observation period, data is analyzed only if H1, H2,
L1 and V1 were operational, while for the S6 observation
period, we require H1, L1 and V1 to be operational. The
total duration of such data taking period for S5/VSR1, in
coincidence with IceCube in its 22-strings configuration
(IC22) is ∼64 days. There are ∼30 days of coincident data

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-9



between S6-VSR2 and IceCube in its 59-string configura-
tion (IC59) and ∼26 days of coincident data for S6-VSR3
and IceCube in its 79-string configuration (IC79).
We analyze GW data from the LIGO-Virgo observatories

using the coherent WaveBurst (cWB) analysis pipeline
[109]. The cWB pipeline searches for short GW transients
with durations ≲1s in the [64 Hz, 2048 Hz] frequency
range, which includes the most sensitive frequency band
of LIGO and Virgo (∼150 Hz). The pipeline looks for
“generic” GW signals within this frequency range, with no
further constraints on their waveform [110]. The cWB
pipeline coherently combines information from all GW
detectors to perform a constrained maximum likelihood
analysis. The output of the pipeline is a list of astrophysical
GW signal candidates, which wewill refer to as GWevents.
For each GW event, cWB records its (i) time of arrival,

(ii) test statistic ρ and (iii) directional distribution. The test
statistic of a GWevent is the coherent network amplitude ρ,
which is proportional to the signal-to-noise ratio. The
directional distribution is specified within 0.4° × 0.4° size
pixels.
The significance of each event is used to calculate the

GW false alarm rate (FARgw), that is, the frequency of
occurrence of events from background alone with a greater
significance. In this analysis we consider only those GW
events whose FARgw, based on their significance, is
≤ 1 day−1. This FARgw was used for electromagnetic
follow-up observations of GW events [92]. The ρ values
corresponding to FARgw ¼ 1 day−1 for the three observa-
tion periods are ρs5=vsr1 ≈ 2.7, ρs6=vsr2 ≈ 3.3, ρs6=vsr3 ≈ 3.4.
Note that the different value for S5/VSR1 is due to the fact
that the H2 detector is also used there.

B. High-energy-neutrino data

The IceCube Neutrino Observatory [17], located near
the South Pole, is a cubic-kilometer Cherenkov detector
optimized to detect neutrinos at the TeV-PeV energy scale.
The full detector consists of 86 vertical strings, with a
set of 60 digital optical modules (DOMs) on each string.
These DOMs are used to detect Cherenkov light from

neutrino-induced charged particles. The present search uses
observational data from IceCube during its construction
period, which can be characterized with the number of
strings deployed at the time of the observation. Deployment
was performed only during Austral summers, resulting in
extended ∼1 yr observation periods using a constant
number of deployed strings (see Fig. 1).
IceCube detects neutrinos through the Cherenkov pho-

tons from secondary charged particles. For the present
search, the secondary particles are muons, which, at> GeV
energies, travel long enough in ice before decaying for an
accurate direction reconstruction. We apply the neutrino
event selection of the standard IceCube point source
analysis [4,111–116]. Due to the large flux of muons
produced by cosmic ray interactions within the atmosphere,
the search is constrained to muons moving upward through
the detector. As the Earth is opaque to muons, atmospheric
muons are filtered out in these directions. The remaining
background is due to atmospheric neutrinos produced by
cosmic ray interactions in the atmosphere, and which
produce muons in the ice around or within the detector.
The direction of the secondary muons are reconstructed to a
precision of ≲1° in the TeV-PeV energy range [117]. At
these energies, the difference between the direction of the
incoming neutrino and the secondary muon is negligible.
Astrophysical neutrinos from individual sources can be

identified by a localized excess in space and/or time above
the atmospheric background. The energy spectrum of
atmospheric neutrinos can also be used to differentiate
between the background and astrophysical signals.
Neutrinos produced in the atmosphere have a soft energy
spectrum (e.g., [118]) with a power law spectral index
of ∼ − 3.7 above 100 GeV, compared to the harder
astrophysical spectrum, with a spectral index of ∼ − 2
due to the expected Fermi acceleration of protons in the
GRB outflow [119].
The present search uses high-energy-neutrino data from

the IceCube Neutrino Observatory in its 22, 59 and 79-
string configurations. Neutrino data was recorded with the
22-string configuration from May 31, 2007 to April 4,
2008. IceCube recorded data with its 59-string configura-
tion from May 20, 2009 to May 30, 2010, and with its 79-
string configuration from May 31, 2010 to May 12, 2011.
The observational periods of IceCube used in this analysis
are shown graphically in Fig. 1 in comparison to the
observation periods of LIGO and Virgo. The number of
detected astrophysical neutrino candidates for the three
observation periods, as well as their subsets that were
detected during times when H1, L1 and V1 GW detectors
were operational, are shown in Table I.
For each neutrino, this analysis uses its (i) time of arrival,

(ii) reconstructed direction, (iii) directional uncertainty and
(iv) reconstructed energy. The present search uses the
reconstructed neutrino energy as a test statistic to character-
ize the significance of the neutrino signal, while the

S5 S6

VSR1 VSR2

V
S

R
3

IC40IC22 IC59 IC79

2007
2008

2009
2010

2011

LIGO

Virgo

IceCube

GW+neutrino search

FIG. 1 (color online). Observational periods of the initial LIGO
and Virgo GW detectors, the partially completed IceCube
detector, as well as the joint GWþ neutrino search periods.
The IC40 observation period was not coincident with LIGO-
Virgo observation periods, and was therefore not included in this
search. See also Sec. II A and Table I for more information on the
GW and neutrino observation periods, respectively.

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-10



reconstructed direction and its uncertainty are also used to
determine the significance of the directional coincidence
between neutrinos, GWs and galaxies.
Note that the energy measured by IceCube is necessarily

a lower limit on the actual energy of the neutrino. If a
secondary muon is produced outside of the detector, an
unknown fraction of the neutrino energy is transferred to
the shower at the interaction vertex and the muon will lose
energy before reaching the instrumented volume. These
two effects are difficult to account for when reconstructing
the neutrino’s energy (c.f. [116,122]).

