Harvard Heidelberg Star Formation Meeting - 11/12/19 AM

[Music] [Music] okay hi hi everyone thanks for coming so this is the sixth annual of hardware title board meeting and this year we're going to focus on linking observations and simulations those of you don't know my friend Thomas here took me out for a tour our coffee in Heidelberg like seven years ago and said we shouldn't have these meetings and here we are so if you enjoy yourself you should thank Thomas and so my job swing and then saying hi thank you very much I owe one very much to everybody that was on the organising committee here could you please stand up so everybody could see where you are well Viale Mike Foley and Kathryn doctor who's going to keep you on time too so watch out so they can also help you and as Kim Jody Blackwell who did a huge amount of organizing for this she's in the back to seem to get herself a nametag later but did you remember Jody knows everything if you need something and so you've all seen the website for the conference and the one thing that it falls to me to explain before we move on is the unconferences and just to show you where that is this is the schedule for the meeting and for those of you who've been merging have you've never been to Harvard Heidelberg meeting before about 10 people okay so talk to the people who are with you that's gonna be the best way to understand what's going on okay but basically like at least half time or so is a little bit unscheduled as of now because we have these things club unconferences because a lot of us works on topics of common interests or tangentially interests and so we'd like to just talk with each other in ways that are not planned in advance so that's what these on since our and in order to facilitate that there are these introductory talks where everybody will talk for two minutes to just kind of say what they're working on these days and unfortunately can't do them all today or we would spend all day doing that so there's a schedule and I'm sure you've all found your name on there and if you haven't uploaded a Google slide or made a Google slide of what you want to show you should do that your many emails explaining how to do that but that's how that works and then we have these framing talks which you can read about it the highlighted research talks and then like I said it's my job to tell you about the unconference talks if you want to and I'm guessing probably no one has there you go there's a Google Doc it just says glue tutorial for what the conference's can be but in fact the way that we will schedule them is there are big white pieces of paper with markers that look kind of like this on the other side of the street where we're going so part of the meeting is going to be here in this room and then the smaller room is where we're gonna have these unconferences and also coffee and stuff like that or in another building called fort timer house on the other side of Garden Street and again just walk with people from here and we'll show you where that is okay and so this horrible-looking document shows you what is going to happen at these unconferences and so all I need to really tell you is that each one has a chair so when you've proposed something so you're gonna write down on these blank sheets here an idea that's like I want to talk about W 49 because it's my favorite region or I want to talk about I don't know unicorns in space okay like whatever you want to talk about you write that down and then other people during a break are gonna put little tick marks if they're also interested in that and then the one people are interested in will get a room during the next unconference session and if yours doesn't get chosen for that session we can schedule it for a later session so in others there's just think there's three rooms for our conference we can have four at a time but there's four sessions five sessions or something about conferences so there's a lot of time to do these and again the experts who have done on this all before just ask one of them if you're not sure what to do but the important thing is that there's a chair who is the person who proposed the topic and your job as the chair is not to say too much but it's also to make sure that no one else says too much so it's supposed to be a conversation and if you want to show slides or images or movies or whatever there's instructions here about a Google Drive folder where you can go whoa 20 URL is yours we just try these 20 minutes you know it's that light bulb all right we'll click the real link will change this so that it's not a tiny URL but anyway there's a folder on Google Drive anyway where you can put any slide or image that you want to show and on top of them in that way people have to unplug and plug in their computer there are computers in each room computer foot pops major okay and then somebody the chair should designate somebody as a recorder somebody should volunteer to take some notes and in that same folder where people upload images and whatever just take some notes in a Google Doc even it's just the name of the unconference if you don't want any of those okay and there is an exception to not talking too much which is that if somebody has a software package that they want to talk about or something else that would get a tutorial you can propose to have explicitly a tutorial and then people should know what they're getting into learn about myself and if there's like related software packages I think it would be fine to find a pal and pass that when the 3-legged to yours and propose one that's a twofer would be great so any questions about the unconferences no that sounds it sort of clear okay and something do I turn it over to the thumbs and then what webpage do I want to be this computer on the Google slides but now all the you go to that place yeah I've there's a bit long link on the home okay so get it home open your home yeah okay and then go to zoo blue slice Indonesia not very convincing yep okay for each song right away yeah okay good morning everybody and my job is to tell you about some work that I've been doing with Ian Stephens son Connor Basu assigned Town Oddie and several other folks and the idea is to try and learn more about magnetic fields in star-forming regions we know magnetic fields are important for many aspects of star formation listed here and so the goal is to be able to get better estimates of the strength and the structure of magnetic fields in star-forming regions present techniques for extracting field strength they all have their limitations so we have our taking two new ways to manage the old properties and start forming regions and they are based on two very old ideas one is the idea of flux freezing which was pioneered by Leon Messel many years ago and the idea is probably familiar to many of you you have a initially relatively uniform medium threaded by a uniform field it is kind of a dense and compress and the field is dragged in because it was a little bit weaker than the forces of doing the compression it gives you an hourglass shape to the field lines and also to the polarization and the field strength scales roughly as the disturbance power of the mean density which has been verified in a number demon studies summarized by Romina paper Isaac question why what is our what makes our contribution new it is that we have gone from the spherical case which was not a lot of fun SL to the spheroid case and that means that we can apply this to regions which look relatively elongated or flattened or or non syrup nearly all preserved regions depart from beautiful symmetry of circle symmetry and when you analyze the problem in the Sphero okay such as a yeah people feel our thyroid then you have to take into account the elongation and the inclination the aspect ratio and so we have now done all that and we are able to do this estimation of field strength case up steroidal objects with flux reason the second old concept is that out pain waves on magnetic field lines are linked to the velocity dispersion of the observers retro lines and also the mean density and dispersion of polarization the gender say our main technique for estimating field strengths from polarization observations and what's new and I'm going to show you is that if you assume a certain level mean level of field fluctuation then that enables you to get the mean field strength from the non several particle whywould and the mean it's budding and this has become especially attractive now that we have so many beautiful maps if ammonia lines which allow us to separate okay so just to illustrate how this might work here's an application to a well-known star form of corn to produce our binary bhr 71 here is the ammonia map made it parks on the 105 parsec scale and estimating the field strength and structure is as easy as one two three here are the one two three yes based first on the turbulence and then secondly on flux freezing and that allows you to get the run of field strength with position and also the field Direction map so here is the Rona field strength running from about a minimum of about 35 micro Gauss in the background level up to a peak of about half a little goes and this is a scanning of water two directions to particular directions since the region is not spirit the final point is that you can use it to evaluate the mass flux ratio we get a value of 2 which means that the star for gravitational collapse we'd now be allowed and that is reassuring since we have already formed two stars in this region so this application can be applied to now [Applause] you see have the vision towards a massive crime scene digital age to investigators a little these are many studies of Mustafa machine by analyzing massive