 Hello, Jorge. Hello, everyone, and welcome to a new web seminar, the number 23 of our series of the Latin American Webinar in San Francisco. My name is Jorge Diaz from the California Institute of Technology, and I will be your host today. Our speaker today is Carlos Arwayas from MIT, who will talk about the serenutrinos and ice group. Carlos received his PhD from the University of Oaxaca, of Wisconsin in medicine, and now he's a postdoc at MIT working on ice cube in the ice geoglutrinos telescope. Carlos' talk today is titled Results from the Search of EV Serenutrinos with Ice Cube. We're really glad to have him as a speaker today, and I remind you that you can be part of this discussion writing questions on the questions and comments using the Q&A system. So remember that you can be part of all of this. I will now hand you over to Carlos, so Carlos, all yours. Thank you, Jorge. Yesterday I will go to talk about our latest serenutrino analysis. I'm going to go into details of the analysis, and you can actually now read about this result. It's in the archive numbers in the bottom of this slide. And so please check it out, and at the end, of course, there will be time for discussion on questions. OK, so the menu for today, we are going to go over three topics. The first thing we are going to discuss is a very fast review of neutrinosilations, and the MSW effect is going to be very important in the ice cube serenutrino search. Then I'm going to explain you how ice cube works. And then the bulk of my talk, most of it is going to be on the actual serenutrino search that we have performed with one year of data. So let's start with neutrinosilations. So neutrinosilations in two generations follow a very simple prescription. The oscillation probability, or rather survival probability, of a neutrinosilation starts with a flavor alpha and remains of that flavor alpha after it travels a distance L and has an energy E, is given by this formula. It's controlled by two parameters. One is the sine square to theta, which is going to gauge the amplitude of the oscillation. And the other parameter is the delta M square, which is going to control the frequency of the oscillation. So in this little cartoon down there, you can see like if I start, for example, with a pure new mu state, as it travels, it can transform to a new E and go back to new mu. And the two parameters will control amplitude and oscillation respectively. So why does this happen? So why do neutrinosilating this fashion? Well, our current picture is that the so-called mass eigenstates are the vacuum propagation eigenstates and the flavor eigenstates are not the same. They are actually related by the PMNS matrix or the leptomixing matrix here represented with this letter U. And the leptomixing matrix U you can see that it has a very large of diagonal components. So it allows for very large flavor transitions. Unlike the CKM matrix, it is quite non-diagonal. We usually parameterize this matrix as a function of three angles and one CP phase. We have measured with reasonably good accuracy these three angles and the CP phase is yet to be measured. So then we have three massive neutrinos. And of course, there's going to be some mass splitting between them. And there is a smaller mass splitting between what we call the neutrino 1 and neutrino 2, which is the solar mass split, which was measured first in solar experiments. And that is around 10 to the minus 5. And then there is a larger mass splitting between the first and second mass eigenstates and the third eigenstate. It was measured first in atmospheric neutrino experiments. And that is around 10 to the minus 3 and electron more squared. The solar mass splitting, we know it's sign. But the atmospheric mass splitting sign is not known. And this gives rise to neutrino ordering problem, which is just to say, if the mass, I guess, say 3, is heavier or lighter than the 1 or 2. And so these very simple equations in this picture actually describes most of the data that we have. And then there are only two remaining pieces in this picture. One of them is what's the value of the CP phase. And the other one is what's the sign of the times square emphatic, which is the neutrino ordering problem. And the next generation experiments in neutrino physics are aiming to measure these two quantities. OK, but there are pieces that don't fit. This puzzle has some pieces that don't quite match what we expect. And the most striking, perhaps, evidence or the most significant evidence of something that doesn't fit is the LS&D experiment. And the LS&D experiment was a new new bar, I'm sorry, a new mutinui experiment. And it saw some excess of events. So in the left plot, you can see the distribution of events that LS&D had as a function of L over E. Those are these black data points. And the background expectation that LS&D had for new E appearance was the green and red histograms. And you can see the green and red histograms cannot explain the data, so the black dots. And so you need to add a third component that comes from new bar, new E oscillations. And that is the blue histogram in that plot. And then using the first formula we discussed, you can convert that particular excess into a parameter space of the mixing angle and the mass square reference. So if you do that, you end up with this plot, which is in the y-axis, delta M squared, and the x-axis, sine squared to theta. And so the green region is the region that's compatible with the data on the left plot. And you can see what's striking is that the delta M squared that's required to explain this anomaly is significantly different from the atmospheric and solar mass delta M squared that we have already mentioned. And we can only have, of course, two mass square difference when we have three new neutrinos. So this is a problem, right? And the solution of this problem, one of the solutions is that