C. Galaxy distribution

The distribution of GW sources is expected to be highly
nonuniform in the volume within a few tens of Mpc probed
by initial LIGO and Virgo. We take advantage of this by
looking for astrophysical sources in the directions of
galaxies, and by assigning a weight to the direction of
each galaxy proportional to the probability of detecting an
astrophysical neutrino from that direction. Here, we assume
that common GWþ neutrino sources are located within or
near galaxies, and that the source population within/nearby
a galaxy is proportional to the galaxy’s blue luminosity (see
[103] and references therein). This estimate also assumes
uniform directional sensitivity (IceCube’s sensitivity
weakly depends on the zenith angle; e.g., [123]).
We take the list of galaxies with their locations and blue

luminosities from the Gravitational-Wave Galaxy Catalog
[124]. We assign each galaxy a weight that reflects the
probability of detecting an astrophysical neutrino from the
given galaxy.
Since we assume that the number of sources in a galaxy

is proportional to B, the probability of detecting a neutrino
from a source within this galaxy will also be proportional
to B. The neutrino flux observed on Earth from any source
within the galaxy source will scale with r−2. Assuming that
the typical number of expected neutrino events from such a
flux from a source is ≪ 1, the probability of detecting a
single neutrino from a given galaxy is therefore propor-
tional with B=r2. This value will be our weight for the
galaxy. Note that the factor would be B2=r4 for two

neutrinos from two independent events from the same
galaxy, but this factor would be very small and therefore we
ignore it in this analysis.
We can also consider the detection of more than one

neutrino from a given source. For this analysis we only
consider up to two neutrinos, since three or more detected
neutrinos from a single source would by itself be a
detection if the neutrinos were all detected within a
500 s time window (see Sec. II D 2). The probability of
detecting a neutrino doublet from a single source, again
with the assumption that the typical neutrino flux from a
source is ≪ 1, is proportional to B=r4.
For simplicity, the analysis divides the sky into

0.1° × 0.1° pixels and sums the contribution from each
galaxy to these pixels (see Fig. 1. in [124] and Fig. 5. in
[103]). We only take into account galaxies that are within
the [5 Mpc, 40 Mpc] range. With the lower cutoff we only
discard a small fraction of the potential events, while these
events may produce a strong enough signal that could be
detectable with GW or neutrino detectors independently.
The farther cutoff takes into account that even in optimistic
models the joint emission can only be detected out to
≲40 Mpc due to GW sensitivity.

D. Analysis algorithm

The joint analysis aims to identify GW events and
neutrinos that originate from a common astrophysical
source [103]. Below we refer to a temporally coincident
GW event and one or more neutrinos as a GWþ neutrino
event. In this section we describe how the significance of a
GWþ neutrino event is calculated.
The joint analysis determines the significance of a

GWþ neutrino event by combining the significances
and directional distributions of GW events, neutrinos and
galaxies. The following measures are used:
(1) GW signal test statistic: ρ;
(2) neutrino energy: Eν;
(3) probability of having more than 1 background

neutrino temporally coincident with the GW (if
applicable);

TABLE I. Number of neutrinos for the three IceCube observation periods considered in the present analysis. First
column (No. of strings): number of strings in the different IceCube configurations, corresponding to the three
observation periods. Second column (Start): start date of observation period. Third column (End): end date of
observation period. Fourth column (No. of neutrinos): number of neutrinos from the northern hemisphere for the
different observation periods. Fifth column (No. coincident with GW data): number of neutrinos that were detected
during the timewhen H1, L1 and V1 were operational (for the first row, the operation of H2 was also required). Sixth
column (Ref.): reference describing neutrino data selection.

No. of strings Start End No. of ν No. coincident with GW data Ref.

22 May 2007 Apr 2008 4852 978 [120]
59 May 2009 May 2010 43339 3363 [121]
79 May 2010 May 2011 50857 3892 [121]

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-11



(4) point spread functions of GW, neutrinos and
galaxies;

The p-values of these four measures are calculated sepa-
rately, and then are combined into a joint p-value. Figure 2
illustrates how they are combined, while a detailed explan-
ation is presented below.

1. Neutrino significance

We start with neutrinos detected by IceCube. We use
the energy of a neutrino to calculate its significance; the
neutrino’s direction and the number of coincident neutrinos
will be used separately (see Secs. II D 4 and II D 3,
respectively). For neutrino i with reconstructed energy

EðiÞ
ν , the p-value pðiÞ

ν will be the probability of a back-

ground neutrino having energy EBG
ν ≥ EðiÞ

ν :

pðiÞ
ν ¼ PðEBG

ν ≥ EðiÞ
ν Þ: ð1Þ

We conservatively assume here that the fraction of detected
astrophysical neutrinos is negligible within all detected
neutrinos, and use the energy estimators of all detected
neutrinos for obtaining the p-values.

2. Gravitational-wave significance

Next, we calculate the significance of GWevent j that is
temporally coincident with neutrino i above. The test
statistic ρj of GW event j is obtained using the cWB

pipeline. We first calculate the false alarm rate FARðjÞ
gw of

the GWevent by determining the rate at which background
GWevents occur with test-statistic values ρbg ≥ ρi. We use
this FARgw to calculate the p-value of GWevent j, defined
as the probability of finding a background GW event with
ρ ≥ ρj that is temporally coincident with neutrino i:

pðiÞ
gw ¼ 1 − Poisð0; 2T · FARðjÞ

gwÞ; ð2Þ

where Poisðk; λÞ is the Poisson probability of k outcomes
with λ average, and T is the GWþ neutrino coincidence
time window.
We adopt an emission time window of T ¼ 500 s

following the recommendation of [125]. We assume that
a source emits both GWs and neutrinos over this 500 s
period, and consequently the maximum allowed time
difference between a GW event and a neutrino is 500 s.
For a detected neutrino, this corresponds to a total time
window of 2T ¼ 1000 s (�500 s) in which a GW event
needs to be in order for the two events to be considered
coincident. This T time window is established based on
the expected emission duration of long GRBs, being a
conservative upper limit on the duration of long-GRB
central engines’ activity. It takes into account the observed
duration of prompt gamma-ray emission, the duration of a
relativistic jet burrowing through the stellar envelope, as
well as potential precursor activity prior to prompt emis-
sion. The time window is likely also sufficient for other
multimessenger transients, such as short GRBs or magnetar
giant flares.