discourse in counts everyone before estimated - to extremes and study a mixed state of happy operations you'll die maybe supports parameters are still much less still reserved which signatures are more dynamic message in my equation in comes a toxin makes bimodal distribution suggesting that magnet close to the stage besides I only started boxing [Applause] I'm walking after three three four eight years now and rendering material about and so today I want to highlight make sure it's something so sorry about that we thought everybody read about this here place identify clear back we're talking so yes so today I wanted to highlight in particular the result of our latest paper or accepted in object where we show the importance of the silver Oh elemental ratio in the chemistry of periods and time forming disks so periods are the acronym for photo doing iterations which are found wherever there is an interface between these shoes and dense gas in the interstellar medium and a famous one is the awesome nebula which have been observed a lot and in which the chemistry Kia was also deeply explored more than 30 species have been detected and in particular those three years I nights I'm showing and so we focused analogy in on those three molecules because they are also observed in disk and before our studio it was not very well understood I mean in particular the abundance couldn't be reproduced so what I've been doing I further develop the PD I'm a drunkard in which I extended the chemistry to these three molecules and explored different parameter on the chemistry so here I'm showing house evidences of these three molecules varying as function of the visual extinction so within the PD are four different overall ratio vying from point four to one point five and well the - boxes represent the observation within the arrow bars and what we show is that when you vary the overall ratio by less than a factor of four there is more than several order of magnitude change in the evidences of these three cyanide and in particular we need each other or ratio of 0.9 so which is hired as a solid value which is 0.5 try to reproduce the abundances of the three molecules so we wanted to test our findings in discs because the out I was earlier of this can be compared as a photo doing iterations so for that I've been using discs that we already published earlier this year in another up J paper and so I took this model and I've focused in particular on the chemistry of Swiss Ian and X his friends so which are both observed in several discs and we show that indeed for the overall ratio an elevated to a ratio of 1 we can better reproduce the observations that I've been up cell so this is an interesting result because it's also explained why you will have more species like safe Swiss Ian result I'm ch3oh in disks because of the three overall ratio and it's also in agreement with other results that are currently found in disc so if you want more detail you can come talk to me or you can look at the paper which is on archive it'll be in there somewhere alright the first thing I'm gonna just kind of mention because it's the main thing so I'm Ian Stevens I'm a postdoctoral fellow on the Smithsonian side and so the main thing I've been working on here is the master survey which is a long CM zoom it's a largest survey ever done by the SMA and it's a survey it's complete of all class 0 and class 1 protostars every single one in the Percy's molecular cloud and we look at size scales from 1 to 28 about 1 to 20 arc seconds what we're sensitive to with SMA and that at this is the only complete survey at those size scales of molecular lines and continuum and we can see you got a lot of beautiful special lines where you see on the left is co2 - one outflow yeah co2 to one and you see these really textbook outflows that we're mapping many that have not been mapped before and we already have 10 published papers with the survey and the data is publicly available what the full catalogs just accepted and then I'm also going to be presenting a poster and it's gonna be a lot of polarization so here are some beautiful polarization images have been taken with Sophia snaky 18.6 this is most of NGC 1333 here's some stuff in Orion B NGC 2060 18 20 71 I've also been trying to do Polar's line polarization of discs and apparently it's very very tough i I don't see any vectors here but these are two discs have been looking at and it seems like better let the GK effect line polarization so that's one less than one percent polarized for these discs which was kind of surprising and then I also talked if you want to talk with Phil or in me about the spherical flux freezing model he's doing we're both working on that here are some things that we're doing when we're fitting a polarization morphologies thanks should I do it this one oh sorry because me thank you [Applause] okay okay right um let's start by thanking the organizers for inviting the G of the talk about chemistry and why it's interesting and why it's difficult so the item is actually chemically a really interesting place even though by terrestrial standards it's a really harsh environment it's it's close to a vacuum terrestrially there's all these high-energy cosmic rays lots of UV photons despite this it's a really chemically rich place so this is just a actual subset because I couldn't fit all the species on the slide of all of the chemical species they've actually been detected in the is M now a lot of these are in circumstellar gas but there's still quite a few of these have also been detected in interstellar clouds so that's really a lot of complexity and we'd like to model that and they're really historically there have been two main reasons put forward for why we might be interested in modeling the chemical composition in stellar gas clouds one is that we know molecules are really good low temperature coolers in particular things like CO and so if we want to get the thermal state of the gas correct we might be interested in in its molecular composition and the other thing is that there's a wealth of data on the physical conditions in the interstellar gas that's encoded in the molecular lines and in the chemical composition and if we want to actually interpret this we actually need to be able to model the chemical composition now I mean this has been done for a long long time biochemists in simplified models starting with kind of one's own models and moving towards kind of 1d static models or even 1d evolving models but really we'd like to do this in in full 3d simulations where you have all of the messy hydrodynamics and this turns out to be quite hard so why why is this hard it's basically because the is M is so chemically rich we've got hundreds of species thousands of reactions so this is an example from the the latest release of the u.s. database faster chemists that's got there almost 500 species in it linked by a border 6000 gas phase reactions and if we're interested in things like Deuter Asian or grain surface chemistry then these numbers just go up even further and so if we really want to comprehensively model all of the chemistry in the is M we have to handle all of this and this is a problem now each of these individual species has a partial density if we add up all these partial densities we get the total gas density and the dense partial density of each of these species has a governing equation which you can write like this so these two terms on the left hand side that's just the usual continuity equation basically describing how the partial density of this particular species is getting add vex it around then we have this term which is representing diffusion this is basically microscopic diffusion if we if we have a chemical in homogeneity even if the gas is not moving around the microscopically it will mix and then we have these terms on the right hand side one representing the creation of new instances of the species so say this is a molecule we're interested in this the formation of the molecule and this term which is the destruction of the molecule now in the is M we have one simplification we can usually get away with making which is to ignore the diffusion term that's because generally the advection term is much more important and it's like the turbulent diffusion is lot much more important than molecular diffusion so we can throw that away but I still leaves us with a set of advection reaction equations one for each the species and now what we generally do in order to solve these we do something called operator splitting so rather than solving this complicated partial differential equation we break it into two things we break it into a normal partial differential equation with zero on the right hand side which just describes the advection of our species and then we have a separate ordinary differential equation which describes the chemistry at the reaction term okay so this just first although de those are easy to solve sorts the problem the problem is that the lots of different processes basically encapsulated in these terms that have a very wide range of different timescales so chemistry occurs on a wide range of different timescales simultaneously you can think of something like h2 formation in the in the is M the h2 photodissociation timescale the unshielded bits the I am is very short the actual formation time is very long so you've got an order of magnitude different in the clock in the characteristic timescales and that means that these ordinary differential equations are what we call stiff now stiff is just a shorthand way of saying they span a wide range of timescales but this presents us with the problem and to illustrate the problem its it's convenient to think of a very much reduced