ebook img

Tracing interstellar magnetic field using velocity gradient technique: Application to Atomic Hydrogen data PDF

0.91 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Tracing interstellar magnetic field using velocity gradient technique: Application to Atomic Hydrogen data

DRAFTVERSIONJANUARY30,2017 PreprinttypesetusingLATEXstyleAASTeX6v.1.0 TRACINGINTERSTELLARMAGNETICFIELDUSINGVELOCITYGRADIENTTECHNIQUE: APPLICATIONTOATOMICHYDROGENDATA KAHOYUEN1,2,A.LAZARIAN1 1DepartmentofAstronomy,UniversityofWisconsin-Madison 2DepartmentofPhysics,TheChineseUniversityofHongKong 7 ABSTRACT 1 0 The advancement of our understanding of MHD turbulence opens ways to develop new techniques to probe 2 magnetic fields. In MHD turbulence, the velocity gradients are expected to be perpendicular to magnetic fieldsandthisfactwasusedbyGonza´lez-Casanova&Lazarian(2016)tointroduceanewtechniquetotrace n a magneticfieldsusingvelocitycentroidgradients. Thelattercanbeobtainedfromspectroscopicobservations. J WeapplythetechniquetoGALFAHIsurveydataandcomparethedirectionsofmagneticfieldsobtainedwith 7 our technique with the direction of magnetic fields obtained using PLANCK polarization. We find excellent 2 correspondencebetweenthetwowaysofmagneticfieldtracing, whichisobviousviavisualcomparisonand through measuring of the statistics of magnetic field fluctuations obtained with the polarization data and our ] A technique. This suggests that the velocity centroid gradients has a potential for measuring of the foreground magneticfieldfluctuationsandthusprovideanewwayofseparatingforegroundandCMBpolarizationsignals. G Keywords:ISM:general—ISM:structure—magnetohydrodynamics(MHD)—radiolines: ISM—turbu- . h lence p - o 1. INTRODUCTION neticfield. Thispropertyofvelocitygradientswasemployed r inGonza´lez-Casanova&Lazarian(2016,hereafterGL16)to t Turbulence is ubiquitous in astrophysics. The Big Power s introduce a radically new way of tracing magnetic fields us- a LawintheSky(Armstrongetal.1995;Chepurnov&Lazar- ingspectroscopicdata.Insteadofusingalignedgrainsorsyn- [ ian2010)showsclearevidencethatinterstellarturbulenceex- chrotronpolarization(seeDraine2011),GL16appliedveloc- tendsover10ordersofmagnitudeofscalesintheinterstellar 1 ity centroid gradients (henceforth VCGs) to synthetic maps media(ISM).TheISMismagnetizedandthereforetheturbu- v obtained via MHD simulations and obtained a good agree- 4 lenceismagnetohydrodynamic(MHD)innature,e.g. see(Li mentbetweentheprojectedmagneticfieldsandthedirections 4 etal.2014;Zhangetal.2014;Pillaietal.2015). traced by the VCGs. As the velocity centroids can be read- 9 The modern theory of turbulence has been developed on 7 the basis of the prophetic work by Goldreich, (1995, hence- ilyavailablefromspectroscopicobservations(seeEsquivel& 0 Lazarian2005),thisprovidedawaynotonlyforobservational forthGS95).Theoriginalideasweremodifiedandaugmented . tracing of magnetic fields but also for finding its strength 1 in subsequent theoretical and numerical studies (Lazarian & 0 Vishniac 1999; Cho & Vishniac 2000; Maron & Goldreich using the GL16 technique that is similar to the well-known 7 Chardrasechar-Fermimethod. 