3. Significance of neutrino cluster

We further take into account that multiple neutrinos can
be detected in coincidence with a GW event. Note that we
do not search for multiple GW events from one astrophysi-
cal source. Two neutrinos are considered to have come
from the same source only if they are temporally and
directionally coincident. Neutrinos are considered tempo-
rally coincident if they arrive within T from each other,
following the same motivation as for the time window for
GW events and neutrinos [125]. We define directional
coincidence as the existence of a common direction from
which both neutrinos could originate from with probabil-
ity p > 0.05.
First, for a cluster of neutrinos, we calculate the

significance of the neutrinos’ test statistic by taking the
product of the p-value of each neutrino within the cluster,
to arrive at a significance

pν ¼
Y
fkg

pðkÞ
ν ; ð3Þ

where the product is over all elements in the cluster.
Second, we assign a p-value to the coincidence of the

cluster of N neutrinos, defined as the probability that a total
of ≥ N neutrinos occur within a T interval. For background
neutrinos detected with rate fν over the northern hemi-
sphere and uniformly distributed in time, this probability is

pclusterðNÞ ¼ 1 −
XN−2

k¼0

Poisðk; fν · TÞ: ð4Þ

Note that the sum goes to N − 2 since we require at least 2
neutrinos for a cluster, i.e. pcluster is a conditional

FIG. 2 (color online). Flow diagram of the joint GWþ neutrino
analysis algorithm, showing how information on neutrinos,
galaxies and GWs are combined into one test statistic. PSFs
denote the point spread functions of GWs and neutrinos, and the
weighted directional distribution of galaxies. See Sec. II D for
definitions of the parameters and for a detailed description of the
algorithm.

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-12



probability, i.e. it is the probability of the detection
of additional neutrinos given that the first neutrino has
been detected. For only one neutrino, we have
pclusterðN ¼ 1Þ ¼ 1.

4. Significance of directional coincidence

We assign an additional significance to the directional
coincidence of GWs, neutrinos and galaxies. We define the
test statistic for directional coincidence to be the likelihood
ratio of the GWevent and neutrino coming from a common
galaxy. We use this test statistic to calculate the p-value
psky of directional coincidence. See Appendix A for the
derivation of psky (see also [103]).

5. Combined significance

After obtaining the p-values for GWevents, neutrinos, as
well as their directional coincidence, we combine these p-
values using Fisher’s method into one joint test statistic:

X2
i ¼ −2 ln ½pðiÞ

skyp
ðiÞ
gwpclusterðNÞpν�: ð5Þ

The background distribution of this combined test statistic
PbgðX2Þ is used to calculate the p-value for a given X2

i :

pðiÞ
gwν ¼

Z
∞

X2
i

PBGðX2ÞdX2: ð6Þ

It is often convenient to evaluate the significance of a
GWþ neutrino event using a joint false alarm rate (FAR)
threshold. One can define a FAR corresponding to GWþ
neutrino event i as FARi ¼ pðiÞ

gwν=tobs, where tobs is the
duration of the joint observation period. This FAR means
that the rate of background events identified by the joint
analysis with X2 ≥ X2

i will be FARi.

E. Background distribution

The background GW-event level is generated by intro-
ducing time shifts between data from GW detectors. For a
given realization, the time for each GW detector data is
shifted compared to other detectors by a duration that is
much greater than the travel time of a GW between the
detectors, as well as the allowed GW transient duration.
This ensures that, in the time-shifted data, no GW signal is
coincident in the multiple detectors. For the analysis a total
of 200 time shifts are produced, and cWB is used to search
for GW events in these time-shifted data.
For a subset of background GW events identified by

cWB, we assign neutrinos randomly selected from the
list of neutrinos detected by IceCube. In practice this is
implemented by changing the time of arrival of each
neutrino to match the time of the corresponding GW
background event. We keep the other parameters of the
neutrino unchanged: the energy estimator and its

uncertainty, as well as direction in local detector coordi-
nates and its uncertainty.
We search for GWþ neutrino events in the set of

generated events using the joint analysis to obtain the
background distribution of X2. Since the events are
generated to have temporally coincident GW event and
neutrino, we further need to account for the fact that a large
fraction of the neutrinos is expected to have no temporally
coincident GW event in the real data. This is accounted for
using the GW event and neutrino rates and the observation
time, and assuming uniform temporal distribution for both
GW events and neutrinos.

F. Simulated signals

We use simulated GW and neutrino signals to examine
the sensitivity of the joint search to astrophysical events.
For this analysis we adopt two different GW signal types.
Both simulated GW signals have linearly polarized sine-
Gaussian waveforms. This is a conservative choice as GW
detectors are more sensitive to circularly polarized GWs
(e.g., [126]). We note that the observed GWs are circularly
polarized for rotating sources viewed along their rotation
axis, which is the case, e.g., for GRBs with highly
collimated jets.
For linearly polarized GWs with polarizations hþ and

h× of the GW strain, we can choose the coordinate system
such that the signal’s polarization is in the þ mode. Our
simulated, linearly polarized GW signal can then be written
as

