[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:00.00,0:00:13.12,Default,,0000,0000,0000,,{\i1}music{\i0}
Dialogue: 0,0:00:13.12,0:00:17.19,Default,,0000,0000,0000,,Herald: Good morning and welcome back to\Nstage one. It's kind of going to be the
Dialogue: 0,0:00:17.19,0:00:21.49,Default,,0000,0000,0000,,second talk about physics on this day\Nalready and it's about big data and
Dialogue: 0,0:00:21.49,0:00:27.15,Default,,0000,0000,0000,,science and big data became something like\NUber in science. It's everywhere every
Dialogue: 0,0:00:27.15,0:00:33.37,Default,,0000,0000,0000,,discipline has it. Axel Naumann's working\Nfor CERN, the accelerator in Switzerland
Dialogue: 0,0:00:33.37,0:00:39.16,Default,,0000,0000,0000,,and he talks about how physics and\Ncomputing bridge in this area and he works
Dialogue: 0,0:00:39.16,0:00:43.18,Default,,0000,0000,0000,,a lot with ROOT, a program that helps\Ntransform data into knowledge. A warm
Dialogue: 0,0:00:43.18,0:00:44.65,Default,,0000,0000,0000,,welcome.
Dialogue: 0,0:00:44.65,0:00:45.26,Default,,0000,0000,0000,,Axel Naumann: Thank you.
Dialogue: 0,0:00:45.26,0:00:51.26,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:00:51.26,0:00:57.85,Default,,0000,0000,0000,,AN: Thanks a lot. So, well you know, when,\Nwhen I was discussing this abstract with
Dialogue: 0,0:00:57.85,0:01:00.95,Default,,0000,0000,0000,,the science track people they tell me:\N"Well, you know about three hundred people
Dialogue: 0,0:01:00.95,0:01:06.00,Default,,0000,0000,0000,,might be in the audience." But well, hey,\Nyou are huge that's much more than three
Dialogue: 0,0:01:06.00,0:01:10.94,Default,,0000,0000,0000,,hundred people. So thank you so much for\Ninviting me over it's a real honor. And of
Dialogue: 0,0:01:10.94,0:01:15.31,Default,,0000,0000,0000,,course originally when talking to 300\Npeople are all science interested I
Dialogue: 0,0:01:15.31,0:01:20.59,Default,,0000,0000,0000,,thought you know I pick something fairly\Nnarrow focuswise but then I learned I'm
Dialogue: 0,0:01:20.59,0:01:24.69,Default,,0000,0000,0000,,going to be in Saal one and that's\Ndifferent, so I decided to make the scope
Dialogue: 0,0:01:24.69,0:01:30.67,Default,,0000,0000,0000,,a little bit wider and that's what I ended\Nup with. I'll talk a little bit about
Dialogue: 0,0:01:30.67,0:01:37.54,Default,,0000,0000,0000,,CERN in society as well if you so choose,\Nyou'll see what that means in a minute. So
Dialogue: 0,0:01:37.54,0:01:41.68,Default,,0000,0000,0000,,the things I'll cover here is obviously\NCERN just a little bit of an introduction
Dialogue: 0,0:01:41.68,0:01:46.10,Default,,0000,0000,0000,,how we do physics, how we do computing,\Nwhat data means to us and I can tell you
Dialogue: 0,0:01:46.10,0:01:51.81,Default,,0000,0000,0000,,it means everything, you heard about that\Nalready, right? How we do data analysis in
Dialogue: 0,0:01:51.81,0:01:56.16,Default,,0000,0000,0000,,high energy physics and just because\Nwe've been doing it for a while and
Dialogue: 0,0:01:56.16,0:02:00.53,Default,,0000,0000,0000,,because I've been doing it for more than\Nten years, I'm one of the guys who's
Dialogue: 0,0:02:00.53,0:02:07.25,Default,,0000,0000,0000,,providing the software to do data\Nanalysis in high energy physics, so, you
Dialogue: 0,0:02:07.25,0:02:11.36,Default,,0000,0000,0000,,know, because we know what we are doing\Nand we have some experience, I thought
Dialogue: 0,0:02:11.36,0:02:18.11,Default,,0000,0000,0000,,maybe you might be interested in hearing\Nwhat my forecast is for data analysis in
Dialogue: 0,0:02:18.11,0:02:25.43,Default,,0000,0000,0000,,general, in the future. So let's start\Nwith CERN. And so if you wonder what CERN
Dialogue: 0,0:02:25.43,0:02:31.51,Default,,0000,0000,0000,,is, you've all heard about CERN, about\Nthe fantastic funds we love to use, then
Dialogue: 0,0:02:31.51,0:02:36.96,Default,,0000,0000,0000,,you've probably also heard that we are\Ndoing science. We were founded right after
Dialogue: 0,0:02:36.96,0:02:41.45,Default,,0000,0000,0000,,the Second World War or soon after the\NSecond World War, basically as a way to
Dialogue: 0,0:02:41.45,0:02:47.46,Default,,0000,0000,0000,,entertain those freaky scientists. You\Nknow that was the idea: peace europewide.