2000; Lithwick & Goldreich 2001; Cho et al. 2001; Cho & 1 Motivated by the GL16 study, in this paper we calculate Lazarian2002,2003;Kowal&Lazarian2010, seeBranden- v: burg&Lazarian2013foraareview).1 TheAlfvenicincom- theVCGsusingHIdatafromtheGALFAsurvey(Peeketal. i pressiblemotionsdominatethecascade. Thiscascadecanbe 2011) and compare the directions of the magnetic fields that X we trace using the gradients with the directions of magnetic visualized as a cascade of elongated eddies rotating perpen- ar dicular tothe local direction ofthe field.2 Naturally, thisin- fields that are available from the PLANCK polarization sur- vey (Adam et al. 2016).3 To do this, we first significantly ducesthestrongestgradientsofvelocityperpendiculartothe improvetheprocedureofcalculatingoftheVCGsandtestit magnetic field. Thus one can expect that measuring the gra- with numerical data. Our recipe for calculating the VCGs is dientinturbulentmediacanrevealthelocaldirectionofmag- presentedin§2, whilein§3, weapplythetechniquetotrace [email protected],[email protected] magneticfields. Wediscussourresultsin§4,andourconclu- 1 WedonotconsiderthemodificationsoftheGS95modelthatwerein- sionsarepresentedin§5. tendedtoexplainthespectrumk−3/2thatwasreportedinsomenumerical studies(e.g.Boldyrev(2006)).Webelievethatthereasonforthedeviations fromtheGS95predictionsisthenumericalbottleneckeffect,whichismore extendedintheMHDcomparedtohydroturbulence(Beresnyak&Lazarian 2. IMPROVEDPROCEDUREFORCALCULATING 2010). Thisexplanationissupportedbyhighresolutionnumericalsimula- VELOCITYGRADIENTS tionsthatcorrespondtoGS95predictions(seeBeresnyak&Andrey(2014)). ThesimulationsalsostronglysupporttheanisotropypredictedinGS95and ruleouttheanisotropypredictionintheaforementionedalternativemodel. 2 ThenotionofthelocaldirectionwasnotapartoftheoriginalGS95 3BasedonobservationsobtainedwithPlanck(http://www.esa.int/Planck), model.Itwasintroducedandjustifiedinmorerecentpublications(seeLazar- anESAsciencemissionwithinstrumentsandcontributionsdirectlyfunded ian&Vishniac1999;Cho&Vishniac2000;Maron&Goldreich2000). byESAMemberStates,NASA,andCanada. 2 GL16 established that the VCGs can trace magnetic field inMHDturbulence. However,thisexploratorystudylacksa criteriononjudgingonhowwellgradientscantracemagnetic fields. Thereforeitisdifficulttojudgewhatistheresolution requirementtotracemagneticfieldvectorsandandwhatare the uncertainties. Therefore our first goal is to introduce a more robust procedure of the VCGs calculation which is to returnthetracingthatisindependentontheresolutionofthe simulationsandonlydependsontheparametersofMHDtur- bulence. Weusedasinglefluid,operator-split,staggeredgridMHD Eulerian code ZEUS-MP/HK,4 a variant of the well-tested code ZEUS-MP (Norman 2000; Hayes et al. 2006), to set upathree-dimensional,uniform,isothermal,supersonic,sub- Alfvenic turbulent medium. We adopted periodic boundary conditions. The initial cube was set with a uniform den- sity, and an initial uniform field. Turbulence was injected solenoidallycontinuously,e.g. see(Ostrikeretal.2000),see alsoAppendixofOttoetal.(2017). Oursimulationshadthe resolution of 7923. We selected two cubes with sonic Mach number M = 5 andAlfvenic Machnumber M = 0.6 but s A differentinitialmagneticfieldorientation(onewasparallelto the z-axis, another is at the angle π/7 to the z-axis). Com- paredtotheGL16,weusedhigherresolutionsimulationsand studiedtheeffectofvaryingmagnetic-fielddirectionrelative tothelineofsight. To trace magnetic field we generated polarization maps by projecting our data cubes along the x-axis and assum- ing that the dust producing the polarization followed the gas and was perfectly aligned by the magnetic field. Let φ = tan−1(B /B ),whereB aretheyandzdirectionofmag- Figure1. (Upperfour)Thedistributionofabsoluteangle(red) y z y,z netic