Sweating the small stuff ▸ KITP Colloquium by Coral Wheeler

[00:00:04]yes uh thank you so much for inviting me I'm very excited to tell you all about some of the work I've been doing over the past few years specifically how I learned to start worrying and love smallest galaxies and of course as with any project this would not have been possible without an excellent group of collaborators which may or may not need to be updated and so first before I get down to why I love small galaxies let's talk about what small is what do I mean by small stuff I'm a galaxy formation theorist but there may be planets people stars people particle people and so maybe you don't think of galaxies is very small but just to give a sense of scale I'm gonna start out with our own Milky Way almost 10 to the 11 solar masses it's very large it contains us of course and it's very wide across 30 to 50 kiloparsecs you know which is huge it looks very much here like a very blurry picture that someone invented oh but if we want to look at a real galaxy we can then go to our next-door neighbor Andromeda now Andromeda is even more massive 10 to the 11 solar masses and these are far from the largest galaxies in the world I think I was seeing on Twitter some time recently people's just existential dread at this IC 1613 or 18 and how much more massive it is but let's not go up let's go down I'm gonna talk about the small stuff if we zoom in to this plot here and then zoom in again now we're starting to get to the level of galaxies that I want to talk to you about today not quite though this is the LMC it is a dwarf galaxy but it's much more massive than the galaxies I'm going to discuss a billion solar masses and yet you know a fraction of the size of its Milky Way which of which it's a satellite then we have Sagittarius now Sagittarius is being currently plunging through the disk of the Milky Way it's been totally stripped and stretched out so it's very large on the sky but it has about the stellar mass that I'm interested in but in what at some point was much more massive so let's zoom in again and zoom in again okay now we're finally at the scale that I'm talking about we've got Draco this is about the stellar mass that I'm focus on on the high mass end it's what's a classical dwarf spheroidal galaxies few several hundred thousand stellar masses in stars but there's even smaller galaxies this is on the lower mass end of what I'm going to discuss today the ultra faint galaxies this um is the lowest mass I believe or one of the lowest mass galaxies ever discovered and you can see it here on the sky or rather you can't see it and you can see where you might see it if you were if you were able to see the stars these things take many complicated algorithms to determine membership but this is an ultra faint galaxy so these are the types of galaxies I'm going to discuss today and it's a really amazing time to be in the field of dwarf galaxy formation particularly observational II we've known about dwarf galaxies this is a very nice plot from Josh Simon's recent review confirmed or candidate Milky Way dwarf galaxies and we've known about these objects for quite some time but you can see we just sort of trickle along finding new ones every few years until the whole dwarf galaxies world was revolutionized for the first time I'd say or one time recently by the the advent of digital surveys like Sloane but now with DES with pan-starrs hypership prime kam and even with Gaia we're currently going another revolution in low mass galaxy formation detections and specifically this time at the low mass end and it's only going to get better or worse depending on how much you like these objects or how much you think they're causing problems with the next generation of ground and space-based observatories LSST coming online very soon should find many more of these objects and of course we're going to need a lot more of analysis or about to not only find more and more of these things I would suggest but learn quite a bit about them so adore theee galaxies are cool in and of themselves but they can they can teach us about a lot of things north galaxies can teach us about the very nature of dark matter I'm going to spend probably the majority of talk of my talk going over this point but dwarf galaxies can also teach us about the cosmic ray ionization background and also about star formation and feedback because they're very sensitive basically to anything you throw at them so if I drag a little too long and so this part don't worry these other two parts won't be as long so Dark Matter okay as we know every galaxy we believe Liz in a giant tri-axial ball of this mysterious substance called dark matter dark matter is a really funny thing because we we don't know what it is we don't know what particle is we don't even know if it is a particle but we know an awful lot about it - was it to quote Paul Rumsfeld I think this would be a known unknown it would be your Dark Matter it interacts through gravity its pressureless which is what allowed it to clump together in the early universe the cold part of cold dark matter means that it was nonrelativistic allowing it to form smaller and smaller clumps and it does not shine or reflect light which makes it very difficult to of course detect and measure but luckily it has of course an impact on baryonic matter which is helium hydrogen stars planets you mean everything in the periodic table oh I hope my volume is off here oh all right so this is a cool [Music] I can't turn the problem is I can't turn the volume off unless it's unplug so that's why I didn't just immediately do that so let's try that again oh cool I should have kept it on all right okay all right so the early universe together and in form structures very specifically according to our best standard framework and as you can see in this we can make massive computer dark matter all these simulations all the exact way in which it's supposed to cluster and this is a prediction of lambda-cdm because dark matter doesn't shine or respond quite how do we test this okay oh my goodness okay probably skipped half what I was supposed a with anyway okay so a lambda-cdm predicts how dark matter should be structured and clustered um but we need to use what we can see to test this luckily we have now with the advent of these surveys such as sloan which is mapped like a third more of the sky we know how galaxies actually cluster in space so we should be able if we assume remember that galaxies form in these Dark Matter halos we should be able to test our model with these galaxies of course then of course we ran into a problem how do we know which galaxies what goes in the Dark Matter halo and like any good physicists we just did the dumbest thing possible and we said let's just