it's included for neutrinos. And that neutrinos happens to have around 1 eV squared of mass difference. And so we have seen some anomalies in the electron appearance in BDME. There have been some claims of some disappearance anomalies in reactor experiments, for example. But if you observe electron appearance, then the product of U mu 4, U e4, must be non-vanishing. And that implies automatically that you must also observe neon disappearance, like we have so-called seen electron disappearance. And so far, we have done long-based experiments and we have self-form noon disappearance, but we have not found it. And so Ice Cube is going to try to see there is mu disappearance compatible to one electron-vault-square-themed neutrino to see if this picture makes sense. So how do we introduce this extra cell neutrino? Well, when one does this, one puts a four neutrino mass, I can state, that can be lighter or heavier than the active neutrinos. So it went into one mass splitting three angles and two CP phases. And so usually, we talked about what we call a 3 plus 1 model, where the three active neutrinos are more or less in the original state, because the mass splitting is so much more than the external neutrino mass splitting. So we can also talk about a 1 plus 3 model, where the external neutrinos is lighter than the active ones. OK, so what does the word say about this model, this picture? Well, there is very large tension. So if you look at the left plot for a paper by Jochen Kopp, you can see here the effective new angle. So the blue line is the 99% exclusion. It excludes everything to the right of it. And the solution for the LSND and reactors and so on are these red balloons. And you can see that the red balloons are the right of the blue line. So that means that there is a tension between the experiments that have not seen cell neutrino-like signatures and the ones that claim signatures. So one can subtract the right minus the blue lines and do a lower fit to put it all together. And we did that very recently with recordings and collaborators. And we find that there are still some solutions that remain after you subtraction processes. And this is shown in the right figure. But OK, so the anomalies still remain after one considers updated data. So what we need to do is we need new measurements to really disentangle what's going on here. So let me introduce what we're going to use in a very simplified model. And so as I was telling you, external neutrino can be actively heavier or lighter than the active one. So if it's heavier, we call it a 3 plus 1 model. And if it's lighter, we call it a 1 plus 3 model. So Cosmori doesn't like to have large neutrinos or neutrino masses, at least in the 1 EV square, one electrical size. And so that puts more strong constraints in the 1 plus 3 model. So we are going to think about what follows in these talks in 3 plus 1 models where the cell neutrino is heavier than the active neutrinos. And so when one introduces the neutrinos I was saying, you introduce two triangles and two extra cp faces. So for simplicity, we're going to set all cp faces to 0. And then there are triangles. So we have to do something about that. So what we decided to do is set two of them to 0 and only take into account theta to 4. And there are good reasons to do this, but perhaps one of the nicest things is that if you take these assumptions, then sine square theta to 4 is exactly the same as sine square 2 theta mu mu, which is the amplitude of mu mu disappearance in vacuum. So it has a very nice interpretation. OK. So how does ice cube connect to this shore baseline anomaly? Those are these LSND problems. So in this figure, in the x-axis, I have the neutrino energy in GeVs. And in the y-axis, in the left one, I have the baseline in inverse GeV. And in the right one, the baseline in meters. And this plot is made such that if you go to the upper corner, that it should be 12,000 kilometers as the diameter of the Earth. And so what you basically tell is if you want to observe one given mass square difference, you need to put yourself in the correct baseline to see the first oscillation. So for example, if I want to look at the solar mass splitting, I have to be in the L little sun symbol, which you can see crosses the Kamlan experiment region, which is an experiment that has ordered MEV energies in hundreds of kilometers. It's in the right position to see that particular oscillation length. Now if I want to look at atmospheric mass splitting, then I have to be at a different baseline. And that's given by this dashed line. The dotted line. And you can see that line then crosses a super camuconde, which measures just in the GeV and hundreds of kilometers. And it also crosses this cluster of experiments, which are more tuned to look at this particular oscillation baseline. Now in pink color, I put the experiment that actually claims some signal of anomalous behavior compatible to 1 eV square telonetrino. And so then I can draw the baseline that would be required to observe that if the telonetrino mass difference would be 1 eV square, which is this dotted line. And you can see that this line crosses more or less the experiments in the short baseline anomalies. And then it's going to keep going. And it's going to grow the ice cube regime, which is basically a very high energy, very high baseline, a very long baseline regime. And so that means that if one believes on this very scaling of the oscillations, that means that the short baseline anomalies are a counterpart of the ice cube, or ice cube would be a counterpart of the short baseline anomaly in the high energy long baselines. So they should be compatible with the oscillations. OK. So what does ice cube typically observe is atmospheric neutrinos, or ice atmospheric neutrinos are producing cosmic rays showers in the atmosphere. We know the flavor composition of the initial