dimensionality version of the problem where we basically have two different time scales are short one than the long one so suppose we have our system that starts in some chemical state it's got not going to start in any kind of equilibrium but any fast reaction so the short timescale reactions are very quickly going to come to equilibrium so the actual chemical evolution is mostly going to occur on some reduced dimensionality surface in our chemical state space so if it's hard to draw a high dimensional surfaces I've just drawn a line and said so basically if we start at some point up here the rapidly occurring processes quickly take us to this line and then the slow reactions move us along this line okay so what is the problem the problem is if we try and solve the OD describing this behavior explicitly so basically calculate the time derivatives at this point in time and if we're not on this slow manifold if we're sitting off of it then we're going to get a very large gradient because we'll pick up the influence of these fast reactions so to accurately follow the evolution we have to take very small time steps to make sure that we reach this slow manifold and don't massively overshoot it so that's fine but every time we go to a new hydro dynamically in our simulation we're going to get displaced away from this low manifold this is this is inevitable we have truncation errors we have roundoff errors we have advection effects the log they're all going to introduce some error that's going to move this away from this equilibrium state and put us in a point where the fast reaction is not in equilibrium so every time we try and solve the IDE we're going to have to take these really tiny time steps which means we wind up taking large amount like enormous amounts of time to solve these Odie's and actually just hold the boring bits the the bits that we don't care about because we know actually the the system is going to be in chemically we equilibrium as far as these fast reactions are concerned so this can be just like her rip equally expensive computationally so the usual ways to get around this is to solve them implicitly so calculate our time derivatives at the end of the time step rather than start at the time step this is a well-established technique for solving stiff ordinary differential equations the problem is it's an iterative technique and you have to do a matrix inversion and that is also costly that has a cost that scales as the cube of the number of reactions you do not reaction species so if you're just dealing with a small handful of species that's fine if you're dealing with five hundred species that's also horrific so the consequence is that the full chemical network is very costly to solve and while you can do it fine in kind of 1d approximations to do so in a 3d simulation where you have millions or tens of millions or hundreds of millions of resolution elements is just not practical so so what do we do well one very so technique we can use to get around this is using what are called reduced networks mmm the idea here is pretty simple if we're just interested in something like ice and cooling we know that most of these of order 500 species are not important if we go back to a list at the start the abundances of many of these things are really small they're not going to contribute in any meaningful way to the thermal thermal balance of the iossef so we only are actually interested from the from a thermal point of view in the advances of a fairly small number of species like C+ atomic oxygen Co h2 a few others I am bother to list here so we can make a really big computational saving if we design our network to just follow the chemistry of these species and not all of the other complex chemistry so this simplified chemical network the the jargon is this is a reduced network and if we can reduce say the number of species we have to track from 500 down to order 15 or 20 remember our cost is going as the species cubed this is like factors of a thousand saving in computational time so this has been exported for a while in the literature there's a number of different examples designed to follow specifically Co chemistry because that's one of the the simpler but more important things um I can I can go into boring detail about this if anybody's interested but I'll save that for the coffee break but the current state of the art there's two kind of networks I would say are state of the art so there's one which if you are interested in speed and don't care much about accuracy there's a network that's Paul Clarke and I put together which basically is a simple description of the hydrogen chemistry plus a a single reaction approximation for Co formation and destruction which is based on the work of Nelson and Langer now this is really fast you you wind up with having basically three reaction equations you have to follow and it's hard to do better than that if you - mobile HT unco it's not very accurate but it's a good choice if you're doing a large-scale simulation and you don't have much resolution in your individual Michael clouds it will give you the right behavior in the limits of low densities and low column densities and high volume in column densities it doesn't do very well in between but if you just want to know is this cloud molecular or atomic is it's a good choice so there's a lot of work already been done using this if you do have enough resolution to actually resolve cloud structure and you want to do a better job of modeling the transition from atomic to molecular gas we can do a lot better there's a few different possibilities in the literature but I think the current state of the art is this network that moon and Gong do published last year which is based on earlier work by again by Nelson and I know that moon and has has improved so I've just got a couple of illustrations of what you can do with these things so this is some work done with the Nelson Langer 97 really simple network so this is a model of the central molecular zone of our galaxy so this is the bar this is gas following along the bar and you see this ring develops in the middle but specifically what let's try that again specifically what we're looking at here is there is a synthetic image of the of the c+ fine structure line surface brightness so we're following where the molecular gas is where the ionized carbon is and looking at what you would see if you were looking from outside our galaxy looking down on the on the central bar and the ring okay and this is a block from my Nana's paper a very complicated plot of a range of different species basically her results of the solid lines and the dashed lines are what you get from aquifers PD our code which is got a much more complex chemical model in it and what you're supposed to take away from these is that there's a good agreement between the solid and the dashed lines in most cases there are a few things like atomic carbon where there's a factory a few disagreement but on the whole reduced Network does pretty well for a much smaller cost than in in the for PDR Network so there's various things you can do with this now one of the one of the first things that we did was to look at the actual role that molecular cooling plays in star formation because as I said like historically people always said Oh molecules are really important because you need molecular cooling to form stars yeah turns out that's not true so this is from a paper we published back in 2012 where we basically ran a series of simulations of the same turbulent molecular cloud just successively turning off bits of the chemistry so you go from the full model with different initial conditions to one that's just tracking h2 formation and destruction to one that's just assuming all of the gases atomic and in the end they all form stars there's not a huge difference in the ability of these clouds to form stars depending on whether their molecular or atomic we've also seen a similar result in simulations of dwarf galaxies so this is some consumed work by HRU who from a few years ago looking at a dwarf galaxy with a metallicity of 1/10 solar and what he did here was very various parameters like strength of the interstellar radiation field of the dusty gas ratio so this is these three colored lines at the top and that makes a huge difference to how much molecular gas you have form in the galaxy this is the the HD mass fraction which is varying by more than order of magnitude as you play around with these parameters but when you look at the actual star formation rate it doesn't really make any difference the only thing that makes any difference to the star formation rate is turning off the feedback which then gives you this green line so yeah turns out while molecules are important in actual molecular clouds if they weren't made form stars anyway with pretty similar properties so from one point of view this is this is good this means that we can probably trust a lot of the simulations of large-scale star formation in galaxies that have been done without molecular chemistry they're not going to be making huge errors from another point of view then it does lead one to think why did we and so much time trying to get this to work in first place um well fortunately there is another reason why the chemistry is interesting which is using the chemistry as a tracer of the physical conditions and fortunately a lot of the things that we're interested in because they play a role in the thermodynamics of the gas are also interesting chemical traces so Co again is the is the obvious one we started trying to model Co because we were interested in it as a coolant but we then get its distribution as a tracer for free so some of these things we can treat