hþðtÞ ¼ h0 sinð2πf0tÞ exp½−ð2πf0tÞ2=2Q2�; ð7Þ

h×ðtÞ ¼ 0; ð8Þ

where h0 is the maximum amplitude, f0 is the characteristic
frequency, and Q=f0 is the characteristic duration of the
GW signal. For simulated astrophysical GWs, two standard
signal types with f0 ¼ 153 Hz and Q ¼ 8.9, and f0 ¼
1053 Hz and Q ¼ 9. The first waveform has a character-
istic frequency that is in the most sensitive band of the
observatories, thus probing the reach of the measurement.
For broadband emission such as compact binary mergers,
this sensitive band will represent the most important
contribution to detectability. The second waveform is at
a higher frequency that may be more typical for some core-
collapse models and SGR flares. The value of Q only
weakly affects the search sensitivity, therefore the specific
choice is not critical [108].
The two are standard gravitational waveforms that allow

the results to be compared to other GW analyses (see, e.g.,
[127]). The sensitivity of GW-only searches with respect to
other waveforms can be seen, e.g., in [126]. Simulated
GWs were added to the recorded data with distinct strain
amplitudes. The GW signal’s root-sum-squared amplitude
hrss is defined as

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-13



h2rss ¼
Z

ðjhþðtÞj2 þ jh×ðtÞj2Þdt: ð9Þ

GWs are added with amplitudes in the range hrss ∈
½10−22; 2 × 10−19� Hz−1=2. For the standard-siren GW with
Egw ¼ 10−2 M⊙c2, this range corresponds to a source
distance of D ∈ ½50 Mpc; 50 kpc�. For each strain ampli-
tude, we generate between 360 − 1114 joint simulated
events.
To generate a simulated GW signal, we first select a

random direction on the sky from an isotropic distribution
of sources with declination δ ≥ 0°, which defines the
directional range in which IceCube is the most sensitive
to TeV-PeV neutrinos. This direction determines the time-
of-arrival difference between the detectors, as well as the
antenna factors that define the projection of the strain
amplitude into the data stream. Then a simulated GW signal
with random polarization is added to the GW detector data.
Astrophysical high-energy neutrinos are simulated with

a Monte Carlo simulation of the partially completed
IceCube detector (see [113] for the IC22 simulation, and
[128] for IC59 and IC79). The simulation considers a
neutrino arriving from the direction identical to the direc-
tion chosen above for the GW signal injection. The time of
arrival of the neutrino and GW signal are chosen to be the
same. Since we do not use any weight for the time
difference of the GW event and the neutrino, choosing
the same time or arrival is equivalent to selecting a random
time for the GWevent and the neutrino within time window
T. The simulation performs the production of secondary
muons from neutrino interactions in or in the vicinity of the
detector, tracks them and their Cherenkov-light emission
through the detector and generates the signals in the DOMs.
Afterwards, the energy, direction and directional uncer-
tainty of the neutrinos are reconstructed as with real data.
The energies of the simulated neutrinos are drawn from a
distribution dNν=dEν ∝ E−2

ν with an energy lower limit of
300 GeV, which takes into account the expected E−2

ν

scaling of Fermi acceleration, which is the acceleration
mechanism typically considered for relativistic, shocked
outflows (e.g., [129]).
Each injected event is randomly assigned a host galaxy

using the probability weights described above. The
assigned galaxy is then treated as if its direction was
identical to the direction of origin of the simulated GWand
neutrino, i.e. it is “placed” in the direction of the simulated
signal. This treatment of the galaxy distribution increases
the effective number of simulated astrophysical signals for
a limited number of simulated GW signals. The obtained
result is conservative: any nonuniformity in the galaxy
distribution increases the advantage of using galaxies,
and since the simulated source direction is uniformly
distributed over the sky, it will, on average, decrease the
significance of simulated signals.

III. RESULTS

Our analysis has found no joint GWþ neutrino event
that was sufficiently significant to claim detection. We
found 13, 56 and 69 temporally coincident GW events and
neutrinos for the time periods spanning the S5/VSR1/IC22,
S6/VSR2/IC59 and S6/VSR3/IC79 observation periods,
respectively, which are consistent with the null hypothesis.
We found no neutrino doublets that were coincident with
a GW event, therefore in the following we focus on
GWþ neutrino events with a single neutrino.
The most significant GWþ neutrino event identified by

the analysis had X2 ¼ 9.21. It was detected on July 27,
2007 at 21:16:54 UTC, during the S5/VSR1/IC22 obser-
vation period. The reconstructed direction of the neutrino
was right ascension 159.6° and declination 20.8°. It hit 22
DOMs in IC22, making it a reconstructed event with an
energy measure typical to atmospheric events. The GW
event reached ρ ¼ 4.11. The p-value of the event, calcu-
lated for the full S5/VSR1/IC22 observation period, is
p ¼ 0.35. The event was therefore not sufficiently signifi-
cant to claim detection, and our results are consistent with
the background-only null hypothesis. For comparison, we
calculate here the X2 value that would be necessary, on
average for a GWþ neutrino signal to reach a joint false
alarm rate (FAR) of 10−2 yr−1 within the S5/VSR1/IC22
observation period, which approximately corresponds to 3σ
significance level. To obtain this X2, we perform a
Monte Carlo simulation as follows. For one realization,
we select N100 ≡ Nobs · ð100 yr=tobsÞ simulated back-
ground GWþ neutrino events [see Sec. II E and Fig. 3
(left)], where Nobs is the number of temporally coincident
GWþ neutrino events found during the S5/VSR1/IC22
observation period. For each realization, we select the
simulated joint GWþ neutrino background event with the
maximum X2 value. To obtain the expected X2, we take
the median maximum X2 value over the realizations. We
find that an event in the S5/VSR1/IC22 observation period
corresponding to a FAR of 10−2 yr−1 would need
X2 ¼ 14.3. This FAR value is equivalent to a p-value
of p ¼ FARtobs ¼ 1.7 × 10−3.
A further examination of data around our most signifi-

cant event revealed that the cause of the relatively large
signal-to-noise ratio of this event was due to large signals
present in the H1 and H2 detectors strain channels that were
caused by an electromagnetic disturbance; this disturbance
is clearly visible in magnetometer and voltage line monitor
channel signals.
To help the reader interpret our results, Table II includes

the variables used describing source properties and the
outcome of the search.