neutrinos and the amount of 200 neutrinos. And basically they're going to interact in various directions. And the directions we can parameterize, we call the synid angle, which is the angle representing the digital diagram. So synid angle equal to pi, or cos theta equal to minus 1, is it means that the neutrinos will go through the crust, the mantle, and the core. And the English ice cube has to cross the whole earth. And as one decreases the synid angle, the neutrinos will have a smaller baseline and cross different earth layers. So why am I making a point here that the neutrinos will cross different earth layer for different baselines? And the important point is the following that in matter there is a so-called Miga-Hebb-Smear-Wolfenstein effect. And what this means is that the oscillation probability will actually modify in the presence of matter. And so the oscillation formula follows the first equation here, so it has the same shape as the vacuum equation. But now we have modified the mixing angle and the mass core reference with the presence of matter. So what's notable here is that if you look at the formula for the tangent of the two times the mixing angle in matter, that has a denominator that if you make it zero, it will make a sine square to theta maximum, so it's one. And that will happen in a specific condition. So here A is the matter potential term. So when that denominator is zero, you will get this resonant enhancement, so-called MSW resonance. So the resonant play with transition actually will happen when the resonant energy, which is this bottom line equation, is equal to this parameter here, so it will depend linearly on the delta M squared. And you have to, it has two sines. The first, upper sign, the minus sign, is for neutrinos and the lower sign is for neutrinos. As you can see that if we have assume a three plus one model, so the standard three is heavier, delta M squared is positive, so the only way to satisfy this equation is to choose the plus sign. So what's going to happen is that for some particular densities where this condition is satisfied, there will be an enhancement between active and sterile anti-neutrinos transitions. So how does this look? So this is kind of a strike of light, but if you take delta M squared to be one electron ball, and then you take the Earth's core density, and you plug this into this equation, you get that this energy comes around one T, and this is going to be a very lucky number because ice cube is very effective at measuring neutrinos in a particular energy range. And so this was pointed out for the first time by Nuno Kawa and collaborators in 2003. And the way this looks is like here in this plot, they have the neutrino energy in the x-axis and the probabilities in the y-axis. And the blue lines are the neutrino, the neonotrino probabilities. The dash ones are actually the neonotrino probabilities and they go to one as the energy increases, as one would expect, but the solid blue ones are the anti-neutrino survival probabilities. And as you can see, one approaches this MSW resonance somewhere around a couple of TVs, one gets this massive conversion between new bars into basically steriles. So this MSW resonance effect is the effect that we're going to try to look for in this analysis. So let me shift gears a little bit and talk about ice cube. So when ice cube is in the South Pole, South Pole is a very large place, so there's a comparison to the U.S. So it's all the way over there. And in the surface of the South Pole, there is the South Pole, the ice cube lab, which looks like this, but the ice cube actually, most of it is not in this lab, as you know, most of the experiment is about one kilometer and a half below the ice. And we have instrumented basically one kilometer of ice with PMTs. And the way that we have done this, we have arranged them in strings, as we have 86 strings on which we have domes separated 70 meters, about 60 domes per string. And each dome contains one PMT that, of course, can convert light into charge. And then we have a digitalization signal on inside the dome. And then the signal goes through the cable and we can process in the ice stop, I'm sorry, in the ice cube lab station, and then we chip it to my somewhere we actually do our post-processing. So ice cube, I think it's quite famous now for discovering the high energy neutrinos, so the astrophysical neutrinos. And the way we did that is we looked for events that started within the detector and did not deposit any light in a veto region that was adequately defined. And then we looked at the events that deposited a very large amount of energy in the fiduciary volume of the detector. So what you can see here in this plot is the deposited energy in TVs. And here the backgrounds are the pink line at most. And you can see as one increases the energy, you see this hard component showing up very clearly. And the two highest events here are better and early. So we name our high energy astrophysical events by map names, and this happens to be these two ones. But ice cube, I mean, has discovered astrophysical neutrinos. We have also found evidence of these neutrinos through ongoing muons. And so what are those events? So in this case, the detector is like this square and a neutrinos coming somewhere through the earth and it may interact inside the detector like in the previous analysis. It also may, and typically does, interact outside the detector. And so then it will produce a muon and the muon will reach the detector and you will not see the starting point of the neutrino. So in these cases, you can see the geometry of the light deposition and that allows us to measure the angular direction of the neutrino to about one degree, a little better actually. And then we can still measure the energy. We have a measurement of the neutrino energy by looking at the energy deposition of the muons. So in