nicely with these reduced networks so we've done a lot of work on C + C Co because we have these in the reduced networks but then there's a lot of other interesting tracer species which are not included in the current reduced networks but we still like to know about and this is I mean good examples of this are the classic high density traces so h HC n CN n 2 h + ch3oh etc etc and to model these to model the full range of interesting traces where we could try and come up with more complicated reduce networks but this is a fairly time-consuming process and it's less of a saving the more chemistry you have to follow so generally the techniques that people use here are more reliant on on post-processing so you run some simulation and then you try and do the chemistry afterwards now you can't do this if the chemistry is important for the thermodynamics because obviously if you do it afterwards you can't account for its influence on the on the actual thermal evolution of the gas but for these things that are not important thermodynamically but are just interesting as tracers there's no problem in just trying to trying to paint them on later so there's a couple of main approaches that people have used here so one approach the probably the simplest approach is to just assume the chemical equilibrium so take a snapshot from your simulate with some complicated density structure and just solve for the equilibrium chemical abundances now there's two different ways of doing this you can you can take the temperature directly from the hydro simulations and keep that fixed and then just solve for the chemical chemistry or you can solve self consistently for the temperature in chemistry in this post processing step and this that latter approach II is what you do if your actual original height hydro simulation was done with an isothermal approximation for instance now the main complication here is that generally we need to worry about the attenuation of the radiation field so generally not just solving the chemistry which would be fairly straightforward but you're also solving for the radiation field at the same time and if you need to account for things like h2 self shielding this actually becomes an iterative process you have to solve for the radiation field and then solve the chemistry and then the chemical states going to modify the radiation field so you go backwards and forwards until you wind up with some equilibrium solution there's a couple of examples of this so one which I'm going to show a result from in a minute is Thomas Bayes vs. 3d PDR code and there's another one turbo can that is going to talk about later this week so I'm not going to say anything about that but this is what you get from turbo chem so this is from a paper by Brent gachy's from a few years ago basically what he's done here is taken a simulation done by cell aradhna which used just isothermal turbulence and post processed it with 3d PDR to produce maps of the different chemical species shown here so you see what you see in a tracer like CEO is actually very different from what you'd see in a higher density tracer like into H+ so this is fine but you have to assume chemical equilibrium so what what if that's not true what if your species are not in equilibrium and we know that some things are probably not going to be some things will go quickly to equilibrium some things like h2 get very slowly to equilibrium so the other approach that tries to capture these time-dependent affair basically uses what are called Lagrangian traces so these are basically just traces that go along with the flow and allow us to record the state of a particular fluid element as a function of time so it's density its temperature and also the radiation field it's saying just as a function of time as that fluid element evolves like we store all that information then we can later run that through our chemical evolution code to figure out what the chemical history of that fluid element was now this has the big advantage that there's no need to assume equilibrium but it has a disadvantage in that generally when we put our phrases in at the start of the simulation we don't know where they're going to end up so there's always the danger you're not going to actually sample all the regions you're interested in you're not gonna sample it well enough and that's something that we we're still kind of struggling with figuring out the best ways to do this so this is an example of a basic basically how you do this so this is from a paper by Azusa Tower currently still in prep so you start with your hydro simulation you select some Reed regions you post-process them with a chemical code in his case he was using edema terminals are chemical and then you stick them on the grid and then you can run radiative transfer on that grid to make emission Maps if you want so this is an example of what you can get so this is the the underlying hydro simulation which is an S pH simulation of a tenth of the four solar mass cloud he's zoomed in on one particular region with a few different cores and put in traces to sample this region and if you do that and then look at the the later chemical State using these tracer derived chemistry you can see we can do a fairly good job of following the evolution of different species and again we see that like your classic high density tracers are really just concentrated in the cause Co is much more extended there's lots of information we can potentially extract from this so we're not the only group that's been doing this there's also paper from John Delaney from a couple of years ago they've done something very similar in that case they were looking at disk chemistry so this is some protoplanetary disk with several fragments formed in it which for some reason he's called John George Ringo and because this is that SPH simulation the SPH particles are actually Lagrangian traces themselves so what john Jonglei did was basically take every single SPH particle look at its history and run it through a chemical model so there's water a couple of million as speech particles in the simulation and apparently if you just kind of use every single node on the cluster for a week you can post-process all of the chemical history of each every single particle and make these detailed predictions for the for the chemical makeup of the disk okay um I have no idea how much time I have left sir oh I've got faster than I thought okay so then I've got time to just show off a few examples of what we can do with models of the chemistry of the is M so these these these are a fairly bias selection of work that we've done and I can talk about them more length in in the coffee break or in the discussion afterwards so this is the first one there's been a question for a long time if molecular clouds form in converging flows of atomic gas how do we actually trace this process the molecules by definition form in the cloud once it's assembled we're not going to be able to trace the inflow by looking at Co we might have some hope of seeing sub signature at ih1 but each one is tricky because it tends to have a large wasti width so it all gets very confusing so people suggested well C plus C plus should be a good tracer of this lower density in flowing material so what does this actually work in detail so this is something that welcome I looked at in a paper we published this year and so these are position velocity diagrams of basically a colliding cloud expand we have two clouds with turbulence we run them into each other and then we make synthetic emission maps to see where the emissions coming from and so we see for the CEO or the to see one fine structure lines this emission is all coming from a pretty narrow range of velocities this is basically the molecular cloud itself after that's formed in the collision whereas if we look at c-plus we see this much broader range of velocities so we are actually successfully tracing the inflow the problem is if you look carefully in particularly if you look carefully this scale yes there's lots of emission out here it's pretty faint its brightness temperatures of tens of million so this turns out yeah that it is possible to trace this material in c-plus but it's really really hard to do that so our original hopes that you'd be able to just kind of point severe at some patch of the sky and map all of this doesn't really work it's too faint to map what you can do is just point at one Patchen and you can get detection so nikola Schneider's got some some nice not yet published c-plus detection 'he's just looking at individual points near kind of high latitude molecular clouds a second example so this is this is now actual data not synthetic data so this is position velocity diagrams of the center of the Milky Way mostly looking at Co this stuff is ignore that and it's been understood for a long time that there are these odd so called extended velocity features where at some particular collective longitude you see these very broad vertical feature in the PV diagram so there's a very wide range of different velocities at one longitude and what I'm not showing here is these are actually fairly spatially coherent as well if you look at longitude/latitude they tend to just sit in one in one space they're not extended on the sky so for a long time there's been discussion about what what are these things it's even being suggested that these might be intermediate-mass black holes falling into the center of the galaxy and the the large velocity which you're seeing here is the gas spiraling around the black hole we have a different picture so this is a basically a fairly complicated plot from one of these galactic centers simulations I showed you earlier so this is the bar the