A. Efficiency and sensitivity

We calculated the background distribution of the
test statistic, X2, using 105 simulated background

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-14



GWþ neutrino events separately for each observation
period, as well as the dependence of X2 on the injected
GWamplitude of a simulated astrophysical event separately
for each observation period. Figure 3 (left) shows the
complementary cumulative distribution function (≡1-
cumulative distribution function; 1-CDF) of X2 values
obtained for the background, as well as for simulated joint
astrophysical events for the S5/VSR1/IC22 observational
period. For a given X2

i value, 1-CDF indicates the fraction
of events with X2 values for which X2 ≥ X2

i . Figure 3
(right) shows a similar evaluation for the GW-only case,
which uses the GW test statistic ρ (for definition see [109]).
Comparing the two distributions, one can see that the
joint analysis is more efficient at differentiating simulated

astrophysical signals from the background. Results are
qualitatively the same for the other two observation
periods. While the sensitivity of the GW detectors is
somewhat better for later runs, the general characteristics
of the data are similar.
We also evaluated the sensitivity of the joint search in

detecting a simulated astrophysical GWþ neutrino signal
for different FAR thresholds. We determined the search
sensitivity as a function of the GW root-sum-squared
amplitude (hrss), and considered “conditional sensitivity,”
given one neutrino has been detected from the source. The
probability of detecting a neutrino from a given source is
separately taken into account in, e.g., the upper limit
estimates below. For a given FAR threshold and for a given
hrss, we define sensitivity as the probability of a joint
GWþ neutrino event with 1 detected neutrino and with
hrss GW amplitude to be recovered by the search pipeline
with a FAR below the FAR threshold. We used different
thresholds: (i) the false alarm rate FARmax;obs, that corre-
sponds to the highest test-statistic value X2

max;obs, measured
during the observation period; (ii) FAR ¼ 0.1 yr−1 and
(iii) FAR ¼ 0.01 yr−1 (see Fig. 4). We fit sigmoid curves
onto the measured data to determine their trend. Not that
the curves saturate somewhat below 1, which is due to the
nonuniform directional sensitivity of the GW detector net-
work, as well as uncertainties in the reconstructed direction.
Results for the S5/VSR1/IC22 observation period are

shown in Fig. 4. We obtained similar results for the other
two observation periods.

FIG. 3 (color online). Complementary cumulative distribution function (1-CDF) of the test statistic of joint GWþ neutrino (left) and
GW-only (right) events for the background and for simulated astrophysical events, for the S5/VSR1/IC22 observation period. Results for
simulated astrophysical events are shown for different characteristic distances, as indicated in the legend. The function 1-CDF grows
with decreasing source distance. We used standard-siren GW injections (Egw ¼ 10−2 M⊙c2; sine-Gaussian waveform with character-
istic frequency f0 ¼ ∼153 Hz andQ ¼ 8.9). For the GWþ neutrino case, the results show events for which one neutrino was detected.
We also express the characteristic distances with intervals of hrss values, corresponding to the amplitude of a standard-siren GW signal at
the characteristic distance. These hrss intervals are shown in the legend in parentheses, in units of 10−22 Hz−1=2. A fit to the background
1-CDF is also shown (see Sec. III C). The distributions for the S6/VSR2/IC59 and S6/VSR3/IC79 observation periods (not shown here)
are qualitatively the same. The results show that the joint GWþ neutrino search identifies simulated astrophysical signals with higher
efficiency than the GW-only case search, and that a large fraction of the simulated signals can be clearly differentiated from the
background.

TABLE II. List of variables and parameters used in the joint
analysis, and their role.

Variable Definition

hrss Observed GW root-sum-squared amplitude
ρ GW test statistic
Eiso
gw Total emitted isotropic-equivalent GW energy

Eiso
ν Total emitted isotropic-equivalent ν energy

nðic86Þν Average No. of detected ν by IC86 from 10 Mpca

X2 Test statistic of GWþ neutrino event
Rul Joint source rate upper limit

aDetection with the full IceCube detector for a source in the
northern hemisphere at 10 Mpc distance.

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-15



B. Source rate upper limit

We use the sensitivity of the joint search, as well as of
nondetection, to obtain astrophysical source rate upper
limits. Here we build on and improve the analysis by
Ref. [86], that derived source rate upper limits using an
effective detection volume defined by the distance at which
sources can be detected with 50% efficiency. We note that a
previous GWþ neutrino coincident search with data from
initial LIGO-Virgo and the partially completed ANTARES

derived upper limits on the GWþ neutrino source pop-
ulation using a different method [87]. Ref. [87] separately
derives horizon distances for neutrino and GW signals
(with 50% and 90% detection probabilities, respectively),
and uses the smaller horizon distance of these two to define
the effective search volume, which in turn determines
the source rate upper limit. The upper limits derived in
the LIGO-Virgo- ANTARES analysis remained above the
currently accepted source rates for long or short GRBs, as
do the results presented here. Limits from LIGO-Virgo-
ANTARES, in principle, apply to a source population on the
opposite (southern) hemisphere as the limits from the
LIGO-Virgo-IceCube search (at ∼TeV neutrino energies).
We derive source rate upper limits using the sensitivity

of the joint search as a function of hrss, determined for
FARmax;obs (see Sec. III A). We calculate upper limits
separately for the three observation periods, which are
then combined into one overall upper limit.