Dialogue: 0,0:02:47.46,0:02:52.35,Default,,0000,0000,0000,,And damn, that's working out really well\Nand so well there's not just Europe
Dialogue: 0,0:02:52.35,0:02:57.53,Default,,0000,0000,0000,,anymore these days. We are located near\NGeneva, we are doing only fundamental
Dialogue: 0,0:02:57.53,0:03:02.27,Default,,0000,0000,0000,,research, so we don't do any weapons,\Nnuclear stuff you
Dialogue: 0,0:03:02.27,0:03:10.23,Default,,0000,0000,0000,,know, these kind of things. The WWW was\Ninvented at CERN but that was just a, you
Dialogue: 0,0:03:10.23,0:03:14.59,Default,,0000,0000,0000,,know, side effect happens sometimes, that\Nwe invent things. But usually we just do
Dialogue: 0,0:03:14.59,0:03:22.50,Default,,0000,0000,0000,,science. So what we do is, we take money,\Nlots off, and brains who like to discuss
Dialogue: 0,0:03:22.50,0:03:27.21,Default,,0000,0000,0000,,and think and come up with ideas and from\Nthat we generate knowledge. It's really
Dialogue: 0,0:03:27.21,0:03:33.00,Default,,0000,0000,0000,,all about curiosity. The things we try to\Nanswer is what is mass? Which is funny
Dialogue: 0,0:03:33.00,0:03:37.37,Default,,0000,0000,0000,,question right? Like we all know what mass\Nis but actually we don't. We know what
Dialogue: 0,0:03:37.37,0:03:42.36,Default,,0000,0000,0000,,mass is in the universe. We understand\Nthat masses attract one another: gravity.
Dialogue: 0,0:03:42.36,0:03:48.73,Default,,0000,0000,0000,,Which is beautifully correct. And in the\Nsmall scale, our particles, we know that
Dialogue: 0,0:03:48.73,0:03:52.94,Default,,0000,0000,0000,,mass is energy and we can't convert them.\NBut we don't understand how these two
Dialogue: 0,0:03:52.94,0:03:58.32,Default,,0000,0000,0000,,things go together. Like there is no\Nbridge, they contradict one another. So we
Dialogue: 0,0:03:58.32,0:04:04.93,Default,,0000,0000,0000,,are trying to understand what that bridge\Nmight be. Part of that mass thing is of
Dialogue: 0,0:04:04.93,0:04:08.65,Default,,0000,0000,0000,,course also what's out there in the\Nuniverse? That's a big question. We only
Dialogue: 0,0:04:08.65,0:04:14.23,Default,,0000,0000,0000,,understand a few percent of that. 90 and\Nsome percent are completely unknown to
Dialogue: 0,0:04:14.23,0:04:20.35,Default,,0000,0000,0000,,us, and that's scary right? I mean we know\Ngravity really well, we can deal with
Dialogue: 0,0:04:20.35,0:04:27.56,Default,,0000,0000,0000,,freaky things like black holes and yet we\Ndon't understand what's out there. Now to
Dialogue: 0,0:04:27.56,0:04:31.85,Default,,0000,0000,0000,,do all these things we are probing nature\Nat the smallest scale as we call it, so
Dialogue: 0,0:04:31.85,0:04:36.19,Default,,0000,0000,0000,,that's particles, we are dealing with\Nthings like the Higgs particle and
Dialogue: 0,0:04:36.19,0:04:43.90,Default,,0000,0000,0000,,supersymmetry. Here's a little bit of a\Nfact sheet. We have about 12,000