colored bits here are gas streaming along the bar and winding up in the center and if you look at this in again in a PV diagram so it can actually long this new velocity we see that we also recover these extended velocity features quite naturally in the simulation but what we can then do in the simulation is say okay well right we see this in PV where is this in actual real space and what we find is that these bits where you have these extremely large extending velocity these are actually out in the bar where individual streams of gas are running into each other so they're nothing to do with the actual CMZ itself they're there basically what happens is some of the gas that falls in along the bar misses the CMG overshoots instead of running into the center it misses it comes around the other side and it runs into the gas that's falling down the bar from the other side of the galaxy and you have a big collision these collisions are spatially coherent and you naturally get these very extended velocity features all right and finally this is suspending some work by going Jones there's a student in Cardiff with Paul Clarke so what we've been doing here is looking at HCN and trying to figure out where does the HD a mission in a dense cloud actually come across so this is our emission map that of HCN in the cloud no it's not this is yeah so this is the total emission in covering kilometer per second you see this complex structure so what go in is then done is basically done that succession of different radius of transfer models where we basically only include gas above some density threshold and as you've changed the density shoulder you can ask when does that emission map start looking like this total emission so if you just include the gas above the number density of 10 to the 4 so this is classically what people have assumed HTN is tracing here you get this this does not look very much like this if you go down to 10 to the 3 then you get much better agreement so in this particular cloud the conclusion that we come to is that actually the HC n emission is mostly coming from this range of densities between 10 to the 3 in 10 to the 4 so the densest is tracing are quite a bit lower than people have previously anticipated and certainly what does a magnitude lower than the HC and critical density which is about a 10 to the 6 and while I don't have a plot of this this seems to be backed up by some observational work we've been doing together with the NS Kaufman and a large crowd of other people so this is the the Lego survey of 3 millimeter lines in gas clouds in the Milky Way so we have some work we're just writing up right now on W 49 and that again is showing us that if you map HT n over the whole of the cloud rather than just in the cause you actually see a mission from all over the cloud which is consistent with it coming from much lower densities than has been previously supposed so again somewhere in this range of 10 to the 3 to 10 to the 4 ok so to sum up modeling the chemical composition of the is M is important for interpreting molecular line observations it turns out not to be so important as we thought for actually getting the thermal stage of the gas right it's the challenging computational problem in 3d for reasons I've tried to explain but this is something that we are actually successfully dealing with now at least as far as the the simple carbon bearing species are concerned and we're starting to get to grips with some of these more complicated high-density traces species but I have to say that while we've got some nice preliminary results there this is an area which is still largely unexplored there's really a lot still to be done looking at these these lower abundance high-density traces and other interesting traces and I think what here's some more of that from yeah then there may be some some effects here but we can't tell with the simulation because this the these tracks are fairly well resolved if we go to higher mass resolution this doesn't change we go to higher mass resolution this changes a lot so the yet we're not resolving all of the fragmentation here so it's dangerous to to over-interpret this we might think the influence that the molecular cooling might have on the IMF is still not entirely pinned down but but I mean something to bear in mind is the molecules are only important over like maybe one and a half or words of magnitude in density so if most of your fragmentation is happening later on when dust is dominating then the molecule is not going to be in port but yeah I mean there's still there's still an element of doubt there I think from a simulation point of view as far as the as far as the star formation rate is concerned the molecular cooling is is not important or if it's important it's not the main thing that we need to worry about the balance of feedback and cloud formation is much more important than the chemical state of what iCloud is made of the well the one advantage that you get from doing it in real time is then you do have it everywhere whereas with traces then you might wind up that you don't have traces in the regions you're interested in so if you're actually interested in in the chemical composition everywhere for it then yeah but I mean that if you look interesting something like Co then that's actually potentially quite a large volume you simulation you want to know what the chemistry is doing so there are our advantages to actually doing it in real time but yeah for a lot of things I think you might as well just post-processor so all of the work that I've shown from from our group is doing time-dependent chemistry okay then I mean this is just looking at one particular output time from this cloud I think probably shortly before star formation has started so I mean this is this is still working progress so an obvious next thing to do is then look at different points in the history of the cloud so yeah strictly speaking we don't know this yet we know at this particular point in the history the cloud this is true no because I mean if you're doing a non-equilibrium model then at this particular time you're not going to be in equilibrium and you can't end you yeah I would have to ask yeah yeah I mean I I think one of one of the big things that somebody still needs to do is really compare what we get from these time dependent models with what you get from just to see me equilibrium and get a better sense of like when when is equilibrium just going to be a fine approximation and when is it not and we don't really have that sense yet but I mean I think with with the different attempts now being done to this in 3d that's that's gonna come soon I mean I think you're gonna say more about that later on yeah but everything else depends on the h2 so if the h2 is not in equilibrium then so yeah I don't have a plot in this particular talk but if you look at what's dominating the cooling balance of of a molecular cloud a low density it's dominated by C+ cooling below about a thousand particles per cc between about a thousand and a few times 10 to the 4 maybe 10 to the 5 probably depends a bit on on exactly where you're looking in the cloud you dominated by the molecules specifically CO and above 10 to the 5 you are dominated by dust you you effectively couple to the dust and then the gas temperatures just tracks the dust temperature now all of those numbers are true for radiation fields close to the usual drain field and for solar metallicity if you reduce the metallicity then you're going to change where those transitions occur but for a kind of typical nearby molecular clouds yeah your your Co is only dominating the cooling between number that's a 10 to the 3 and 10 to the 5 which is potentially important because that's the that's the density of issue form the pre stellar cause well no I mean the the statement I made was basically if you didn't have the molecules they're not much you change the the the big difference if you don't have Co if you just have assumed all of your carmen is there a c-plus is then instead of cooling down to order ten Kelvin you call downs 20 Kelvin and that's basically it you can get almost as cold it's not quite as cold so it will change the structure a bit but I mean you're still dominated by non thermal motions and really whether whether your minimum temperature is 10 or 20 that turns out to make very little difference from the point of view of star formation okay yes yeah yeah generally the dust is done fairly simply so I mean in in our models we just have a fixed um kind of dust composition and size distribution that just calibrated by local is M and then when we look at different metallicity 's we just scale that up and down and don't actually change its nature there are efforts to do that more self consistently and actually look at how the dust might change as a function of metal st or environment but that gets quite complicated but yes I mean you have to include it the dust is is vital for getting the chemical structure right if you don't have the dust shielding you're not going to get the right chemical structure in these clouds and if you don't have the H deformation on dust grains you'll you could just gonna stay atomic yeah and I mean for I think understanding star formation I mean one thing to take away it's like above number density 10 to the 5 at so the metallicity dust is the dominant coolant so like over most of the orders of magnitude of density as we go from from the low density is m2 forming stars it's dust dust is the thing that's doing your cooling it's not the molecules the molecules are really a transient phase they're interesting because we see them but they're not in the end all that important yeah so chrome is basically is a it's a