We can express the sensitivity of the search, which is
characterized above as a function of hrss, as a function of
the source properties. The observed GW strain amplitude
hrss depends on the total isotropic-equivalent GW energy
Eiso
gw emitted by the source, as well as the source luminosity

distance D:

hrss ≈
kG1=2

πc3=2
f−10 ðEiso

gwÞ1=2D−1; ð10Þ

where k is a constant that depends on the polarization of
the emission, and G is the gravitational constant. It is
k ¼ ð5=2Þ1=2 for circularly polarized emission that is
expected from a rotating source whose rotation axis points
towards the Earth, and k ¼ ð15=8Þ1=2 for linearly polarized
signals (e.g., [130]). We conservatively adopt this latter
value in the present analysis.
Let εdetðEiso

gw; DÞ denote the probability that the joint
search detects an astrophysical GWþ neutrino event with
test statistic X2 > X2

max;obs, given that one neutrino is
detected from the source. This probability is shown as a
function of hrssðEiso

gw; DÞ in Fig. 4.
We now estimate the probability that a source will

produce a GWþ neutrino event with test statistic
X2 > X2

max;obs. Let N ¼ nνðD=10 MpcÞ−2 be the expected
number of neutrinos detected by IceCube from a source at
luminosity distanceD in the northern hemisphere, where nν
is the expected number of detected neutrinos for a source
located at 10 Mpc. Note that nν depends on both the
detector and the source. The probability that at least one
neutrino will be detected from this source is

pð≥ 1jD; nνÞ ¼ 1 − Poisð0; NÞ: ð11Þ
We further take into account the source’s neutrino beaming
factor fb, which is expected to be similar to the gamma
beaming factor. The beaming factor is defined such that f−1b
is the fraction of the full sky towards which neutrinos are
emitted. We take fb ¼ 14, which is a typical upper limit on
the beaming factor of low-luminosity GRBs [131]. The
probability that a randomly oriented source at distance D
will produce a GWþ neutrino event that is detected with
X2 > X2

max;obs can be expressed as

PdetðD;Eiso
gw; nνÞ ¼ f−1b pð≥ 1jD; nνÞεdetðEiso

gw; DÞ: ð12Þ

We point out that Pdet is determined by the source distance
and the source’s GW (Eiso

gw) and neutrino (nν) emission
parameters. Note further that Pdet is a probability for a joint
event, while pð≥ 1jD; nνÞ is a probability for the neutrino-
only case.
We place a limit on the rate of joint GWþ neutrino

events at 90% confidence level. For this confidence
level, we need to determine the source rate upper limit
that corresponds to an average of ≥ 2.3 events over
the measurement duration tobs that would produce a

10
−22

10
−21

10
−20

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

h
rss

 [Hz−1/2]

1 
−

 F
al

se
 D

is
m

is
sa

l P
ro

ba
bi

lit
y

4.3e−22

GW+ν (observation)

GW+ν (10−1 yr−1)

GW+ν (10−2 yr−1)

GW     (10−1 yr−1)

GW     (10−2 yr−1)

46 4.6 0.46
Distance [Mpc]

FIG. 4 (color online). Sensitivity of the joint analysis and the
GW search as a function of GW hrss, for different FARs. The FAR
values are shown in the legend. The threshold “Observation”
corresponds to the FAR of the most significant observed GWþ
neutrino event. For the simulated GW events, the standard-siren
event was used (sine-Gaussian waveform; f0 ¼ 153 Hz;
Q ¼ 8.9). For the joint analysis, the efficiency is calculated given
the detection of one neutrino from the source. For comparison, the
top axis shows the source distances corresponding to the different
hrss for the standard-siren event. For the “Observation” curve, we
show h50%rss ¼ 4.3 × 10−22 Hz−1=2 (full circle).

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-16



GWþ neutrino with X2 > X2
max;obs. Assuming a uniformly

distributed source population, the rate density R can be
constrained below an upper limit Rul:

RulðEiso
gw; nνÞ ¼

2.3
tobs

R
∞
0 PdetðD;Eiso

gw; nνÞ4πD2dD
: ð13Þ

After calculating this source rate upper limit for the three
observation periods, we combine these results to obtain one
source rate upper limit for the whole search. Upper limits
can be integrated by summing their inverse, such that

1

Rcombined
ul

¼
X3
i¼1

1

RðiÞ
ul

; ð14Þ

where the sum is over the three observation periods. The
resulting combined upper limit, Rcombined

ul , is shown in Fig. 5
as a function of source emission parameters. For GW
emission, the source emission strength is expressed with
Eiso
gw. Characteristic GW energies range from Eiso

gw ∼
10−2 M⊙c2 for compact binary inspirals [71] to Eiso

gw ≲
10−4 M⊙c2 for core-collapse supernovae [67] (c.f., the
horizontal axis of the figure). For neutrino emission, the
source emission strength is expressed as (i) isotropic-
equivalent neutrino energy Eiso

ν including all neutrino
flavors, and (ii) as the average number nðic86Þν of neutrinos
that would be detected from the source at 10 Mpc distance,
scaled to the full IceCube detector (with 86 strings). These

two neutrino measures are connected as Eiso
ν ¼ κnðic86Þν ,

with κ ≈ 0.7 × 1049 erg for an E−2 neutrino spectrum [86].
To convert the sensitivity of the different partially com-
pleted IceCube runs to that of the full IceCube. We assume
that sensitivity scales with the number of strings, which is a
good approximation over the whole range of declinations
searched; the relevant effective areas are presented in [120]
and [121]. Predicted characteristic neutrino emission ener-
gies include Eiso

ν ∼ 1051 erg [41], and Eiso
ν ∼ 1050 erg

[42,82] (see the secondary y axis of Fig. 5).
To put these results in context, we compare the obtained

rate upper limits to those of potentially related astrophysi-
cal sources. One of the potential sources of interest is
core-collapse supernovae with rapidly rotating cores, which
may drive relativistic outflows that result in high-energy-
neutrino emission, and can also emit detectable GWs
[9,132–138]. The local (z ¼ 0) rate of core-collapse super-
novae is ∼2 × 10−4 Mpc−3 yr−1 [139], significantly below
the upper limits that were obtained with the present search.
While the present results do not constrain current

astrophysical models, we establish a baseline for future
measurements by advanced GW detectors. The nominal
sensitivity increase of advanced over initial GW detectors is
a factor of 10, which corresponds to a factor 1000 increase
in sensitive volume. For the cases in which the GW
component limits the joint sensitivity, this translates into
a factor of 1000 improvement in the expected source rate
upper limit upon no detection. For emission cases in which
the neutrino component is limiting, this improvement will
be smaller. For comparison, taking the standard-siren GW
emission with Eiso

gw ¼ 10−2 M⊙c2, and a neutrino emission
comparable to typical GRB gamma-ray emission, Eiso

ν ¼
1051 erg [41], the upper limit obtained by the present search
is 1.6 × 10−2 Mpc−3 yr−1. In contrast, we project a source
rate upper limit of 4 × 10−4 Mpc−3 yr−1 for a 1-year
observation period with advanced LIGO/Virgo and the
full IceCube, which is comparable to the core-collapse
supernova rate.