assume the biggest galaxy goes in the biggest halo and see what we get and doing that simple or a dumb thing whatever the initial simplest thing called abundance matching does a really good job of predicting dark matter structure across large scales Milky Way and larger this is huge chunk of the universe so one of these objects is or one of these images is the real universe the structure of actual galaxies the other is a mock universe made from the boys bolshoy simulation and I'm gonna ask you guys to guess which is which we're looking for team real universe who thinks the real universe is red I know one okay who thinks is blue all right if you guys were up my talk last here at the kitv that's not fair or maybe you saw the man here I'd flip it upside down and switched it around so usually that gets you but anyway the point is they're statistically indistinguishable lambda-cdm does a fantastic job of predicting the way galaxies should be strung clustered on large scales what what about a small scale see dark matter according to lambda-cdm behaves in a hierarchical manner right so these giant clusters have halos these halos have sub halos so if we take a Dark Matter halo that we expect to host the Milky Way that's still where abundance matching is going to work really well and we zoom in on should be yeah it's the feel ya via lactea of stimulation so this is what we would expect to host a galaxy like Andromeda or the Milky Way we can see that lambda-cdm predicts that it should have hundreds or even thousands or even ten thousands of these tiny other sub halos orbiting around it however we only if you look out into the real sky around the Milky Way instead of hundreds or thousands of satellite galaxies we only see about 50 this is known as the missing satellites problem and if you guys find if any of you guys find yourself like groaning and getting tired why is someone talking about the missing satellites a problem again well you're not alone in fact I once heard Carlos Frank say he would leave the room if he had ever heard anyone mentioned the missing satellites problem again yes yes I mean that's been shown there's there's there's been a few papers including including bite yeah people at UC Irvine they shown that they it destroys not only galaxies but the actual of some pillows themselves so that's one way to do it but to be honest you don't actually need to do it and in fact Andrew Wetzel was here at the kitv at the Gaia conference and I think he repeated four or five times there is no missing satellites problem standing from about right here and even without even without the destruction by a disk by a disk potential it's true down to the point where we're observational incomplete around the Milky Way so down to about ten to the five and I'm just gonna try to promote this sort of new jargon we always need new jargon in astronomy so this is james bullock and mike BK we're talking about bright dwarfs at the very high mess and your LMC smc classical dwarfs these you know are a few million solar masses and then ultra phase bright dwarfs and classical dwarfs there really is no missing satellites problem they match it's within error between so this is the number of galaxies as a function of stellar mass and it matches surprisingly well and this is you know without any tweaking necessary but look a alter phage worse right we don't know yet how many ultra fade dwarfs are around the Milky Way we're certainly not observational II complete at this mass level and even more we don't even know the number that is really predicted by lambda CDM right we about know about fourteen but how many will we see one hundred two hundred three hundred maybe in the next few years we'll find out so this is one of the ways that dwarf galaxies can tell us about the nature of dark matter even counting them up and comparing them to our best theory can help discriminate between two different Dark Matter models so here we have CDM which is what I've been talking about until this point on the left and for CDM even though even if there aren't that many tiny dwarf galaxies it may not be an M coffin for CDM because we don't expect every single tiny some halo to have form stars or form to galaxy we expect that when the first stars did form they heated up the intergalactic medium to certain temperature that prevented these tiny Dark Matter halos with their shuttle shallow potential Wells from holding on to their gas so that's the solution for lambda CDM uh if that doesn't work there's another solution we just changed our Dark Matter model it's the Dark Matter particle it's a little less massive a little warmer then did small there's a cut-off and such that these tiny Dark Matter halos don't even form at all so what I want to do and one of the things that kind of a themes I'm going to say is run these lambda-cdm simulations to see can we fix the system within lambda-cdm or do we need to throw out our model altogether now there are other people in the fire group running simulations with warm Dark Matter I'm gonna stick to cold dark matter but yet we can learn about the nature of dark matter however if we're gonna make a firm testable prediction for the number of satellites that we expect to see around the Milky Way which is the only way we're gonna be able to test the theory well then we need to know what's the minimum mass for galaxy formation even though we know that the smallest halos will never form stars we don't know what that if there's a sharp mass cut off below which the Dark Matter halo cannot form a galaxy and if there is a sharp mass cut off we certainly don't know what that mass is and and that's quite I think striking in 2019 of course there was way back in 2008 a little bit of a suggestion that maybe we do know this mask l-lois regard measured the dynamical mass they saw in the velocity dispersion within three hundred parsecs for all of these satellites of the Milky Way dwarf spheroidal over I think nearly five orders of magnitude in luminosity and Lois found that they were all had the same dynamical mass they're consistent with occupying Dark Matter halos of about three times ten to the nine now one sort of conclusion that was put forward is well well that's it that's the minimum halo mass coral what are you talking about this is it it's three times ten to the nine however there is an alternate explanation and is that it is a selection effect if you have these Alexei's and with the same velocity dispersion if for some reason maybe maybe supernova explosions or heating of some sort there's a floor on the velocity spurgeon if you take that stellar population and put it in a lower mass Dark Matter halo as James Bullock showed in his 2010 paper then it puffs out the half-light radii becomes much larger it becomes more dim yes yeah