this upper right equation, I have the beta block form and you can see that at the hundreds of GVs, the energy of the muon and the DE-DX of the muon start to have this linear relationship. And so that means that if we can measure the DE-DX, then we can have a proxy for the muon energy and from that we can then extract the neutrino energy. And so how this actually looks in our reconstructive quantities is shown in this plot in the lower right panel. And so you can see the muon energy proxy, there's overestimation of the muon energy and the y-axis is the neutrino energy and about one TV, they start having this linear relationship aspected by muon energy loss process. So basically by looking at the geometry, we can look at the direction pretty well and by looking at the energy deposition, we can get an estimation of the energy. So this is how one of these two muon events look like. Here, each little dot is one of our doms and then the balloon size is the amount of charge deposit so the bigger the balloon, the more charge was deposited and the color specifies the time, so right is earlier times and blue is later times. So that means that this event starts somewhere in the bottom, they take all and made its way up to the top, we have this particular direction and again what we do is basically we can obtain this direction which is the red line here, is our best fit reconstruction direction by minimizing on the geometry and time of this charge deposition and then by looking at the energy position along this line, we can estimate the energy. Okay, so there was an analysis that was for this MPRL, some years ago that we looked for this high energy neutrinos, it contains about 20,000 events and we are going to base our analysis on this particular analysis and this is how the distribution of these true going new new events looks in energy. So in the high energy tail, these are dominated by the MAPET neutrinos, there's physical components there, it's very strong. Then with a look at lower energies, then it's mostly dominated by this non MAPET atmospheric neutrinos. And of these neutrinos, we have actually used the lower energy ones, not in this particular cut selection, in our selection, but we have used these low energy neutrinos with deep core to actually measure the standard atmospheric parameters pretty well. And we have policies in PRD last year. So let me just hope to what external neutrinos are to exist. So external neutrinos are leaves, analysis leaves in the middle of this distribution, but it's neither in the low energy tail where the standard oscillation happens or the high energy tail with those astrophysical neutrinos have been observed. It's in the middle of the distribution. And the reason is that, again, there is this amazing conscience that if you take one electron per square cell neutrinos, much difference, you happen to have a resonance energy around the middle of this distribution. So it's a very good place to make measurements because we have lots of data that we can use to constrain this. Okay, so this is where we are going to leave. So it's going to be a little more complicated than just looking at the energy distribution, just to use information from the direction and the energy. So what's going to happen is like, you can see here a little error there, and neutrinos can come from different directions. So if they come from where the blue arrow points, that's cost theta equal minus one. And if they come from the green arrow, where the green arrow comes, that's cost theta equals zero. So then I can put this so-called zero rounds where in the y, in the horizontal axis, I have the cosine of the angle. So where the blue arrow points is, it means neutrinos go through the earth, and where the green arrow points and these neutrinos just come through the horizon. I have the zero rounds for neutrinos and neutrinos. So one thing that you can see already is that if I am at very high energies and I'm crossing the earth, neutrinos will get absorbed by the earth. This is just the earth capacity, which is this blue corner in both neutrinos and neutrinos. And then the other striking feature is that for neutrinos, as we have promised, there is going to be this MSW conversion around a TV. This happens for these parameters that are L, C, D, minimum like. And so what we are going to try to do is we are going to try to identify these resonance. And so how does this resonance depend on the sternotino parameters? Well, it's a simple relationship is that the resonance energy is linearly proportional to delta M square, as you can see in the equation up here. So that means that if I increase my delta M square, my blob, this blue depletion in antinotrinos will go higher in the energies. And if I make it smaller, it will go smaller in the energy. The angle of xenotrinos is small, so cos theta to one is close to one as a subline effect. But basically we have very good handle on the delta M square. Okay. So this of course is not so beautiful when one actually looks at this and reconstructed of the point is where this is energies meaning or any resolution is good, but not amazing. So in this plot, I show in the upper left panel is my parameter space. I have again sine square to the fourth that's the amplitude in vacuum. And then delta M square for one is the y axis. And then I plot this case is gray value which is a representation of the L, C, D minimum anomaly solution. And then I pick three parameters points here. It's like this layer over the stars, one, two, three. And for example, if I'm at parameter point one, then the effect that I see or the effect that I see with respect to more hypothesis is shown in the right upper panel every by one. And you can see that it makes deep somewhere around the air crossing direction. So cos theta equal minus one and it's a 10% effect and has this very specific energy signal correlation. Now if I move from