C. Detection sensitivity improvement
with the joint search

An advantage of combining data from searches for GWs
and neutrinos is the reduced total FAR. Requiring temporal
and directional coincidence from a joint event are effective
filters in reducing the background, beyond the identifica-
tion capabilities of single-messenger searches. Low FAR
searches are critical in identifying an astrophysical signal
with high significance. This will be particularly important
for the first GW discoveries. Below, we compare search
sensitivities requiring low FAR for the joint GWþ neutrino
search and the GW-only and neutrino-only searches to
characterize the improvement we can expect from the joint
search.

5−2

−2

−
1.5

−1.5

−
1

E
gw
iso   [M

sun
c2]

n ν(ic
86

)    
[#

 in
 IC

86
 a

t 1
0M

pc
]

10
−4

10
−3

10
−2

10
−1

10
0

10
1

10
2

log(R
ul
combined / Mpc−3yr−1)

−3 −2.5 −2 −1.5 −1 −0.5 0

10
49

10
50

10
51

E
νis

o

FIG. 5 (color online). Joint GWþ neutrino source rate upper
limit of the present search as a function of isotropic-equivalent
GW emission Eiso

gw and neutrino emission. Neutrino emission is
given both in terms of average number of neutrinos detected by
IceCube with 86 strings from 10 Mpc (nðic86Þν ), and in terms of
emitted isotropic-equivalent neutrino energy (Eiso

ν ; in all flavors).
The results here assume an E−2 neutrino energy spectrum (see
Sec. II B). The results shown here combine measurements from
the three analyzed observation periods (S5/VSR1/IC22, S6/
VSR2/IC59 and S6/VSR3/IC79).

MULTIMESSENGER SEARCH FOR SOURCES OF … PHYSICAL REVIEW D 90, 102002 (2014)

102002-17



To estimate the sensitivity of GW searches for low FAR,
we take the S5/VSR1/IC22 observation period, which
represents the longest duration of joint operation. To be
able to estimate GW-only search sensitivities with low
FAR, we extrapolate the distribution of the GW test statistic
to increase the effective number of test-statistic samples.
We fit the complementary cumulative distribution function
of the GW test statistic ρ (see Sec. II A) with the empirical
function a1ðρ − ρð1Þ0 Þγ1 þ a2ðρ − ρð2Þ0 Þγ2 . The distribution of
ρ, as well as the fit, is shown in Fig. 3. With this empirical
distribution we can estimate the average largest ρ for a
given measurement duration. This in turn can be used to
calculate the expected ρ threshold corresponding to a
given FAR.
The efficiency of the joint search and a GW-only search

in detecting a signal below a given FAR as a function of
GW root-sum-squared strain amplitude hrss is shown in
Fig. 4 for different FARs. The detection efficiency is given
as a conditional probability, given 1 detected neutrino. The
figure shows the difference between the efficiencies of the
GW-only and the GWþ neutrino searches. One can see
that the sensitivity of the GWþ neutrino search is much
less dependent on the applied FAR threshold than the
sensitivity of the GW-only search. The joint search is
therefore increasingly advantageous for smaller FARs.
To quantitatively compare the sensitivity of the GW-only

and GWþ neutrino searches at low FAR, we consider the
hrss value for which the searches reach 50% efficiency. We
denote the corresponding hrss value with h50%rss . This is a
meaningful comparison since h50%rss is characteristic of the
distance to which a GW search is sensitive to. The obtained
efficiencies for different FAR levels are shown in Table III.
One can see that, for FAR ¼ 0.01 yr−1, the difference
between the h50%rss values of the GW-only and GWþ
neutrino searches is about a factor of 3. Since the probed
search volume scales with ðh50%rss Þ3, the volume covered by
the joint search increase by a factor of ∼30. Further, for a
5σ discovery, the difference between the GW-only and
GWþ neutrino searches is likely to grow even further. The
joint search can therefore significantly expand the set of
potentially detectable sources. The number of detected joint
sources, nevertheless, will be affected by the probability of

detecting a neutrino from a distant source as well as the
beaming angle of the source.
The above results also highlight the importance of GW

detector characterization efforts. We see from Fig. 4 that
decreasing the FAR reduces the efficiency of the GW-only
search much more significantly than that of the joint search,
due to the non-Gaussian tail of the background GW-event
distribution. For a 5σ discovery over a 1-year observation
period, assuming that the fit curve is valid for higher test-
statistic values, the required GW test statistic would need
to be ρ ≈ 300. This value is outside of the scale of the
injections we performed, and shows that the non-Gaussian
tail of the GW background distribution makes it difficult to
raise the significance of a GWevent. This result exemplifies
that addressing the heavy-tail background distribution in
GW detectors, e.g., in the context of detector characteri-
zation, will be critical for GW-only searches.
To compare the GWþ neutrino search to the neutrino-

only case, we emphasize that the joint search is in principle
capable of detecting joint events with a single detected
neutrino. A neutrino-only search, in the same time, would
require at least 3 directionally and temporally coincident
neutrinos for detection, without any electromagnetic or
other counterpart (e.g., [86]). To compare the GWþ
neutrino and neutrino-only cases, we consider the differ-
ence between the characteristic distances from which 1 or 3
neutrinos are detected from a given source. For simplicity,
we consider the case in which the respective detection
probability is 50%. Let Dν;1 be the distance at which the
probability of detecting at least one neutrino from a given
source is 50%. This corresponds to an average of ∼0.7
detected neutrinos. For the same source, detecting 3 or
more neutrinos with 50% probability corresponds to an
average of ∼2.7 detected neutrinos. The ratio between the
characteristic distances of the neutrino-only search and the
GWþ neutrino search (given a GW trigger) is therefore