there's quite a few assumptions that go into this it's and it's maybe not the best I think there you know there's been improvements to the modeling but I think it is interesting to note that if for some reason there's a floor on your velocity dispersion if you have galaxies and lower mass Dark Matter halos they may be invisible James call these you know another clever name let's what James well we're all kind of known for that stealth galaxies and there might be hundreds or thousands of them okay so how do we go about figuring out this problem right we can't just keep we don't have enough ultra feints to do this abundance matching thing it clearly doesn't work so the way I want to go ahead and solve this problem is to run a different sort of simulation one that actually forms these tiny galaxies right fully hydrodynamic gas physics star formation and that with enough detail to actually simulate the tiny galaxies of interest now if to really simulate ultra faced with uh with a lot of resolution it's still really computationally prohibitive to do something on this mass scale this would be a Milky Way and drama to sort of pair so to kind of sidestep the issue for now until we can get faster computers I'm gonna cheat a little and zoom in on an isolated dwarf galaxy so I just want to be clear there are no milky ways in any of my simulations that I'm gonna show you to you today they're far away from a more massive neighbor these are dwarf galaxies Dark Matter halos of about ten to the ten solar masses now these are large enough that we think we know the type of galaxy that forms in them something of about a million solar masses and because dark matter behaves hierarchically again these things have their own sub halos you can see that even a 10 to the 10 solar mass Dark Matter halo should have some sub structure some of them are less than this 3 times 10 to the 9 right so if I'm trying to find out if it's possible to form galaxies a low mass dark matter they're massive enough for low-mass enough to be interesting and we might be able to predict whether or not Worf galaxies have satellites so that's basically what I've been doing for the last several years starting with my thesis and moving on through Caltech I've been part of the feedback and realistic environments or fire project with code primarily written by Phil Hawkins and fire is a is a code it has all the bells and whistles it's primarily particle based although it's a slightly more complicated hydro solver but it's kind of crowning achievement was to be one of the first to really get sort of realistic feedback now feedback is when a star forms it's basically energy mass and momentum fed back into the surrounding gas from That star and includes both radiation pressure from young massive stars stellar winds from Oh AGB type stars ionizing radiation both locally and cosmic background and then of course the supernova here and you can see this every time this sort of puffs up there that's the supernova exploding really puffing up that gas and one of the sort of hallmarks of the dwarfs and fire is that they're very bursty they're having a lot of expulsion of gas oh another cool thing about fire or yeah is that it allows us to push to a very high resolution higher especially the new the new version so I'm just going to kind of give a sort of survey of resolutions up until what I've done of course actually is changing everyday there's people are pushing pushing down on resolution and so when we have a particle based simulation we talk about your dark matter particle mass your I say star particle mass but it's really a gas or baryonic particle mass that then turns into a star when it reaches certain conditions and then spatial resolution or the softening link between the simulations before my 2015 thesis your typical dwarf galaxy resolution still had a bearing on a particle mass of well 10 to the 4 which you could do a lot with is very interesting but remember the galaxies I'm trying to form were a thousand solar masses so you're not going to be able to do that much structurally with a gas particle mass or star particle mass that's larger than it's larger than your okay it's larger than your galaxy so in 2015 in my thesis I pushed it down to what we're then the highest resolution Cosmo is called zoom in simulations around regice cero 250 solar mass baryonic particle and just about a parsec spatial resolution um and we did you know I think we're able to look at a lot of things about the dwarf galaxies but again we're talking about tens of particles for an ultra faint so we tried to go even higher so my latest paper which is you can find on the archive here is oh yeah so I think I was just beaten out recently so it's the highest spatial resolution cosmological zoom in simulation ever run a redshift zero there's another one out that has 20 solar mass baryonic particles but I want to point out um this spatial resolution is it's sufficient to resolve the internal structure of within the blast radius of an individual supernova with a hundred elements so this this is important because the inability to resolve this set off Taylor phase of expansion in a supernova has led to a lot of sort of debates on how you handle that sub grid approach and so this is the first time actually that we'd be able to been able to get in and get that internal structure ok so sorry the SPH particle mass is 30 solar masses yeah sorry I should have said I should have said gas particle there the gas port this our particles are just a tiny bit less they just from from mass loss okay so I talked about this as a particle based simulation when you have high or lower resolution whether the traditional way we do it is to treat a single star particle as a it's a single stellar population it's like a star cluster that evolves from a gas from a gas particle when it reaches a certain temperature density and other criteria and you basically IMF average all of the properties the winds the supernovae and we use a Krupa IMF but the problem is once you start to get down to about 30 solar masses or anything less than 100 it you're you you can no longer represent a complete population right you're gonna have a fraction of an O star and a fraction of a type 2 supernovae so what we have to do now is stochastic Li sample the IMF so now each star particle has a certain random sampling of these and in the case in the rare case that it's a massive Oh star then this 30 solar mass particle is going to go around and it's going to behave as a single massive Oh star doesn't gain in mass but it's going to have the full supernova explosion and it averages out over all the particles to still conserve mass and this has been tested at lower resolution by Cooney soo if you're interested in the details of that okay so here's what