one to two, then the MSW energy goes down. So I expect my disappearance to have been a lower energy. So which is what you see in the second panel. And now if I move to point three, point three has a bigger angle. So the effect, the vacuum system are more stronger and so you have a stronger effect and also it happens on a little higher energies because we have increased L times square. So basically for each point in this parameter space there's a map in this reconstructed energy synid map space. And so what we basically are going to try to do is given the data which of these maps fits the best with some losses parameters and so forth. But this is basically the idea. Okay, it's important to know that we want to look for our 10% effect. So that's more the size of our signal. And for that it's important to consider the statistical errors that our sample is going to have. So we are going to use one year of samples as around 20,000 events. This is the meaning that we are going to use. And so in the region of interest is going to be the MSW region is around three, is around log 10 of three, so one around a TV. And so in that region of interest our statistical errors are going to go between 10% and like 4%. And so it's going to be enough to measure this very particular correlation we're going to look for. This is going to give us quite a lot of power. But that also means that what we have tried to achieve in this analysis is to control the systematic uncertainties to the 10% level. And systematic in this analysis are very important. Some of them are going to be more important than others of course. And so we try to parameterize all of possible systematics that we could think about. And so here I have more or less listed all of them in order in importance. So one of the most important systematics is going to be the actual efficiency of our domes which is not a completely certain quantity. Then we have the uncertainties in the atmospheric flux, cosmic race hour and stuff like this, studies on the ice and so on and so forth. I'm going to go through a couple of these just to exemplify. So first let me tell you what are the fake flux dominates in this particular region. So atmospheric flux in this analysis is basically dominated by the Keon components so that we can assume that most of the neutrinos actually come from Keon decay. And there is going to be a subdominant component in the pion decay which is the threat line. Prompt, so which are charmesons decays are going to be totally subliding. So we are going to assume that our atmospheric neutrino flux has two components. One is the one from Keon decay and the other is from pion decay. So what we did is we took a bunch of cosmic remodels. So this is the cosmic ray spectrum in the region of interest more or less like power which is 10 to the power, below 10 to the power g is in primary energy. So we took a couple of cosmic ray models and then we took a couple of hydronic models and we solved the cascade equation for each of these combination of hydronic model and cosmic ray model to calculate the yield of pion and Keon neutrino fluxes. And we did this for these six combinations and then we also took the Honder-Eiser model which has also some predictions of the expected pion and Keon components. And then when we said, well, our atmospheric flux is going to be parametrized by basically adding the pion and Keon components, there is some relative uncertainty in how big these two components should be. So we allow for a for a for a nuisance parameter to gauge this and so it's going to be a Gaussian and this parameter. And then to account for the uncertainty in the cosmic ray slope which in this energy rays is very good approximation by a power load. We account for this to do this spectrum in the shift and then the overall normalization of the atmospheric flux is going to actually float. Initially we let it flow before the prior and then we're going to discuss the case where we constrain that. Then the dome efficiency of the detector, well, the efficiency of dome is not precisely known and what actually matters is the precision is the actual, I'm sorry, efficiency of the dome when it's coupled to the ice and there is a cable around the dome that actually changes the overall precision altogether when you consider the dome. So this is what this is going to do is going to shift your energy distribution as you increase or decrease your dome efficiency and it has a civilian seat and dependence too. Then the other thing we discuss is the uncertainties given to the ice. So the ice has different absorption scattering coefficients as you go as function of depth and we have measured this through calibration runs and we have a bunch of models that explain most of the features that we think are in the ice and the way we try to estimate a certain the ice in this particular analysis is we took our benchmark model and then we increase its absorption by 10% so 10% extra photon absorption and then separately will be the plus 10% scattering of the photons and that gives us these changes in the expected event number so you can see it more or less but if you do that it's a very featureless change in the TV regime and it's a 4% change and so it's smaller than our statistical uncertainty so we think that by just considering these discrete variations of the ice possibilities we cover the uncertainties in the ice and then there is another effect which is the whole ice so basically what happens is that when we drill the hole and we put the water back in it has some air bubbles and these air bubbles are going to increase the scattering in the direction of where the hole has been drilled and this is going to scatter the photons and then what's going to happen is the relative efficiency of our domes along the direction that's pointed out by the red arrow is going to be decreased so we have a bunch of models that try to