package with a whole bunch of different chemical networks so they have a variety of different reduce networks I'm not sure if they have a version of the gong Network yet but I mean those are basically largely taken from from the existing literature and they've just put them all together in a nice simple package so I mean chrome is a really useful package if you're interested in putting chemistry into your simulation you don't already have an implementation because everything is there in one package but it's not really an alternative way of doing things I mean it's just a nice package for doing the things I've been talking about yeah so we've looked at this a bit and I mean the particular issue of freeze-out and co cooling as was looked at extensively by Paul Goldsmith about 20 years ago it turns out the freeze-out really doesn't change your thermal balance at all it's because Co so optically thick you can freeze out 99% of it it's still optically thick so you still get the same amount cooling and so Paul's got a really nice paper showing this where you just look allow different amounts of the co2 to freeze out into the grains and just solve for the thermal balance of it of a model core and it really makes very little difference and we've we've done some models of this in simulations and yeah we find something pretty similar yeah but I mean it's the dust temperature that's important they're not the gas temperature you're not gonna get much freeze out if your gas is 10 Kelvin if your dust is still 20 Kelvin the other than one see dust gets down to 10 Kelvin as well then then you do get have that being a big effect although something that we have found is also important is properly accounting for normal breeze out if you just have fries out and thermal evaporation then you find that freeze-out should be really effective and should completely change what you see in NC oh and once you then put in non thermal desorption due to cosmic rays that completely changes again and it goes back to being just important that identities which I think much better matches what we actually see observation I wish I had a good answer to that question now I mean this is something that we're currently looking at and other people have looked at so from the point of view of actually regulating star formation I would say the cosmic rays are probably not that important except they may play play some role in actually driving galactic scale outflows from the point of view of what you get in terms of an IMF that could be a lot more interesting because cosmic rays give you a way to heat the gas at identities even when it's very well shielded from the background radiation and if you can heat it enough then yeah that's potentially going to change the scale on which you're fragmenting and what you get is an IMF but that's difficult to model there are claims that cosmic rays could give you a different IMF nobody's actually done fully resolved simulation demonstrating this because it's hard so it's something we're working on that I think we don't yet really know how important that is compared to other effects that we know every important like feedback from the stars you've already formed yeah and then from the point of view of tracer chemicals cosmic rays can be important for clearing out a lot of your low-density molecules so Thomas B's Bess and others have done quite a bit of work showing that if you have a very high cosmic ray flux you lose a lot of your low-density Co even in regions that have a lot of dust shielding and that's potentially going to be an important effect if you've got a very large number of cosmic rays it's going to change what your Co and other low density traces are actually tracing or another way in which cosmic rays might be important but HC n HC n is very easily excited with electrons so you can have enough cosmic rays so that your electron fraction is high then you actually get much more HD on a mission than if you're just relying on molecular collisions so in usual in normal molecular cloud conditions that's probably not important but in someone like the CMC that might actually be be pretty significant it's a it's not been much explored yet mainly because there's not enough people to work on it and the reason that I've shown you two examples taken from SPH is it's easier to do in SPH because the SPH particles themselves are the tracers so you don't have to put in additional tracers but really there's no huge problem with doing this with with any kind of like separate Lagrangian tracers so you could you could do this in a repo using the the tracer particles you could do this in flashy or I'm not sure I think there are traces in Enzo but I mean as long as you have some kind of Lagrangian tracer that you can modify to store the quantities you need to store then you can do this it's just it's easier in SPH because there's less bookkeeping it's something I've thought a lot about the the difficulty is figuring out when to switch so yeah I mean I think yeah there's as potentially a really promising technique but it's not immediately obvious how to do this in a way that ensures you're always self consistent I mean there are there are some techniques from actually from combustion chemistry that try and do this but i I've never had trying to actually work on this for faster chemistry I mean is that that there's a lot we can learn from other fields here because I mean the folks who do combustion or atmospheric chemistry they're dealing with advection reaction equations so they have a lot of techniques that are mostly not used Astra mystery that we could profitably learn from I mean this some little work been done on this for designing reduce networks I think Thomas can talk more about that but there's there's a surprising amount of stuff that there's just never really been tried out for Astor industry oh okay so how one my name is for long as my fellow here surfing so I'd like to show some results for our recent ARMA survey which looked at young discs in Carson region so what makes the surveillance special that our sample is selected to be less unbiased in terms of this brightness so with the typical resolution of 15 au we found that many discs show substructures in their mini meters at a screen distribution and so the most common types of structure can theta is the excel metric caps and rings in systems and we rarely see sparrow patterns or high contrast as Musa variations so we found that there's no correlation with this gap or wing location with Stella nasty which tells us that the Iceland happiness may not work in all cases but what if they are caused by planets then those young planets should be low mass Neptune's in outer disk well how to form networks outside thirteen are you in one to two million year time scale is do or challenge but in in addition to those beautiful rings and gaps we also found that many disks that are spatially resolved but without any textual features as showing here so those smooth disks are distinct from most wing discs in their thirst emission radius so this plot really compares the disk radius and brightness for the twosome house you can see clearly that a larger disk we always host the strings but a smaller disk and not intrinsically finger so the reason behind the size difference is still unclear but I think it very likely related to the different initial conditions in the disk formation stage so I hope that this result will be interested to people who working on this formation series in this room we also have a few papers coming out from surveys so if you're interested in the results protect me afterwards thank you [Applause] okay so I'm Anna we all know me by now um so I'm not gonna go into too much detail about the results here but I will just kind of give a brief intro know what I work on so I use numerical simulations to study how feedback affects massive star formation I do this by developing new techniques used to model feedback so I've done a series of papers studying how radiative feedback effects a crucial into massive stars and also how the initial dynamic state of the cores affect the accretion onto these stars and fragmentation of the cores and now I'm working on and finishing up some stuff where I also include how outflows and wind feedback so from massive stars so once the star gets hot enough you will expect that these stars will have isotropic fast winds that can actually shock heat the surrounding material as you see in this picture here but the main result too is that in all cases so far as once you have a thick accretion disk around the star a feedback will kind of evacuate material along the poles of the star but the accretion disk will actually confine the effects of feedback and you still get a creation onto the star I'm also looking into how magnetic fields of in conjunction with feedback effects accretion onto the stars and you get very different results when you include magnetic fields so come talk to me if you have any questions thank you [Applause] yeah my name is Henrik boy I'm from the max Blum site in Heidelberg and I have neither a slide nor a poster so but I hope that you talk to me nevertheless in the end and in general I would say my interests been vast range of scales from is m and star formation physics vast range of scales I would say from almost Milky Way structures to hundred parsecs or longer whether they are atomic or molecular down to the high mass star formation processes and sub thousand au scales and we do this in different observational approaches via different large surveys for example the for survey that's an H 1 and H and other trace or survey of the Milky Way and you will see more of this by UN and unis later on smaller scales we're doing this for example