here's what I've done the other problem with high resolution is you can't run that many of them so in 2015 I ran four dwarfs to classical these are ten to the ten don't in dark massive solar masses in dark matter these are two isolated ultra fates in the more recent high-resolution version we also have to classical Dwarfs and one ultra faint and as you can see um kind of the main point of my thesis was that we fought every one of the massive dwarfs here has a satellite galaxy so we predict satellites around isolated classical Dwarfs um but it weren't more generally well in 2011 we formed in fact there are two satellites which you can see here in the triangles we formed 11 galaxies total in the high resolution region so some ultra fates that are isolated in their own right but still in the high resolution region and these all have a hundred star particles or more and you can see here there's the abundance matching relations expanded from higher mass so using these simulations um we make a prediction oh yeah sorry so we make a prediction that galaxies exist in all Dark Matter halos so every dark matter halo greater than about 5 times 10 to the 8th is populated by a well resolved galaxies a hundred star particles or more now that's not testable I'm interested in making testable predictions or it's not testable yet so we predict that there should be satellites of isolated Dwarfs and there should be these things should be everywhere in the field once we have the ability to find these ultra fates outside of the virial radius of the Milky Way we predict you know many whoo of them depending on if you mean of satellites of more massive Dwarfs or not so these things are everywhere however I'm sad to say I still don't have a prediction for the minimum of mass of galaxies information that's one of my primary goals and yet in order to figure that out we have to push farther down here this is my particle limit this is 100 particles so this is the more broad stellar mass scale amassed relation you can see galaxies are forming and Dark Matter halos are single star particles depending on what you want to call a galaxy single star particles are forming in tiny Dark Matter halos so what your minimum mass is always going to append on depend on your resolution and depend on what you'd consider a galaxy this will be my next paper we'll be investigating this lower mass and I'll get a drop the particle count and look at the extreme massive galaxy formation what oh sorry so this is from one galaxy this is one of my classical Dwarfs this is a stellar mass Haila mass relation for every thing that forms at least one star particle here's where I cut off from the previous plot here this is just everything that's a hundred star particles or more but if I if I get rid of that limit then this is what I get so these are the ones from this one halo and this is uh everything or these are the yeah everything that forms one star particle or more yes in a sub halo yes in a sub halo you can see them here they're very beautiful one other cool thing we can do with this higher resolution so these are kind of very simple images but my colleague che garrison Kimmel who makes all the beautiful fire images might have seen has has mocked up a false color image of what this door would actually look like in real life you guys ready I don't know what you were expecting this is a tiny dwarf galaxy they're not that interesting visually its density temperature self gravitation and yeah genes unstable and molecular fraction they vary some of them it depends on its orbit and that's actually my big project for my next postdoc is gonna be smashing these things into a milky way to see what happens because that's the one thing I mean I would say that's the main thing missing for my simulations and it's going to come up again but it's hard to do that with this kind of resolution one yeah sorry well I'd yeah it depends on I think at equal mass yeah because now there you have more if there's just higher mat by nature being higher mass will have more substructure probably do that for the same beautifu mass match than those so that might be difficult um okay so one cool thing we can do with hundred stellar particles that were I wasn't so much able to do before is to put these things on a mass size plane so this again is a half mass radius or you could say half flight radius stellar mass and the all the black open circles or classical observations of dwarfs the black points with error bars there des your one in your two result I don't want to point out like if you want to post a simulation on this plot you really need you need to be able to resolve it right you can't what's the half-light radius of a single star particle and just to give a sense of where kind of standard simulations lie I'm gonna put the resolution spatially in mass for a couple simulations you might have heard of this is a big box simulation a lustrous so it has a particle mass nearly a million solar masses and a spatial scale of almost a killer almost a kiloparsec now it's not designed to do this of course it's a big box simulation it does a lot of other stuff but just to give you a sense you would only be able to look at this portion of the plot dwarf simulations until recently did it got a little better they are able to get to about a hundred spatial scale and ten to the four but again even if you're falling in this region you might just be a single star particle and you can't get these des Dwarfs even our own lot a simulation which again is a Milky Way simulation but it's one of the highest resolution Milky Way simulations isn't doing much better my this I'm talking about or off the plot so we're well able to resolve this okay so the high mass object that I've placed on here these are just the low resolution versions everything in the high mass and actually lies right on top of these observations and in fact they form a really nice line right there's kind of just going along like this almost suggesting some sort of trend between half a flight radius and stellar mass and in fact a universal trend between at least halo mass and in half mass or half light radius was proposed back in 2011 by Andre cross off and it's over you know many orders of magnitude however you can see these objects are really kind of bunching up oh I should point out these lines here are lines of constant surface brightness so this is 30 magnitudes per square arc second and what I've marked as very dramatically is unobservable is 32 and a half that's probably where LSST can just about get to in terms of surface brightness right so for me you can imagine for the same stellar-mass you know if you have a morphs right out you're gonna have a lower surface brightness okay so this suggests though that maybe this is nothing more than a