field. The intensity I, velocity centroid C and stokes and relative angle (blue) in a synthetic map of size 792x792 parametersQ,U werecomputedby: for sub-regions of size 33x33, 50x50, 99x99, 198x198, re- spectively. The Gaussian profile emerges when the patch is (cid:90) 1/8ofthetotallengthofthemap. Theprofileiswell-defined I(r)= ρ(r,x)dx when it is 1/4 of the map. (Lower four) The distributions of (cid:90) absolute angle (red) and relative angle (blue) from observa- C(r)=I−1 ρ(r,x)v (r,x)dx x tion data for sub-region of size 50x50, 100x100, 200x200, (cid:90) 300x300(relativetoGALFA-HIdataresolution)respectively. (1) Q(r)∝ ρ(r,x)cos2φdx (cid:90) totheSoberoperatorusedinSoleretal.(2013).Wesmoothed U(r)∝ ρ(r,x)sin2φdx ourdatawithaσ =2pixelsGaussiankernel. The statistical properties of gradient fields can determine themeandirectionofmagneticfieldsinasub-regionofinter- where r is the vector on the y − z plane. The polarization est. We divided our synthetic maps into sub-regions and ex- angleisgivenbyφ2d =0.5tan−1(U/Q). Polarizationtraces amined the statistical behavior of gradient vector orientation themagneticfieldprojectedalongthelineofsight. (hereafterabsoluteangle(AA))andrelativeangleφbetween WecalculatedvelocitycentroidsfollowingGL16butmod- gradientsandfields(hereafterrelativeangle(RA))withinthe ified the VCGs calculations to increase the accuracy of the region. The upper four panels of 1 shows what distributions procedure. In particular, we performed cubic spline interpo- oftheAAandRAlooklikewhensizeoftheblockdecreases. lation,whichusesathree-pointestimatetoprovidethemaps As the block size increases, the mean gradient direction be- forgradientstudy. Theresultingmapis10timeslargerthan comesmorewell-defined. Thealignmentbetweenthegradi- the original one. To search for maximum gradient direction entandmagneticfieldalsobecomesmoreclearasblocksize in each data point, we selected a neighborhood of the radius increases. Wefindthatastheblocksizearrivesat 100×100, vectorr ∈(0.9,1.1)pixelsintheinterpolatedmap.Theinter- asharpdistributionemergeswithwell-definedmeananddis- polationprocessisaccuratewitha3oerror,andiscomparable persion. BymeasuringthemeanoftheAAdistributions,we determine the mean magnetic field direction within the re- 4MaintainedbyOtto&Yuen,(https://bitbucket.org/cuhksfg/zeusmp-hk/) spectiveblock.TheRAdistributionstellsushowaccuratethis 3 Figure2. (Upper16panels)ThedistributionofAA(red)andRA(blue)inasyntheticmapfromrun-2withblocksize198×198. By detecting the peak of the AA distribution, we determined the mean magnetic field direction within the block. (Lower) The predictedmeanmagneticfieldvector(red)comparedwiththerealmagneticfieldvector(blue). Thebackgroundistheintensity ofthesyntheticmap. 4 prediction of magnetic fields is. We shall call this treatment High Frequency Instrument (HFI).5 We performed the same sub-block averaging in the following sections. Notice that, procedureasindicatedinSection2. WecheckedtheAAand sub-block averaging is not a smoothing method. It is used RA,asshowninthelower4panelsof1,topickanappropriate to increase the emphasis of important statistics and suppress blocksizeforagradientvector. Forthegivencase, a100× noise in a region, and provide an estimate on how accurate 100blocksatisfiestherequirementintherecipe. Thevelocity thisaveragingisbytheAA-RAdiagram. Ontheotherhand, gradientvectorsareplottedwithpolarizationvectorsinFigure smoothingdoesnotprovidesuchanestimate. Adetaileddis- 