Search for neutrino point sources with IceCube The image shows the estimated origins of the 28 first very high energy neutrinos reported by IceCube (Nov 2013 in Science). In the image, the galactic plane is shown. If those neutrinos are produced by sources in our Galaxy, we expect to have a high correlation between them and the galactic plane. But this correlation is small. Even if we might have neutrino sources in the Milky Way, IceCube has shown that some of these neutrinos are of extragalactic origin. Search for neutrino point sources with IceCube Jacob Feintzeig WIPAC − May 21, 2014

Cosmic rays energy spectrum Origins of cosmic rays at energies above GeV? Supernova remnants Gamma-ray bursts Active galactic nuclei Cosmic rays are high-energy charged particles that constantly bombard the Earth from all directions. They are created in powerful cosmic engines and reach energies higher than any other particle in nature or in a lab. The highest-energy cosmic rays have been measured to have an energy over a million times the energy at the LHC (the particle accelerator at CERN, Geneva, Switzerland, so far the most powerful accelerator made by humans). Most cosmic rays are protons, but they also include heavier nuclei as well as electrons. Cosmic rays are charged and thus bend on their way to Earth due to the magnetic fields in the galaxy and in intergalactic space. When cosmic rays reach Earth’s atmosphere, they create a shower of particles including neutrinos and muons. The energy spectrum of cosmic rays is well studied, but their origin is a 100-year old conundrum. How and where nature can accelerate protons –most cosmic rays are protons- to those extreme energies? Cosmic rays were the first particles to be names “cosmic messengers”, since they can bring us information from the far and extreme universe, from the powerful cosmic engines where protons can be accelerated to energies around and above 1020 eV. Scientists think that cosmic objects such as supernovas, gamma ray bursts or active galactic nuclei can be the sources of cosmic rays. By learning more about cosmic rays, scientist expect to learn more about the insides of these powerful cosmic objects, the most extreme environment in our universe.

Messenger particles Cosmic rays are not the only messenger from high energy sources - a cosmic ray source is also a beam-dump. Cosmic rays inevitably interact with radiation and gas surrounding their source, e.g., Energy escaping the source is distributed among cosmic rays, gamma rays, and neutrinos. Cosmic rays are not the only cosmic messenger. Neutrinos and gamma rays can also tell us about the most extreme places in the universe. In the image, we have a schematic of a cosmic source, such as a black hole. The source accelerates protons to very extreme energies. These protons will interact with the radiation and/or gas around the source and will create secondary particles called pions. Pions are hadrons, as protons are, since they are made up of quarks. Protons, and neutrons, are made up of three quarks, while pions are made up of two quarks. Charged pions will create neutrinos, as shown in the image, while neutral pions will create gamma rays. The next three slides show imagesof possible sources (supernova remnants, gamma-ray bursts and active galactic nuclei).

supernova remnants An example of a supernova remnant.

…and if the star collapses to a black hole… gamma-ray burst happens in seconds, not thousands of years beamed along the spin axis of the black hole simulation not image Click on the yellowish top-left corner image. It’s a video of a gamma-ray burst. It’s not a real image, but a simulation.

active galaxy particle flows near supermassive black hole

Searching for neutrino sources in IceCube This is a sky map of the first 28 high-energy neutrinos ever observed. They were published by IceCube in Science magazine in November 2013. These neutrino events were found in two years of IceCube data. We know we have only a few events, but we would like to understand if the information from these neutrinos is enough to point to the first source of very high energy neutrinos ever reported. This would be a scientific breakthrough, since we have been looking for these sources for more than 100 years. Explain: What is a sky map? It shows all of the visible universe in one image. Here we are using equatorial coordinates, i.e., the center of the Earth is our central point. We could also use Galactic coordinates. Learn more about celestial coordinates here: http://en.wikipedia.org/wiki/Celestial_coordinate_system In the image, the color scale denotes how much “clustering” there is, how incompatible the observed events are compared to the background expectation. Darker colors mean more clustering, more likely to be a point source. Crosses (x) and plusses (+) show different types of neutrino events in IceCube. Plusses are cascade-like events while crosses are track-like events, more common for muon neutrino events. Learn more about neutrino events in IceCube here: http://icecube.wisc.edu/masterclass/neutrinos

A simulated sky map But look now at this sky map. Here, “fake” neutrinos have been distributed randomly in the map. How different is this map from the real one shown in the previous slide? This is a simulated “random” dataset, we know the neutrinos are not telling us about any source.

A simulated sky map with a source Now we have changed seven of the random neutrinos to seven neutrinos produced by a source. Where is the point source? Students will identify the spot with more events. How do you know? Students will explain that neutrinos here concentrate in a small region, while the others are just scattered all over the map. Why are the seven events not all coming from the exact same place? Explain: Concept of angular resolution Even though neutrinos are expected to point to the origin of their sources, our measurement of their direction is far from perfect. We call this the angular resolution of our measurement, this is the precision with which we can measure the direction of the neutrino. There is an intrinsic angular resolution due to the detector design. E.g., the spacing of the sensors within IceCube limit the angular resolution of our measurement. But also, the signal that the neutrinos leave in the detector will limit the our angular resolution. When neutrinos produce a track-like event, we can measure the direction of the incoming neutrino with less than one degree of resolution. But when the neutrino produces a cascade-like event, our angular resolution is around 15 degrees.