selection effect let's see where our tiny Ultra fades lie on this plot I'm gonna put him down oh they were actually there the whole time they are really really puffy they're really puffy these things have you know some of them nearly a kiloparsec in half mass radius definitely a few hundred a good you know nearly order magnitude or more higher than observations ah but you know maybe that's not so crazy there have been all sorts of new discoveries in the low surface brightness sky the most famous being the ultra diffused galaxies and these objects are known for having basically the mass of dwarfs but the half-light radii of the Milky Way there they're much more massive than the objects I'm talking about but they have very low surface brightness --is there definitely could be a too many satellites problem and that would be one way we could maybe err on the side of well no yes there might be - there might be too many satellites problem mmm okay definitely could be these objects yeah so I hope to find out more about that but yes I we used to call it the found satellites problem which too many satellites is probably better have found satellites probably oh but it's possible is definitely possible okay so oh yeah and even more recently the lowest surface brightness galaxy ever detected is was with the Gaia data released to Antalya - I placed it right it's got a met surface brightness of thirty two point three magnitudes per square arc sang it which is somewhere you know in between here it's again more massive than the objects I'm talking about but it suggests maybe this apparent trend between massive size is really a selection effect and we just can't see all the objects up here right and if we could if they are out there um yeah hopefully we might be able to see a few of them with LSST okay so I think I've at least convincingly explained why we can't see any of these observations up here but you're probably wondering why we don't see any of the simulations down here right why are all of my galaxies puffy I was uh it is possible and this is something I'd like to look into about possible numerical heating um if you want to talk to me about that Joe and I were talking about the possibility because we use different dark matter and stellar masses anyway sorry so uh why don't we see any object here so one thing I wanted like on this theme of selection effects that I think I've been mentioning at least twice in my talk so we call it a theme on the theme a selection effect I thought maybe this is a selection effect what if let's say we have a surface brightness profile of our dwarf galaxies and what if we can only see the inner bright center of these objects what if like we do a surface brightness limit so I'm gonna pick sorry there's two numbers on you I'm gonna pick 32.3 magnitudes per square arc second in honor of aunt Leah - right what do we can only see these bright centers where would these objects and move on the chart so keep your eye on the colored points here whoa okay that's quite a difference right so if we these things the 32.3 magnitudes for square arcsecond they move along lines of constant surface money but look our higher mass ultra fans do start to go over towards the des Dwarfs this is just the very preliminary work here but it could be that some of the des dwarfs are actually more the bright centers of more massive dwarfs with undetected stellar halos beyond observational limits I'll be looking into this more in the future again this would mean you know could mean all sorts of things in terms of how we're misunderstanding the properties of these objects yes that's also yes that's yeah it could be strict yeah it's just stars and half-mast radius yeah I would again I would say one of the weaknesses of my work so far then I'm going to correct soon is the lack of a milky way I'm in some ways comparing apples oranges but it's really the best we can do now okay so still on dark matter how else can to work galaxies teaches bard this is dark matter part two yeah yep yeah you have to there's a lot oh yes you have to decide what your galaxy is basically first we yeah that's true we should yeah we should yeah well they might be missing stars too yes well yeah and that that's what I'm suggesting might be happening I spoke with Josh Simon and Alex G about it at Carnegie it's something I can look into it they said it might be easier to preferentially miss stars on the outskirts of these objects again it's not really a surface brightness cotton it's effective surface brightness but okay alright again so don't worry this is the second part of darkness okay so I didn't write it out here but another challenge to stand the lambda-cdm at the low mass and maybe that doesn't produce so many groans yet as the missing satellites problem is the core cost controversy and this suggests basically this came about when we looked at the dark the density profile of in Dark Matter only simulations and when that was done it was found they had a steep cusp e density profile at the center and this was you know at all galaxies masses however when we measured the Dark Matter density from galaxies rotation curves for some Dwarfs it's just not the case it's much shallower it's a flat low-density core now this is another way we can sort of use dwarf galaxies to distinguish between models because although there's a lambda CD there is a potential way out that I'll talk more about about something heating up the dwarf galaxy another way out is to allow dark matter to have some sort of self interaction cross-section allow it to scatter off itself if it does this then it heats itself up and has and creates darkmount of course at least for Dark Matter only simulations over a wide variety of halo masses okay but again back to lambda-cdm this is what I do is I try to see if we can solve these problems within lambda-cdm I'm not wedded to it but that's just so far when I'm looking into because if only if we can make these firm predictions do we know if the model is correct or not okay pause McGovern I don't have this really nice little cartoon that shows how baryons can heat up the Dark Matter creating a core you have your star particle orbiting a clump of gas supernal goes off very quickly gas is blown out so the star particle moves out the gas falls back in slowly so the dart the stellar particle stays at its orbit if you do this if you keep blowing the gas out having it fall back in blowing it out and having a fall back in you can create a dark matter core this has shown in many different simulations by many people but I will point out some work by someone at fire TK chan also think is almost not at UCSD anymore um used basically showed and I've just shown some of them here that for about 10 to the 11 solar masses in Dark Matter a little bit for 