parameterize this effect and we have also the expectation where this effect doesn't happen so that's the no whole ice expectation so what we did in this analysis we basically look at the difference between having this effect and not having this effect and it turns out that this is a small difference so we just consider the two scenarios okay so this was a blind analysis so we did a bunch of clear blinding checks we figured for a new hypothesis in the whole samples when we didn't see it and then we look at the projection so that did not allow us to look at this specific correlation and didn't see it so blindness was preserved these are the energy and signal distributions you can see they are pretty good fits and the red line is the nose they are a lotrino hypothesis so we found that we had a very reasonable chi-square and the nuisance parameters and the nuclear parameters and so on and so forth they all come out to have reasonable values so they give us green light to forward we look at the pools per bin and then an oyster hypothesis which is shown in this plot here this distribution has about 200 bins and in these 200 bins there are five or six plus two sigma fluctuations which is compatible with the expectation of statistical calculation we have tested that this is quite a compatible realization and so this is going to be fine so before I go forward to the end of my story let me encourage you to co-conderize us one of us Ben Jones who was at MIT he's now at UT Arlington and the other one is Joseph Alcerra who was at UW-Miles and he's now at the Fick Valencia and so we decided to open the box and we found no significant evidence for serenotrino and this allowed us to place new bounds so in this diagram in this plot here you see our results for 99, 90% confidence level in what we call the shape-only analysis where normalization of the atmospheric flux is allowed to float freely of course when you do this analysis in this parameter space you're going to find a best fit point somewhere so we actually found it at this delta n squared 10 delta n squared beta 0.76 which is up then in the corner in this parameter space and this particular point happens to be very likely realization of the no-ster hypothesis what we did with the following exercise so we used Monte Carlo with many examples compared to no-ster hypothesis and then we fitted it using our software and then we see what the delta n squared n squared we obtained and that gives you the distribution of points you see in the right diagram here and you see most of the solutions clustered around this upper corner and so what's happening there is that in this high delta n squared regime the oscillations are going to be very fast so they're going to average out which is kind of simplifying this little diagram down there and so they're going to be basically indistinguishable from the no-hepothesis and so sometimes no-hepothesis will feel a little better than the other one and this happens 19% of the time so it's a very quite, it's a very typical minimal location so then of course we are in this high delta n squared regime which where the normalization prior pulls to normalization value pulls to close to two so we said well what happens if we put posteriori and 40% prior normalization and so we can see that the sensitivities of putting and not putting the prior are actually the same but the result is weakened slightly so here in this plot shape only we put no prior and normalization normalization can be whatever it wants right past shape normalization is our constraint within 40% of what Honda says and then the red lines, I'm sorry yes so the solid lines are the right past shape so that's with the prior and the dash lines are the shape one result so it's weakened and the result is weakened it's like we have a likelihood problem in the shape only case so we saw this little problem and we obtained the solution in this upper corner then we put a posteriori, a prior that's intention with this particular solution and that makes the likelihood problem worse and we get a weaker amount okay so we can compare this to our expectation and this is the Brazilian plots that make its function of mass so for each fixed delta m squared you have the median sensitivity and the expected 60% and 95% bansal containing the sensitivity you see that this is very compatible with what we actually get from the data which is the red line so how does this compare with other measurements? Here is the, all these appearance results are 90% here we have minus super case, CHS, minimum cyber and you can see that in the one 2.1 EV range we have improved the constraints on this parameter by about one hour of magnitude so that's pretty neat so now how does this look if we look now with the LSD, minimum balloon which is attracted from copetol so here I have the shape and shape just rate results and you can see that in both cases at 99% confidence level the minimum cyber combination from copetol is excluded, 90% confidence okay so we have performed cells for UV strain neutrinos we found no significant evidence we improved the work limits by about one hour of magnitude in that mass range of between 1.1 and 2.2 and actually I see it has more years of data rate to analyze and we're just getting started we're working on follow up analysis for more details please see this paper on the archive so thank you Carlos for this very interesting seminar let's see we have questions let's first check for the people and be connected to the hangout I have a question for Carlos so Carlos it's very naive in the sense do you expect because in your analysis you assume that you fit your CP phases to zero for the sterile neutrino with the active and everything so do you expect that non-zero CP phases would have an impact in your analysis or could be it kind of so the CP phases tend to cancel so there are two other effects so they're okay there are