Dν;3=Dν;1 ≈ 2; ð15Þ

corresponding to a search volume increase of a factor of
∼10. Note that beaming will be the same for the neutrino-
only and the GWþ neutrino search, it will therefore not
affect this difference. For the GWþ neutrino case, a
limitation will come from the finite distance from which
GWs can be detected (up to hundreds of megaparsecs for
advanced GW detectors, depending on the source’s GW
emission).

D. PeV neutrinos

Following IceCube’s recent detection of > 100 TeV
astrophysical neutrinos [28,29], it is beneficial to consider
the extension of the present GWþ neutrino search in this
direction. While some of the > 100 TeV neutrinos have
poor localization as they show a cascade-like hit pattern in
the detector, the present analysis pipeline can incorporate

TABLE III. Distances and hrss thresholds at which 50% of
astrophysical sources with standard-siren GW emission are
detectable for different FAR values, for both the GW-only and
the GWþ neutrino searches. The GWþ neutrino case assumes
that 1 neutrino has been detected from the source.

Search FAR [yr−1] h50%rss [Hz−1=2] Distance [Mpc]

GW-only 0.1 9 × 10−22 5
GW-only 0.01 1.7 × 10−21 3
GWþ neutrino 0.1 5 × 10−22 9
GWþ neutrino 0.01 6 × 10−22 8

M. G. AARTSEN et al. PHYSICAL REVIEW D 90, 102002 (2014)

102002-18



this limited direction reconstruction, as well as the temporal
coincidence, to find joint GWþ neutrino events.
We compared the time of arrival of the > 100 TeV

neutrinos detected by IceCube to the time periods when
LIGO-Virgo were operational. While some of the neutrinos
were detected during the nominal LIGO-Virgo observation
periods, in the present search we considered only times for
which at least three detectors were operational. None of the
> 100 TeV neutrinos had such triple coincidence. While
this means that we presently did not determine meaningful
limits on joint sources of GWs and > 100 TeV neutrinos,
these searches can present an interesting direction for the
advanced GW detector era, even in its early stages with
limited GW direction reconstruction capabilities.

IV. DISCUSSION AND OUTLOOK

We searched for common transient sources of GWs and
high-energy neutrinos using the initial LIGO and Virgo
GW observatories (S5-6/VSR1-3 observation periods)
along with the partially completed IceCube neutrino
detector (22, 59, and 79 strings). We found no sufficiently
significant joint event to claim detection. We used the
results to derive a limit on the rate of common sources for a
range of GW and neutrino emission parameters. Taking
the standard-siren GW emission with Eiso

gw ¼ 10−2 M⊙c2
around 150 Hz and a neutrino emission energy Eiso

ν ¼
1051 erg comparable to typical GRB gamma-ray emission,
we constrain the source rate below 1.6 × 10−2 Mpc−3 yr−1.
While these results do not constrain current astrophysical
models, we establish a baseline for future measurements by
advanced GW detectors. We project a source rate upper
limit of 4 × 10−4 Mpc−3 yr−1 for a 1-year observation
period with advanced LIGO/Virgo and the full IceCube,
which is comparable to the core-collapse supernova rate.
We investigated the improvement of a GWþ neutrino

search over single-messenger searches with respect to
(i) the obtained source rate upper limit and (ii) search
sensitivity. We characterized the latter as the prospects of
identifying an astrophysical signal with low FAR, which is
necessary for discovery. For the first time we carried out
a comparison with low FAR (1=100 yr). We found that
the GWþ neutrino search performs significantly better
than a GW-only search. Compared to the extrapolated
GW background distribution, the horizon distance of the
GWþ neutrino search increased by a factor of ∼3. This
difference further increases with decreasing FAR. This
demonstrates that the added information from multiple
astronomical messengers will be especially important for
high-significance detections. The difference for these
detections is more pronounced than for source rate upper
limits that allow higher FAR. Our results further indicate
that the heavy-tail distribution of the GW-event test statistic
limits the prospects of high-significance single-messenger
detections, making GW detector characterization a critical
step in increasing sensitivity for advanced GW detectors

(e.g., [140,141]). For characterizing the different sensitiv-
ities of GWþ neutrino and neutrino-only searches, we find
that the possibility of detection even with 1 observed
neutrino for the joint analysis increases the search volume
by a factor of ∼10 compared to neutrino-only searches that
require at least 3 coincident neutrinos for detection. This
increased volume, nevertheless, may be limited by the
range of GW detection.
The search algorithm developed for this analysis is

readily usable with data from any number of advanced
GW detectors and with neutrino observatories such as
IceCube or KM3NeT. Observations may be especially
interesting with early advanced GW detectors with only
two operational observatories. The search can also be used
to search for the highest-energy (∼PeV) neutrinos coinci-
dent with GW signals, even with limited directional
information.
With IceCube now completed and with the construction

of advanced LIGO-Virgo finishing within a year, the next
generation of multimessenger searches can commence
soon, expanding our reach towards the high-energy, tran-
sient phenomena.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the support of the
United States National Science Foundation for the con-
struction and operation of the LIGO Laboratory, the
Science and Technology Facilities Council of the United
Kingdom, the Max-Planck-Society, and the State of
Niedersachsen/Germany for support of the construction
and operation of the GEO600 detector, and the Italian
Istituto Nazionale di Fisica Nucleare and the French Centre
National d