10 to the 10 and a little bit less for 10 to the 12 you can form these cores in dark matter and it really 10 to the 10 solar masses which is the classical Dwarfs that I've been telling you about is right at the sort of transition between ultra fates where in fact we're they're not predicting course and ultra fates and ten to the ten where you can form a core now this is one way it's been suggested that we can distinguish once and for all between self interacting dark matter and non interacting dark matter is if we find a core and an ultra faint warf galaxy yeah it's basically um yeah yeah it depends on the stellar mass of the halo mass relationship shown by ariane dentist in teo here there's just too much - too much dark matter the potential well is too deep for it to really have much of an effect and then at higher mass you have kind of explosions going off not it's not blowing all of the gas out of the Milky Way and back in again right it's just kind of popping off in the spiral arms that's why there's really a key stellar-mass - hail amassed relationship where you have a core of the you've got the best between about for a stellar master hail amassed relation we bout ten to the minus two to ten to the minus three okay so my simulations do form we don't I mean we're really kind of right at that transition phase so we do form one core and this is the only galaxy that has a stellar mass to halo Master lit ratio of above ten to the four to about ten to the minus three and it's a very you see it's pretty good core a particularly higher res but that's just because it forms a little more still in math um and everything else has cusps down about at least 100 per sec parsec so you can see this is the density is a function of radius this is more basically it gets a little more strict than the power criteria where we consider it resolved where we can trust numerically our results and you see we have cusps okay so we can form chorus great everyone can for course I want it this is the more important point that I want to make this is our lowest Dark Matter halo mass of any of our galaxies it form it's only two point seven times ten to the seventh it does form a little bit of stars but it definitely has a stellar mass to halo mass ratio of way below ten to minus five it kind of that's kind of a core there I just want to point out it's not going to break lambda-cdm if we see tiny cores I want to point out when people talk about cores they're talking about kiloparsec scale quarks I want you to remember this I want you to remember this so much I did a really terrible haiku here for you guys it's really bad ultra fake or seen less than a few hundred parsecs lambda-cdm so remember it's okay it's good I tried to make it better than that I apologize okay all that time forty minutes I spent on dark matter I'm going to now kind of go quickly through the last two ways that dwarf galaxies can teach us about realization and then stellar feedback okay so already talked about how early in the universe the ionizing background would like prevented the gas from cooling into the smallest halos this was shown in 2000 by James Bullock who showed that basically if all galaxies that form after realization below a certain matthos don't form and you can actually match the number satellites with the server with the observable halos at this mass so this may help solve the music's the Chumley's problem but what effect does it have on the galaxies that do survive ultra fates have shown by Tom Brown at all in 2014 uniformly ancient stellar populations these are six Ultra fate satellites of the Milky Way again they're all Milky Way satellites and they have identical color magnitude diagram consistent with all having their star formation shut down by redshift of two and this is about where the epoch of Rihanna's ation occur so it's one way to interpret this is hey that's Rihanna's ation however their satellites right I've heard people you know many people say oh they all got shut down when they fell into the Milky Way it's hard to distinguish because we only have Milky Way satellites so one some work that was done by Katie Rodriguez Wimberly at UC Irvine used Dark Matter only simulations to look at the in fall times of the Dark Matter Hill as we think host alter fate and even if the abundance matching is a little off you can see this is a probability that they fell in by a certain redshift the probability of these objects fell into the Milky Way or any host this is for all six the probability that all six fell in by redshift of one let alone two is less than 1% so that's one way to do it right these things all fell in at least some of them fell in after redshift of one no it does the way I want to do it is I can actually just look at the galaxies that form in my simulation so here's their stellar mass halo Master elation color-coded this is the cumulative fractional star-formation history I'm just going to color code it all the ultra faints aren't cyan oh the massive galaxies in magenta and you can see every ultra fate and again there's no Milky Way at all so now there's no Milky Way to fall into we do have some satellites but they're satellites of dwarfs and all of them have their star formation shut down prior to red shift of two meanwhile the more massive galaxies have ongoing star formation almost a redshift is zero one of the higher mass ones does actually self quench this is interesting it kind of freaked me out at first but this is low if you know anything this is lower than the mass at which everything should be star-forming according to Marla Gia's work there are some galaxies classical works that we think self question only like two or three but anyway so these things almost certainly have their star formation shut down by reorganization so I make another testable prediction eventually when we find an ultra fade that's not a Milky Way satellite I predict that they will all have uniformly ancient stellar populations and if you're not convinced I worked on a project brief project that never really got published with Oliver Albert where he ran one of my dwarfs ultrafeed dwarfs with no realization and indeed it does continue Farrukh story forming stars to redshift zero and then he went on to do a little more detailed analysis using a milky way I can't remember the mass resolution but you can see this in stellar mass this is the line is that the the halo mass at which 50% of these objects are dark if you change Rihanna's ation from the earliest to the latest allowed by the POC results and it changes by you know like point two decks there anyway so Rihanna's ation is having a big effect on these objects see I told you that was gonna be very short next one is a little bit longer but not much okay um I think I can do it five minutes okay how did