two angles that have made zero and they have also made zero of the CP phases the two angles I made zero there's a T-34 angle the T-34 angle does have some effect on the result but it can be shown that serenity to zero is actually the conservency thing to do the standard the standard three-neutrino CP phases doesn't play a role here but the sterile neutrino CP phases do play a role but it's a subleading role to the T-34 angle and actually because we have an addition of neutrinos and anti-neutrinos and the CP phase acts in opposite ways it tends to cancel the CP phase effect so it's a subleading effect but okay so the reason we also took this particular model is that it's quite a benchmark model so it's very simple to interpret but of course it's now the task of the global fitters to discuss this for example let me there is this balloon here this is the LSD minimum balloon and so this balloon, so LSD minimum measures new neutrinos and oscillations and so they are measuring this combination which I put here which is sine square root of mu e which is the product of these matrix elements we are measuring actually sine square root of mu e which is the product of these elements here so to say that you exclude it or not you need to assume so for me to be able to put the LSD minimum in this angle I have to assume a value of mu e4 and so the way we do it so we take mu e4 to the westward value according to Copp et al which is this number now you can always of course make mu e4 bigger number and so what that does it makes the minimum sine balloon move to smaller angles in this sine square root of mu e4 and that then escapes a 19% level bound but of course if you do that you increase the mix the disappearance of the e-sectors which are more constrain from than coming from reactors so there are tensions I mean there are effects of the other angles in the overall problem but this is something that that the global feats community has to sort out because of the many pools is coming from different directions and actually me and Janet Conrad's groups we are working on updating our global feats so in corporate is also the ice cube information Thank you and another short question because this is kind of I mean really I don't know what can be done with ice cube I mean in the sense of all the possible different scenarios that you can cover but are there other type of physics beyond known physics that can mimic the same signature that there are neutrinos in your I mean can you have some caveats in your analysis that can you use not only to constrain the neutrinos but other type of physics so it's your question, let me try to understand this is your question is are there any systematic that could mimic the signal or standard models or some physics systematic mimic the signal or is your question is there any other new physics that's look similar to that if there are other new physics that can mimic the signal apart of standard training well I don't know all the new physics possibilities because they are of course too many but yeah this is a very concrete MSW looking feature right so the deep is very the effect is very characteristic because you follow this this MSW resonance condition right so in that sense it's rather specific but this energy range between you know hundreds of GVs to little couple of TVs or tens of TVs we have a lot of data there we have hundreds of I'm sorry tens of thousands of events and so of course this data will constrain more and more for example any science can be constrain also which is something we are working on and Lorentz violation for example can also be constrain using the upper most tail of the sample can be also constrain which is something we are also working on thank you anybody else in there can I ask anything to Carlos I do have a question for you Carlos so if I can go with it so one of the questions that I have is that you mentioned and of course we will know that the matter potential changes for when you have neutrinos or antinutrinos right and show the formula there so the question I have is how do you separate the neutrinos from antinutrinos in ice gear in order to use appropriate this formula so event by event basis we cannot separate the neutrinos and the neutrinos of course at least speaking they will look exactly different in the detector right but the issue is that we measure the sum of neutrinos and antinutrinos and the sum will be modified when you have the antinutrinocomponent have the MSW effect so actually in our sample so the initial atmospheric neutrinos and antinutrinos ratio is about an order one number but the neutrinos and antinutrinos cross section is about a factor of two larger than the antinutrinos one and so that makes the sample has about 30% of the sample is antinutrinos and 70% of the sample more or less are neutrinos and so basically you will only have the effect on this 30% of the sample now of course I can switch my cell neutrinos so it's not 3 plus 1 or 1 plus 3 so I switch the sign of delta M squared and in that case the resonance happens in neutrinos but neutrinos we have more than antinutrinos say and so the effect will be enhanced or larger so we expect that the constraints will be even stronger in that scenario yeah but we cannot distinguish then I guess my point is we cannot distinguish them even per event yes of course on the distributions we can definitely if there is a difference in one of them we can we can we can take it apart so you can basically suffocate them in a statistical way is it similar to what a super-common candidate? not quite I mean we add the components together right okay we don't have a really we don't have like a this analysis doesn't doesn't use like a like for example a booster decision tree or whatever to try to separate them right no we just add them together and then we we look at the sample all right then okay there is one question from the Q&A