3. Inthisregion,mostofthegradientvectorsalignverywell cussion of how white noise affects the sub-block averaging withpolarizationvectors. Thedetailedstudyoftheobserved andsmoothingisprovidedinanextendedpaperbyLazarian deviationsfromtheperfectalignmentwillbeprovidedinour et al. (2017), where the a companion new measure, namely, subsequentpublication. synchrotronintensitygradientsarestudied. Following GL16 we provide a comparison with the align- ThebenefitsofourapproachcanbeseeninFigure2.Wedi- mentmagneticfieldastracedbypolarizationandtheintensity videdthewholesimulationdomaininto16blockswithequal gradients. Theemissionintensityofatomicisproportionalto size,andpredictedthemagneticfielddirectionineachblock. itscolumndensity.Thecolumndensitygradientswereshown Asonecanseefromthesefigures,theVCGstracewellmag- toactastracersofmagneticfieldsis(Soleretal.2013).Figure neticfields. Wealsoconfirmedthisforsyntheticobservations 5 shows the histograms of relative orientations between ve- whenthelineofsightwasatdifferentanglestothemeandi- locityandintensitygradientvectorstopolarization. Inagree- rectionofmagneticfield. ment with the theoretical expectations as well as the results Chandrasekhar & Fermi (1953, C-F) provides an expres- in GL16, our improved procedure of calculating the VCGs sionrelatingthestrengthofofplane-of-skymagneticfieldby shows that the latter are much better aligned with polariza- dispersionofturbulentvelocitiesδv andpolarizationvectors tioncomparedtotheintensitygradients. Indeed,nearly80% δθ in magnetized turbulence (For an improved C-F method, of the VCGs are within 45o deviation from the polarization seeFalceta-Goncalvesetal.2008): directioncomparedto61%oftheintensitygradients. (cid:112) δv 4. DISCUSSION δB ∼ 4πρ (2) δφ 4.1. Structurefunctionsofvelocitygradients Themeanmagneticfieldstrengthcanalsobecalculatedus- The structure functions of polarization and gradient fields ing the same concept in sub-block averaging.The dispersion can also allow us to study how well-aligned they are. As ofVCGsandthatofmagnetic-fielddirectionsarenotexactly the statistics of polarization are dependent on the Alfvenic thesame,butthedifferenceissmall. GL16introducedafac- MachnumberM (Falceta-Goncalvesetal.2008),theclose A torγ of∼1.29toaccountforthisdifference. Inourcase,us- relationship between rotated the VCGs and magnetic fields ingourimprovedprocedureofgradientcalculationwegetthe suggests that gradient statistics should have similar behavior dispersionoftheVCGsinblocksthatisjust1.07-timesthatof to the polarization statistics. To compare the VCGs to po- polarization. Thestandarddeviationoftheratioofthedisper- larization in synthetic maps, we extended the sub-block av- sionsis0.05. AsillustratedinGL16,thefactorγ varieswith eraging algorithm to every point of our map, and computed parameters of MHD turbulence. Elsewhere we shall provide the structure function in terms of the orientation θ of gradi- afittingexpressionforγasthefunctionofMsandMA. This ent/polarizationvectors: shouldfurtherincreasetheaccuracyofobtainingthevalueof SF (r)=(cid:104)(θ(r(cid:48))−θ(r(cid:48)+r))2(cid:105) (3) magneticfieldstrength. Moredetailsonthetechniqueofob- 2 taining magnetic field intensity using only spectroscopic in- Thestatisticsofdustpolarizationareimportantforstudying formation and no polarimetry will be provided in our forth- magneticfieldturbulence(Falceta-Goncalvesetal.2008)and comingpaper(Yuen&Lazarian,inpreparation). forcleaningtheCMBpolarizationmaps. Ifwewanttodothe