doris galaxy teaches about star formation to feedback well the really cool thing to look at is mass metallicity relation this is shown showing some updated do you most spectroscopy of the FE over H versus stellar mass from Evan Kerby his recent work and for satellites and isolated worse and maybe for ultra faints over many orders of magnitude mass there all seem to have a universal mass metallicity relation and strangely enough it's roughly continuous with the overall slope from higher mass now more recently I think I'm not the only one who has questioned Evans decision to neglect this point here and fit a universal profile to it many of us have decided this actually is a flattening and if you add more points on there it actually really does look like a flattening but you know you got to do what you got to do so here the data points again um yeah possible flattening okay I have that here so um I'm gonna put my points on here and really myself and other simulations that are finally pushing down to these this high resolution or for the first time the ones we're putting these objects on the plot and it's awful it's really bad we don't really get it up here not exactly and we're woefully off off course here at the at the low mass end I would like to point out that fire simulations both gas and stellar militant metallicity above 10 to the 7 do match the mzr so this is only a dwarf problem I would I would but I think these things were also affected by Rihanna's ation yeah I mean I think I think I'm gonna point I think the difference is more yes is more of the presence of a milky way but I will yeah I will get into that yes yeah I have I have to but yes exactly yes no no exactly very good yes absolutely um however we do have some isolated objects here so this issue you know and also it's just a few supernovae um we don't have explicit pop3 treatment or this we just have a metallicity floor and these are all at the metal of seafloor basically um our yield yield tables are also very old and for they're not for low metallicity they're for solar metallicity the type to supernovae yield tables are you know we just haven't been able to tie it down for very low metallicity yields so we're using like woolsley and weaver 1995 for solar metallicity so and that might have something to do with it but let's talk about the point everyone made which is the Milky Way yes so this is not a published figure it's a little messy you can this is the same data again Fe over H but what I've done is I ran a very high resolution like a triple latte this is 880 solar mass baryonic particles I ran one to redshift of four and a half and and Andrews actual triple latte had already passed four and a half so they're a little bit lower resolution but I'm going to show where they go on this this is where they start at resolute that's yeah so we're good to show nine eight six anyway okay you can see nine eight seven six five nine eight seven six five this is kind of where they go and it kind of looks good for the ultra fades we're getting a few up there and in fact I also check the gas phase metallicity of a few Laura's even low resolution simulations and about 20% of the IgM is enriched by redshift of is enriched to this level so what is possible however we have a big problem up at higher mass here we're not really gonna match here I'd like to extend this a little bit higher but because eventually we do end up matching I don't not sure how that works okay so it explains the offset for ultra fates but not for massive doors again this problem also occurs we think in alpha at first we thought it was our type one super delay time distribution but it's probably again these type 2 supernova yields um if this is true this kind of testable prediction I'll put a question mark at it isolated ultra fates may have higher at the / H then the satellites we again we have to detect these isolated ultra fates to know for sure and then in my last minute I'd like to some just say that I'm not the only one to find us and in fact I'm only gonna go over two papers this is I think this was seen in maybe even as far back as boville and riccati 2009 something similar to this but I'm going to talk about two recent work Springs at all used nearly as high resolution simulations as we have very high resolution also they ran to redshift 0 with gasoline so similar code they actually match a higher mass so I'm gonna look into a little more about what you know exactly why that might be I haven't investigated this full yet no they actually um when the author's sent me an email and said when they first saw this they assumed it was resolution but then they saw my paper and they realized maybe there's a bigger bigger issue and then even more recently this paper really blew me away hospital s latest work they run one single ultra faint at 10 to the 9 in solar masses and Dark Matter a bunch of different ways and they look at you know they have their food so when they do radio transfer they multiply their supernova energy their high res has 20 solar mass so better Speight better mass and slightly lower spatial resolution but it's a grid-based code they even have radiative transfer it's a completely different code and they you you could see so it's a little the two plots are a little confusing is here the axes are different but you can see my points here are these pale red ones and actually where we match in stellar mass we also match in F V over H except for the rate of transfer they and their rating of transfer runs happen to happen to match much better but everything all of the tracks run right through right you know right through where my data points are so they're still hitting the same mass metallicity relation maybe just a little bit farther evolved so I think this is a problem that's just starting to be discovered and maybe it has a really easy solution so I mean this is all new stuff maybe it's gonna take a little bit longer maybe we need to wait for those low mid olicity type 2 supernova yields I'm not sure if it's probably just a Milky Way but more work on this in the future but it's something we definitely need to take note of okay in conclusion all right using the highest yeah add in another caveat there because of Oscars work a spatial resolution simulations cosmic simulations run duration 0 I'm predicting there's all sorts of ultra faints out in the field hundreds of them they just happen to be including a satellites of Dwarfs I can't tell you the lowest mass for galaxy formation I will continue to work on that even though they're everywhere we can't see them not yet but hopefully with LSST or you know better algorithms yes we can create cores and Dwarfs but maybe tiny core cores in alter fate Dwarfs so and just remember if there's there's already been some claim of