with the core survey that's a large program at the Platte for the beer which dissect 20 high mass star forming regions and again you will also see more by Colleen and Sue you about these kind of things and then we we continue and do Alma projects to do the fragmentation on the smallest scales but also magnetic fields that is zooming a little bit out I would say I'm not so much an equilibrium person but I think that the whole star formation nice information process are really something extremely dynamical and we try to trace the gas flows from the largest spatial scales of hundred parsecs down to in the end the very inner protostars via the disks and it's yeah it's a very dynamical process and one thing we could do in a unconference is how to trace these dynamical process the tracing is incredibly difficult and how to do this again that would be one of the things thank you very much [Applause] hello my name is will Armentrout i met neither harvard nor heidelberg I'm a postdoc at the Green Bank observatory thank you to everyone for the invitation to be here most of my work has been on tracing high-mass star formation regions across the Galactic disk so we're using h2 regions as tracers and we have a catalog of 8,000 h2 regions and h2 region candidates one thing you can do if you have a galaxy-wide census I start to trace some Galactic structure because high master stars form preferentially in spiral arms we're able to try to start to trace some of that structure with these regions in particular today I'll be talking about the outer scoutin Centaurus arm which was discovered by a little telescope that is on top of this building or one of the adjacent buildings the mini telescope on top of this building it is an extension of the scoutin Centaurus arm into the outer first quadrant so if the Sun is right here at this dot we're looking on the outermost reaches of the first quadrant it's really the boundary for a high-mass star formation within the Milky Way it's about 20 to kiloparsecs from the Sun 15 kiloparsecs from the galactic center and the reason it wasn't discovered until 2011 is because of warps above the plane of the galaxy it's so far from the galactic center that it's affected by galactic warp which we can see through integrated h1 here bends up to about 4 degrees above the Galactic plane and so we've discovered 17 new h2 regions in this outer arm and are turning to the Green Bank telescope at 89 gigahertz and 110 gigahertz to trace some of the molecular gas associated with this star formation so here's one sample region this was actually discovered through Palomar plates in the sharpless survey 60 years ago or so so this region was known but it wasn't known to be this distant and again we have 16 other very distant star formation regions here I would be happy to talk with you about any of those hello my name is Sameer Surrey I'm a postdoc at MPI a I'm gonna shamelessly introduce not one but two of my projects the first one is as Henry introduced briefly is this fragmentation and high mass star formation within the framework of the core project does this work yep and so we observed this particular object at very high resolution with with the high frequency band at naima and this object is particularly interesting because it was discovered 35 years ago by Charlie Lauda as an outflow source but despite the fact that it was speculated that it should be fragmenting it has never been observed at high resolution until now because it's out of Alma's reach so with knowing why we detect at least five cores at this particular frequency and the continuum emission and this is also shown in the line emission that we detect and because we have lines we can also characterize the discs around the object my second project is concerning star formation feedback and this is in collaboration with the Carmen or Orion survey team we discovered this this arc kind of arc like structure in South in the southeast of on c1 and then later we looked at it in multiple wavelengths and we discovered that this is not a simple art but it's composed of many of arcs and then it also encompasses this is an a.1 it shows the PDR or on this region it also encompasses two Starla scores and ripples so instability ripples and we discovered that this object is not powered or objects are not powered by one one source of feedback but three so they're located in ONC pore and on so you want South and this is very interesting so if you're interested in high mass star formation feedback or this fragment please come talk to me [Applause] all right cool all right so for those of you you know me my wondering like what's this one person from Texas doing here so the answer is that I was a student with Alyssa she's still writing letters for me thank you but the longer answer is that as Alyssa mention our Heidelberg was the idea of a however high over a conference actually came about about seven years ago and that's when I actually started grad school in in astronomy so I think it's fair to say that Habra Heidelberg marks the progression of me as a an astronomer and and today I'm going to tell you about a story that we've learned with machine learning about the progression of course and so some review of the history here there were three classes and then they would war so as the the progression of as we've got better and better observations people start to be able to analyze earlier stages of star formation and thanks to works done by few and others people start to think about star formation in ten scores and even earlier stages of course and the fascinating thing about star formations that it connects the turbulent molecular components the cows with the stars so all the way from you know like a few hundred particles per cc or even smoke even lower - you know stars stuff right stuff but even though the most so most exciting episode of star formation take place within these found cores and most studies focused on them you start to wonder what was before that my career as an astronomer is exciting but no my life before becoming an astronomer also matters so so how do we learn about about these cores before they become they'll actually ourselves evasions gave us quite important hints I'm back so these here I'm showing max of two different regions on the Left facing a DA in the future on the right people reason in Perseus and the color codes corresponds to the language or the project version we we measured from with ammonia observations and as you can see the brighter regions show where the rockets version is slower or thermally dominated essentially so throughout all these different regions we observe a whole group of what we call cultural regions so these are essentially regions where we observe velocities version that's thermally dominated and as you can see at the periphery of these structures the changing about this version is actually quite sharp but all these coherent structures actually come in different flavors so for example on the right hand side the first directly observed coherent core host star formation at a time scale of a few tens of thousands of years and on the left hand side are with the GBP ammonia survey we see a lot of these coherent structures that are not gravitationally bound and we have some guests about that all right so so yeah so these are also already very dense structures all right so if I take a look at at the balance between different energies for example on the left hand side parting on the x axis is the kinetic energy of these structures and on the y axis is the gravitational potential energy as I said we observed a whole population that's not gravitationally bound when I compared it to the energy provided by the pressure of the the ambient gas curve emotions then you know the the pressure the turbine pressure could be enough to keep these structures together and these structures also have very shallow density profiles so comparing the radio identity profiles to a critical bombers fear these structures have shallower profiles at the right hand side I I also show that these the density profiles of these structures are shallower than previous observation start its course and it seems that these structures are very are closely related to two turbines in the dense gas component of a cow so for example here I'm showing a histogram a distribution of the centroid velocity or line-of-sight velocity of all these structures and the the shape of this distribution resembles what we see as the average line profiles traced by ammonia so so it's intriguing right so but you know when I tell people about you know we've found a lot of gravitationally I'm down course yeah they are coherent but they're not really found myself gravity some people told me that yes probably has nothing to do with star formation they are interesting but no like we can if we want to study stuff for making we don't need to care a lot about them and I say not really so if you take a look at a logical evolution the rules don't is the survival of the fittest but we also found many weird bizarre species that's not the fittest and why because the evolution history is a weight okay because the evolution history is a dynamical one right so it just keeps going on so like at any time step you don't have to be the fittest but you can evolve to become to survive essentially so after moving to Texas I started looking into the evolution of the density structure using energy simulation and and what I do to these simulations that first extract density structures with ten program you can you know it's like it you can apply your own favorite but I love dendrogram it allows me to extract all the independent energy structures in these simulations and and I do two things on the one hand I extract because