sameusingVCGs,itisimportanttesttowhatextentthestatis- 3. APPLICATIONTOOBSERVATIONDATA ticsoftheVCGsaresimilartothoserevealedbypolarization. The left and the middle panels of 4 show the power spectra With the tested procedure in hand, we selected diffuse re- P (k) and second order structure functions SF (r), respec- gions from observation surveys. We acquired data from the φ 2 tively,oftheVCGsorientationandthepolarizationangle. In Galactic Arecibo L-Band Feed Array HI Survey (GALFA- termsofthespectra,bothVCGsorientationsandpolarization HI).WecomparetheVCGsdirectionstothePLANCKpolar- angles exhbitit a −2 slope. We also examined the structure ization data. In diffuse media, polarization of emitted radia- functionsforpolarizationandtheVCGdistributionsfromthe tionisperpendiculartolocalmagneticfielddirection(Lazar- observationdatausingthesameprocedure.Therightpanelof ian 2007; Andersson et al. 2015), i.e. the same way as the Figure4showsthestructurefunctioncomputedusingobser- VCGs. To adapt the difference of resolutions, we adjust the vationdata,the+1slopealsoemerged. block size used in Planck to reflect the same physical block GALFAisreferringto. 4.2. Comparisonwithothertechniquesandearlierpapers TheregionweselectedfromGALFA-HIsurveydataspans right ascension 15o to 35o and declination 4o to 16o. The 5 We use the planckpy module to extract polarization bin size along the velocity axis is 0.18 km/s. We analyzed data in a particular region with J2000 equatorial coordinate: 353GHz polarization data obtained by the Planck satellite’s (https://bitbucket.org/ezbc/planckpy/src) 5 Figure3. Rotated VCGs (Yellow) map obtained using GALFA-HI data. Red vectors are polarization directions obtained from thePLANCKdata. Thedirectionspresentedinthisfigureshowthedirectionbutnotthemagnitude. Thebackgroundshowsthe columndensityofatomichydrogen. ThispaperpresentsthefirstapplicationoftheVCGstoob- ploredbySoleretal.(2013).Thealignmentofthesegradients servational data arising from diffuse media. By comparing withmagneticfieldisalsoduetothepropertiesofturbulence. theresultsobtainedwiththeVCGsandPLANCKpolarimetry For instance, Beresnyak et al. (2005) showed that GS95 tur- data, we have demonstrated the practical utility of the VCG bulence can in some situations imprint its structure on den- for tracing of magnetic fields and obtaining statistical infor- sity. However,densitydoesnottraceturbulenceasdirectlyas mationaboutmagneticfieldinthisdiffuseregion. velocitydoes. Therefore, weexpectmoredeviationsofden- Thegradienttechniqueshavebigadvantageoverothertech- sity gradients from the magnetic field direction compared to niques for estimating magnetic field direction and strengths: thevelocitygradients. Ourstudyconfirmstheconclusionsin These techniques only require an easily available centroid. GL16 that the VCGs provide a better tracer. We expect that Unlike the PLANCK map, the VCG maps do not require thedensitygradientsarerelatedtothefilamentswhichalign uniquemulti-billiondollarsatellitesbutcanberoutinelyob- withmagneticfieldsasreportedinClarketal.