cores in ultra faint worse but depending on the size of the core that could be completely consistent with lambda-cdm um dwarfs are ancient independently of whether there are satellites or isolated ultra faint works and yeah this there's this issue with the mass metallicity relation it's probably a massive neighbor for the ultra pace maybe both maybe type 2 yields we're going to look into cluding an explicit treatment of population three star formation instead of just a metallicity floor maybe that'll figure something and then yeah just works are awesome lambda-cdm is fine but we should keep challenging it it's very important to challenge it we're gonna kill it or make it stronger and it's time to start thinking about how to build bridges computationally achievable bridges between just have a higher and higher particle size to actually forming stars because at some point I mean if we have a solar mass star particle what do you know what might as well have a real star but we need bridges between that in the meantime thank you know it's yeah so they they're the yeah I could show that the tiny ones the tiny ones the really tiny ones all form their stars by redshift of 10 the really tiny ones and the the more massive ultra fates form their stars by redshift of 2 so they have a little bit of salt there's like a sub grid self shielding aprox I didn't need to unplug it yeah so you can see here's colour coded but yeah so that's what I was showing here these things all form before a reorganization really kicks on and then the higher map everything is affected you can see pretty much everything is rolling over maybe not this one most of them are being affected by Rihanna's ation turning on but these things have some gas left and keep chugging along but then they sort of they don't create any more fresh gas likely we don't have any of those yet oh yeah I guess I'm think I'm sorry I'm thinking of like tiny ultra face but yeah you're right it's more like more massive though yeah yeah but you're right ultra faint is technically up to 10 to the 5 I just always I see such a differences like even right here so sometimes when I I need to get I told you about the nomenclature and then I'm messing it up so yes the the for the tiniest ones we don't have any satellites but yeah yeah they don't yeah these the ultra are my Ultra faced don't have any gas the higher mass ones do but not the tiny ones what do you mean I don't know yeah I don't know I don't think not with my son right yeah yeah right yeah they're getting it right yeah yeah it was something I thought of because of Fras paper and also just a lot of things that's come up I really need to go back and run all the standard fire tests that were run back in the day because back then we weren't even really forming ultra faint so we have these have been tested on Dwarfs maybe one ultra faint like one higher mass ultra faint but I haven't looked at different prescriptions I've just been using so far the standard fire well I saw that what a Gertz at all did and I've been wanting to do that just one ultra faint and then just changing changing everything but I haven't done that yet but I imagine I imagine a lot would change the interesting thing about Dwarfs is when you change things I mean what changes and what stays the same because even just sometimes a little bit of stochasticity can you know even just running on a different machine okay it can unfortunately change the change things enough but that's why I'm going to run more of these you need a sort of to get a statistical sense and like and ask yourself what's robust like you they're always uniformly ancient as long as I don't keep Rihanna's ational that's one thing that doesn't change but yeah that's something I need to do we do at medal mixing now yeah yeah I think that I don't know if I'll be honest I don't know if I check this for the previous runs where we before we did metal mixing but that would be something I could easily look at easily check for me good comparison because I didn't look at metallicity in my 20 2015 hey per yeah back then before we had mixing we just actually we had a huge spike in the middle of City distribution function at the floor but I don't know what the average would have been they said we said we matched it back then but again I think we were averaging the abundances before taking the log forth instead of doing it the way observers do which is taking a fee over H of each star and then averaging once we did that that's when we notice the initial discrepancy but it's after check that yeah so I mean I think that's it is true that maybe the first order issues our galaxy formation issues but I think again with large enough sample size and statistically you can start to get control on those things and if you hold all else equal and change the Dark Matter you also see differences so I think it's dwarf galaxies or ultra feints are really good for ruling out models so maybe I can't say this is specifically it but you can try you can test the Dark Matter model on it and use dwarf galaxies I would say to rule out a class and models and to maybe narrow in so maybe you always have to make you know there's always some uncertainty with with your galaxy formation and to be honest your start we're really learning about star formation that's really the biggest uncertainty but through a combination of running you know a sweet holding things fixed all fixed but the dark matter I think you can rule out some parameter space and in terms of dark matter and as eventually as we go on figuring out galaxy formation better then we'll have even more constraining power yeah they pretty good yeah that's a really good question and I should check that but I would imagine they all form before realization that would be yeah where's that here it is yeah I haven't maybe a long time ago I looked at something like this for lower resolution but I don't remember off the top my head but yeah I mean if it's just a single star particle like howdy it could just be passing through could be there's I mean these things have stellar halos too especially for a satellite but so I wouldn't trust it but even yeah I would expect all of these to be uniformly agent but I'll check yeah I mean well it's thing is we already know the Milky Way is shredding up dwarf galaxies in the streams but what the what the what the you and you're talking about is even like empty Dark Matter halos right that the sub Hale is themselves are getting destroyed so I don't know I don't think we would be able to detect that but we do some we do see streams in with Gaia like according to the conference then what I've seen there's none more so that is that is happening yeah we I mean we have to at some point factor that into our missing satellites problem probably [Applause]