I've I know where the density structures are I observed the velocity dispersion profiles the radio profiles of the rozzers version and the radio profiles of density and I put them through a principal component analysis which basically would tell me the how different or how similar these different profiles are in a high dimensional space 600 on the other hand I'm also tracking these structures so these two are independent from each other so the result is that so here I'm showing a distribution of all the structures extracts from simulation in the PCA space and so you know as if two structures are close to each other in this space then they have similar distributions of dispersion intensity so eventually what I found is that on the Levantine core coding the distribution with the size of coherent regions on the right hand side and color coding the distribution with how steep density profiles are right so at the end of day I find that there are three different groups so towards the top there are non coherent structures with very shallow density profiles towards the bottom left have coherent structures with mild density profiles and towards the bottom right I have essentially cell forming course and actually I can extract that with some simple cross turing algorithm and and if i overlay the tracking results on top of it what I find is that you know there's is actually evolutionary track connecting these three groups so going from unbound over densities in very turbulent median you can you know they eventually become coherent structures and then they would go to to become the star-forming course that we're familiar with so this is great so and if I take a look at the change in the resistors and profiles and the density profiles of pop is the versus detergent profiles you see that the change from the first group to the second group is that computer regions start to grow in these Rogers turbines is being dissipated meanwhile the density is not has not grown and grown a lot during this step and then only on here we go from a group to two group three we start to get the growth intensity and gravitational info so really what we're seeing here are it's a two-step process that in the first step or three steps so I formed these structures in terms of media as turbine perturbations maybe and then the turbines within these structures starts to get dissipated and then and it only after that do we start to get gravitational even for info right so I also you can also estimate the timescales and quite intrigued indeed like these time cells are pre fairly well with what Mordecai and Eve Ostreicher calculated as the turbulent energy dissipation timescales and lastly but most importantly is that if I put these structures in a similar theory analysis so again kinetic on the x-axis gravitational energy along the y-axis the three roots actually occupy different spaces in this in this the real space you can call it so so there seem to be there seems to exist a evolutionary track that brings all the sensory structures in the turbine median to close to a very own equilibrium and until like only after that these structures start to grow and become protector a mouth and and start forming so and you know like there is always some variations do you to know you can start with seeds of different sizes and methods essentially so right and that also explains this whole population that we observed with ammonia they seem to be at different stages and they could form out of seeds of different sizes and masses so that's the I want to show you what where these structures are so so in simulation these structures seems we find structures at both outside and within the densest filaments in the simulation and so these structures they essentially formed everywhere in this in this MHD box and they as they evolve they do not have to be bound as I said they evolve dynamically and they can gravitate gravitate towards the denser regions and have the mass reservoir for for them to to accrete and become star forming alright so so just to quickly summarize the observation those bound and unbound coherent structures give us important constraints they have these coherent structures have to form constantly and they do not have to form gratification adapt and there seems to be a prolonged coherence phase in the evolution of course and the formation of course seems to start with the turbine seeds followed by a outer anticipation and after that what we are most familiar with the gravitational uniform alright so I can take questions [Applause] very nice can you use these three categories to identify observational stage a particular object is in and when you do that do you find that young ones are near young ones old ones are here old ones or is it all just kind of intermixed it doesn't matter location doesn't matter yes you can do that but yeah that that's of course much time so now we have a full evolutionary picture for you essentially we need to add you know identify these structures their population in cows they would be super interesting to look at but I haven't for the observations but I think I can do it with the simulations and exactly I think it would be a very interesting thing to look at yeah how long are they expected to spend in agencies oh there it is yeah so yes so so that that's that's what I would think about our feels clustering psyche because now we have a full evolutionary picture we can do population analysis of observations and to know and also the formation of pores as they you know their relation amongst themselves yes so there are there are there is a nonzero chance for proof one to go directly to prove three and not all Group one structures no they are just dense restructuring a target medium they some of them are dense enough to for them to be close to Fowler almost found already so they think you know quick create a priest stuff and become star-forming yeah yeah and some somethings first as well so so there is actually a large proportions you can see myself so we have a lot of different again using the analogy to biology you have a different species and not all of them will will survive the sample some of they are like evolutionary dead end but it does not have to be bound to proposition about you survive because there's a dynamical process so that's one so we've only so this is isothermal MHD with gravity wave so the analysis is some on post of decaying and thriving turbulence and we only started like the initial analysis on summation with feedback the takeaway is that there's always these three stages the physical processes we put in only affect the length of the time scale century but it does not vary by more order of magnitude and the survival rate as I show here so with feedback as you can imagine some of the structures may get destroyed disrupted yeah yeah that would be super interesting because everything's ready for that exactly so uh okay so I haven't but I can't comment on how easy that would be because there as you can imagine checking these structures is difficult that's why there isn't many studies before this and the most difficult thing would be to those tiny structures like if you identified you know tiny tiny structures at different time steps how to correlate them but if you have like broader bigger structures then it's actually easier to to track them and that's let me our our it next step so again going back to my algorithm because centric wins does I only applied endocrine mounted density so it does not excuse any structure it basically takes up all the independent entity structures and what I'm guessing is that just with turbines it sometimes it kind materials can get disturb is not to to have like local density Peaks other because like I have not done a systematic analysis down exactly what creates that but I would think that's they naturally formed in a turbine environment because what does it very size scales say these things so so they're they're all scored but like these structures start out with like you know stop something that go go up to that point a few parts AG something smaller than point one four point scale yeah yeah sure Gayle's and see yeah so so other than there yeah so other than the size scales the time to actually are so so I just did this comparison so just the other week and it's actually consistent this group one time scale is actually consistent with the kind of turbulence dissipation that's more like high and these are became two and I think in their papers they're finding that most of the dissipation occurs because of the existence of magnetic field because it get like you can easily channel out the programs within these sensitive structures and I think that to here no visit question so what destroys this hmm what destroys what poses the sources oh yeah so okay so so again it's it's very dynamic holy alright so for example well a lot of the group one structures are not curved attention right now okay so and as you can imagine if eight if something that's static they stay there and that's not rotation rate bounce they will just get dispersed but something are lucky now that's they may be net the the package of material within these structures already have some momentum in it and they gravitate toward sensor region they find some mass reservoir and they they are naki enough to evolve on so so these structures gets dispersed naturally in a turbine medium as they are to instructors and Besant initial look with simulation with feedback is that feedback destroys them right so you know like I think yeah so so that that's my best sensor right [Music] okay

Channel: Harvard Astronomy Video

Views: 74

Rating: 5 out of 5

Keywords:

Id: LECsG62XEP4

Length: 110min 18sec (6618 seconds)

Published: Tue Nov 12 2019