(2015). There- tainedwiththeexistingspectroscopicsurveys. Byusingdif- foreweexpectthattheVCGstracemagneticfieldsbetterthan ferent species, one can distinguish and study separately dif- thefilaments. ferent regions along the line of sight. Combining the VCGs We,however,havetostressthatthisregionisonlyapartic- thattracemagneticfieldsindiffusegaswithpolarimetry,e.g. ularexampleonhowVCGworks,whichdoesnotrepresentit ALMA polarimetry, that traces magnetic fields in molecular isapplicableeverywherewithoutcautionsonthelimitations. clouds,onecanstudywhatishappeningwithmagneticfields One should understand that both density and velocity prop- asstarformationtakesplace.Thismaybeawaytotestdiffer- ertiesareimportantcomponentsofMHDturbulentcascades. ent predictions, e.g. the prediction of magnetic flux removal Therefore, the deviations of the gradients from the magnetic through the reconnection diffusion process (Lazarian & A. field direction are informative. For instance, we observe an 2005,2014;Lazarianetal.2012). adifferentbehaviorofVCGsanddensitygradientsinthere- The alignment of density gradients were previously ex- gions of strong shocks as well as in self-gravitating regions 6 Figure4. The power spectrum (Left) and second order structure function (Middle) of the sub-block averaged velocity gradient (blue) and polarization (red) from the synthetic map.. Both power spectra and structure functions show very similar behavior. Structurefunctions(Right)ofthesub-blockaveragedvelocitygradientsandpolarizationanglefromobservationdata. OurworkprovideapromisingexampleonhowtheVeloc- 1.0 ity Centroid Gradient (VCG) technique introduced in GL16 VCG tracesmagneticfieldsininterstellarmedia. Inthepaper: IG 0.8 1. We provide a new robust prescription for calculating the nt VCGsandtestthisnewapproachusingthesyntheticdataob- u o e c tainedwithMHDsimulations. ativ0.6 mul 2. We show that with the new prescription the estimates of u d c magneticfieldstrengthbasedontheC-Fapproachcanbeim- alize0.4 proved. m or n 3. We apply the VCGs to the available high latitude HI 0.2 GALFA data and demonstrate an excellent alignment of the direction of the VCGs and those measured by PLANCK po- 0.0 larization. 0 10 20 30 40 50 60 70 80 90 Relative pseudo-angle between VCG/IG to polarization (degrees) 4. We show that the statistics of the fluctuations measured Figure5.Acumulativehistogramshowingtherelativepseudo- by the VCGs and polarization have the same slope for both angle between sub-block averaged VCG and intensity gradi- synthetic and observational data, which suggests that VCGs ent(IG)topolarization. Totracethemagneticfielddirection, couldpotentiallybepromisingtoolforaccountingforpolar- werotatedgradientvectorsby90degrees. izedforegroundswithinCMBstudies. 5. Thedifferencesbetweenthedirectionsdefinedbythepo- larization, the VCGs and the intensity gradients carry infor- (Yuen&Lazarian,inprep.).Thereforethereisimportantsyn- mation about the turbulent interstellar medium and this calls ergyofthesimultaneoususeofVCGs,density/intensitygra- forthesynergeticuseofthethreemeasures. dientsandpolarimetry.Addingtothelistthenewlysuggested techniqueofsynchrotronintensitygradientsthatisdiscussed WethankSusanClarkforherhelpwithGALFAdata. We inanewpaperbyLazarianetal. (2017)increasesthewealth thankAviLoebandDiegoF.Gonzalez-Casanovausefuldis- oftheavailabletools.Thisopensnewwaysofexploringmag- cussions. WealsothankPaulLawforhisgeneroushelpwith neticfieldsinthemulti-phaseISM. PLANCK data. The stay of KHY at UW-Madison is sup- Wewouldalsoliketopointoutthatwhilethepolarimetry ported by the Fulbright-Lee Hysan research fellowship and directions in Figure 3 seem to be well aligned over signifi- DepartmentofPhysics,CUHK.ALacknowledgesthesupport cant patches of the sky, this does not mean that there is no theNSFgrantAST1212096,NASAgrantNNX14AJ53Gas turbulence there. The correspondence of the VCGs and po- well as a distinguished