system it was sent by Mauricio Agustamante but he said this is a question from Secarpio in Lima he said I do not understand why you introduce a 40% normalization prior post-unblinding um well that's a good question I mean the the analysis um as it was designed prior to unblinding was thought to be a shape only analysis so we did not concern in normalization of atmospheric flux um then up then a posteriori we decided that it was a good idea to incorporate the information on atmospheric flux so we look at the to a literature to see what is the quality and certainty uh in the atmospheric neutrino normalization flux and so that turns out to be this 40 percent number um and so I think I think it's valid if you have information uh you can certainly include in the program but yeah this this number this 40 percent comes out of the literature so we can just pick it you know randomly I see it is actually very well justified so Jose if you're connected I hope that answer your question if not you can still have some time so you can some of the question again I don't have actually another question so when people check about the sterile neutrino sometimes they get confused with the idea that sometimes you have to say a three plus one model and there are many constraints on this three plus one model and when people say from talking about the reactor normally they check up about the sterile neutrino and when people consider they say the cosmological bounds they say there is this room for another for a sterile neutrino so which of all these sterile neutrinos is the one that you're talking about yeah um I I'm a little confused so I mean the the reactor anomaly and the lsnd anomaly both point to the same type of neutrinos so these eb square scale neutrinos cosmology doesn't like Mercedes neutrinos so you take like standard land of cdm cosmology and standard sterile neutrino you will see that there is a tension between the sterile neutrino solution that the lsnd likes and and cosmology right um but this is of course assuming you know land of cdm and assuming that sterile neutrino is just you know the minimal sterile neutrino we can call it right but sterile neutrinos of course has been bsm particles may come with extra bsm physics that may change the game so actually have a slide that they got from georgraphil but let me see if i find it um that kind of represents this issue right so he got this from hanistan but um you can see that if you you know if you take more cosmological data and you just stick with land of cdm you know you can get quite quite strong constraints on the sound neutrino masses and that would be quite big tension with the um with the having a one eb square sterile neutrino right but of course the issue is that you are adding data sets in different ways and then it turns out there is this other axis which is the what he calls the model space right so if you add new physics you know this this this this can be interpreted in different ways also yes if you lay land of cdm and put all the data the bounds are like around point two or so and they've got quite a big tension with the standard sterile neutrino thing but of course both cosmology and sterile and particle physics could be different and that makes this okay yeah i see there is one more more question from the q-name uh from Diego Rostrepo he says the so he says the the point limit on the number of neutrinos now points in powers three instead of four impose any constraint on the others in the anomaly yeah again this is uh again the same point right i mean uh yes cosmology largely doesn't like sterile neutrinos um so if you take that one in particular related to also the highl parameter right so it's it depends uh how strongly do you take this cosmological bounds and results right yeah but it it certainly doesn't favor it okay so i have a question maybe you said it during the talk no how how how much do you think that this is going to be the improvement on the on the bounds that you found with i don't know in 10 years of the ice to when ice to is going to be at the end i don't know how how much you expect that this bound improves so we haven't we're under the currently starting because of course as you as you accumulate statistics on my bigger analysis systematic that we are so bleeding or and not so important now become very relevant right so if this is analysis not statistically limited but systematically right and in those cases it's very it's difficult to predict exactly what we will end up with i don't have a plot and a benchmark number to give you but i do have uh let me see if i find it i do have uh this slide that tells you what we are thinking of how we can go forward and we we're going to forward in two ways one way we're doing it is we're going to do this analysis using now five years of data which we have in disk of course what i was saying when you put more statistics you know systematic like the ice become more important so you have to deal with them and then more carefully um and another way to go forward is to select better events right so so for that for example we can select only starting events and we have a better note you know energy constructs also we can actually have information not only on the track energy but the shower energy that improves our energy proxy let's say of course you pay a price because in your statistics it's reduced because you're selecting better events right so we are trying both approaches we're starting which one would be better in the end and so it's at the moment understand what will be a future but we have ideas like like this and the data is the data is certainly on this so we just have to work hard okay so anybody else have any any questions if not well we thank Carlos again and all the all the viewers of this seminar and see you next time to know the Latin American where we're next on this okay thank you bye Carlos thanks again