A simulated sky map with a source So, how can we define the signal we are looking for, i.e., a neutrino source? We can look for events clustering around one specific point. But, what is the background of this signal? How many events distributed evenly across the sky, either atmospheric muons, atmospheric neutrinos, or astrophysical neutrinos not from source, will also show up in the signal region? How can we quantitatively figure out if there is a point source in our data? Teachers lead discussion on counting number of events in a bin, or a well-defined region. Main points to get across: A simple analysis is to count events in an on-source/signal bin To estimate the background expectation we can use sky maps with data events scrambled To estimate the p-value, i.e., the significance that a number of events in a signal region is really a source, we will count the number of events in the signal bin in real data, repeat this for many scrambled sky maps, and determine the proportion of random sky maps with equal or greater numbers of events in the signal bin compared to the data. The size of the bin should be about the size of the angular resolution (we will use 15 degrees for IceCube, which is the resolution for cascade-like events)

A simulated sky map with the signal region As an example, for a given map, we will count the number of events in signal bin. In this case 1. Red region is a 15 degree circle around the galactic center – our on-source/signal region. Instructor can also illustrate the analysis more by using this map to start making a sample histogram on the board of the TS for the scrambled distribution, and demonstrate how to calculate a p-value if the data had 3 or 4 events in the region, etc.

Counting events Break up into groups, hand out scrambled maps Each group has 10 to 20 scrambled maps. They have to count how many maps have 0, 1, 2, 3, … events in the signal region.

Histogram of analysis for scrambled trials This is a template for the histogram that the class will create. Teachers can either draw this on the board, or project this slide onto the whiteboard and the students can draw on top of it. Each group will come up and mark tallies in each bin for the scrambled maps they analyzed. At the end, the teacher will use tallies to draw a standard histogram on the axes.

And here comes the real data… Have each student at their desk count the number of signal events (2) and calculate the p-value from the histogram on the board. Then, the teachers does the same for everyone to see. Teacher leads a discussion about what this result means? What does the p-value mean? Is this significant? How often we will just get at least two events in the signal region due to background events? We could introduce the concept of a “sigma,” drawing a gaussian on the board. Can we say there is a neutrino point source at the galactic center? Do you see other possible sources in this map? Discussion: What are the problems with this analysis? There are some events near the signal region but not in it. Should these be included? (flip to next slide, where each neutrino is shown with the angular uncertainty curve around it) What if you moved the signal region to the left? You could get 5 events! How come we didn’t do that? We can explain the concept of introducing a bias in our measurement. In science, we are not allowed to change the way we look at our data once we know what the data looks like. If we do so, we might introduce a bias in our results, this is where we can find something just because we thought that the data was showing something. We had decided to look at the center of the galaxy since we thought this was a region where we should get more neutrinos. We cannot afterward just search exactly in a place where neutrinos seem to accumulate. What if there was a track nearby? If you have 1 degree angular resolution, you don’t need a 15 degree signal region.

And here comes the real data… Same map, with angular uncertainties, to aid discussion. Here each neutrino is shown with a curve around it that shows the uncertainty in the measurement of their direction. Crosses, track-like events, do not show it since the precision with which IceCube measures the directions of these neutrinos is very high. Plusses, cascade-like events, show the real precision (angular resolution) for each neutrino…on average it is 15 degrees, but depending on the shape and location of the event in our detector, we will measure the direction with a poorer resolution.

Point source analysis in the Science paper But, what if there were many events clustering around other spots in the sky? Discussion: What would a better analysis include? (Classroom-wide brainstorming session, can write these on the board) -Knowledge of the shape/size of the angular uncertainties for each event -Some sort of adjustable binning/unbinned method to avoid events falling in/out of bins -Some sort of scan across the sky to look for point sources everywhere In the real IceCube analysis, researchers calculated a p-value for every point of the sky, taking into account the resolution of each event. The reason why they did that is that neutrinos might be produced in the galaxy, but they can also be produced anywhere outside our galaxy. The most significant cluster, one with five events, had a final significance of 8%, far from what would allow IceCube to identify a source. Explain the map and how the real measurement was made: -Color scale denotes how much “clustering” there is, how incompatible the observed events are compared to the background expectation -Darker colors mean more clustering, more likely to be a point source -Analysis knows about size/shape of angular resolution for each event – track-like events have very narrow hotspots, galactic center cluster is much wider -Analysis has no bins – near galactic center, all five events are contributing, according to how close/far away they are and how big their angular uncertainties are -Analysis looks over the entire sky -P-value calculated in same way: analysis repeated on scrambled datasets -P-value for galactic center alone: 5% Discussion. What next? Wait for more data Look for correlations with data from other experiments Improve the angular resolution for cascade-like events.

Point source results with 3 years of data Previous map showed neutrino events found in two years of IceCube data. This map shows 9 more neutrino events found in one more year of IceCube data.. Discuss locations of new events. Do we think there’s a point source at the galactic center? Do we think there’s a source anywhere else? Are they clustering around the galactic plane? There is still not a significant result, but the maps seems to show that there are two contributions of astrophysical neutrinos: one of galactic origin and one of extragalactic origin. We are waiting for more data to improve these results.