visitor PVE/CAPES appointment at larization directions can be understood only if the media is the Physics Graduate Program of the Federal University of turbulent. Thepowerlawbehaviorofthestatisticsrelatedto RioGrandedoNorte,theINCTINEspaoandPhysicsGradu- boththeVCGsandpolarizationdirectionsconfirmsthis. The ateProgram/UFRN. factthatthepowerlawdoesnotcorrespondtotheGS95slope isduetotheeffectsoftheemittingregiongeometryasitdis- REFERENCES cussedinCho&Lazarian(2002,2009). Adam,R.,Ade,P.A.R.,Aghanim,N.,etal.2016,Astronomy& 5. CONCLUSIONS Astrophysics,594,A8 7 Andersson,B.-G.,Lazarian,A.,&Vaillancourt,J.E.2015,Annu.Rev. Goldreich,P. &Sridhar,S.1995,TheAstronomicalJournal,438,763 Astron.Astrophys,53,501 Gonza´lez-Casanova,D.F.,&Lazarian,A.2016,ArXive-prints, Armstrong,J.W.,Rickett,B.J.,&Spangler,S.R.1995,TheAstrophysical arXiv:1608.06867 Journal,443,209 Hayes,J.C.,Norman,M.L.,Fiedler,R.A.,etal.2006,188 Beresnyak,A.2014,ApJL,784,L20 Kowal,G.,&Lazarian,A.2010,TheAstrophysicalJournal,Volume720, Beresnyak,A.,&Andrey.2014,TheAstrophysicalJournalLetters,Volume Issue1,pp.742-756(2010).,720,742 784,Issue2,articleid.L20,5pp.(2014).,784 Lazarian,A.2005,MAGNETICFIELDSINTHEUNIVERSE:From Beresnyak,A.,&Lazarian,A.2010,TheAstrophysicalJournalLetters, LaboratoryandStarstoPrimordialStructures.AIPConference Volume722,Issue1,pp.L110-L113(2010).,722,L110 Proceedings,Volume784,pp.42-53(2005).,784,42 Beresnyak,A.,Lazarian,A.,&Cho,J.2005,TheAstrophysicalJournal, Lazarian,A.2007,J.Quant.Spec.Radiat.Transf.,106,225 Volume624,Issue2,pp.L93-L96.,624,L93 —.2014,SpaceScienceReviews,181,1 Boldyrev,S.2006,PhysicalReviewLetters,96,115002 Lazarian,A.,Esquivel,A.,&Crutcher,R.2012,TheAstrophysicalJournal, Brandenburg,A.,&Lazarian,A.2013,SpaceScienceReviews,Volume Volume757,Issue2,articleid.154,20pp.(2012).,757 178,Issue2-4,pp.163-200,178,163 Lazarian,A.,&Vishniac,E.T.1999,TheAstrophysicalJournal,Volume Chandrasekhar,S.,&Fermi,E.1953,TheAstrophysicalJournal,118,116 517,Issue2,pp.700-718.,517,700 Chepurnov,A.,&Lazarian,A.2010,TheAstrophysicalJournal,Volume Li,Z.-Y.,Banerjee,R.,Pudritz,R.E.,etal.2014,arXiv.org,1401,2219 710,Issue1,pp.853-858(2010).,710,853 Lithwick,Y.,&Goldreich,P.2001,TheAstrophysicalJournal,Volume562, Cho,J.,&Lazarian,A.2002,PhysicalReviewLetters,vol.88,Issue24,id. Issue1,pp.279-296.,562,279 245001,88 Maron,J.,&Goldreich,P.2000,TheAstrophysicalJournal,Volume554, —.2003,MonthlyNoticesoftheRoyalAstronomicalSociety,Volume345, Issue2,pp.1175-1196.,554,1175 Issue12,pp.325-339.,345,325 Norman,M.L.2000,6,6 —.2009,TheAstrophysicalJournal,Volume701,Issue1,pp.236-252 Ostriker,E.C.,Stone,J.M.,&Gammie,C.F.2000,56 (2009).,701,236 Otto,F.,Ji,W.,&Li,H.-b.2017,arXiv:1701.01806 Cho,J.,Lazarian,A.,&Vishniac,E.2001,TheAstrophysicalJournal, Peek,J.E.G.,Heiles,C.,Douglas,K.A.,etal.2011,TheAstrophysical Volume564,Issue1,pp.291-301.,564,291 JournalSupplement,Volume194,Issue2,articleid.20,13pp.(2011)., Cho,J.,&Vishniac,E.T.2000,TheAstrophysicalJournal,Volume539, 194 Issue1,pp.273-282.,539,273 Pillai,T.,Kauffmann,J.,Tan,J.C.,etal.2015,74 Clark,S.E.,Hill,J.C.,Peek,J.E.G.,Putman,M.E.,&Babler,B.L.2015, Soler,J.D.,Hennebelle,P.,Martin,P.G.,etal.2013,16 PhysicalReviewLetters,115,1 Zhang,Q.,Qiu,K.,Girart,J.M.,etal.2014,1 Draine,B.T.2011,Physicsoftheinterstellarandintergalacticmedium (PrincetonUniversityPress),540 Falceta-Goncalves,D.,Lazarian,A.,&Kowal,G.2008,TheAstrophysical Journal,Volume679,Issue1,articleid.537-551,pp.(2008).,679

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.