Dialogue: 0,0:04:43.90,0:04:47.50,Default,,0000,0000,0000,,physicists who are working with CERN. We\Nare basically the workbench that you saw
Dialogue: 0,0:04:47.50,0:04:54.66,Default,,0000,0000,0000,,in Andre's talk before. We are the table\Nthat physicists use, okay? And, so they
Dialogue: 0,0:04:54.66,0:04:59.05,Default,,0000,0000,0000,,come to CERN and once a while about\N10,000 physicists a year, or they work
Dialogue: 0,0:04:59.05,0:05:02.81,Default,,0000,0000,0000,,remotely most of the time from about 120\Nnations. So you're seeing it's not
Dialogue: 0,0:05:02.81,0:05:10.65,Default,,0000,0000,0000,,European anymore, this is a global thing.\NCERN in itself has about 2,500 employees,
Dialogue: 0,0:05:10.65,0:05:15.49,Default,,0000,0000,0000,,you know those scrubbing the table,\Nsetting things up and so on. And our
Dialogue: 0,0:05:15.49,0:05:21.19,Default,,0000,0000,0000,,table is right here. In the far end we\Nhave the Alps, it's in Switzerland
Dialogue: 0,0:05:21.19,0:05:25.99,Default,,0000,0000,0000,,as I said, so the Alps are\Nalways close, with Mont Blanc, we have the
Dialogue: 0,0:05:25.99,0:05:31.64,Default,,0000,0000,0000,,Lake Geneva we have the Jura, the French\NMountains on the lower end here, it's just
Dialogue: 0,0:05:31.64,0:05:37.41,Default,,0000,0000,0000,,beautiful. It's really nice, but we\Nneeded to stick a 30-kilometer ring in
Dialogue: 0,0:05:37.41,0:05:43.86,Default,,0000,0000,0000,,there somewhere and people would have\Nhated us had we put it like this. But
Dialogue: 0,0:05:43.86,0:05:49.67,Default,,0000,0000,0000,,luckily people were smart back then in the\N70s, and built a tunnel much better. So
Dialogue: 0,0:05:49.67,0:05:55.23,Default,,0000,0000,0000,,now we have this huge tunnel, and we send\Nparticles through in both directions near
Dialogue: 0,0:05:55.23,0:06:00.35,Default,,0000,0000,0000,,the speed of light and the tunnel is\Nfilled with magnets simply because if you
Dialogue: 0,0:06:00.35,0:06:08.11,Default,,0000,0000,0000,,don't use a magnet the particles will fly\Nstraight but we need them to turn around.
Dialogue: 0,0:06:08.11,0:06:13.56,Default,,0000,0000,0000,,Here you see what it's looking like, you\Nalso see these big halls there that have
Dialogue: 0,0:06:13.56,0:06:21.88,Default,,0000,0000,0000,,access shafts from the top and that's\Nwhere the experiments are. That's sort of
Dialogue: 0,0:06:21.88,0:06:29.21,Default,,0000,0000,0000,,a sketch of one of the experiments. So the\Nthe LHC is one of the, no, is the biggest
Dialogue: 0,0:06:29.21,0:06:35.89,Default,,0000,0000,0000,,particle accelerator at the moment, it's a\Nring with 27 kilometers circumference, 100
Dialogue: 0,0:06:35.89,0:06:40.30,Default,,0000,0000,0000,,meters below Switzerland and France, it\Nhas four big experiments and several
Dialogue: 0,0:06:40.30,0:06:45.27,Default,,0000,0000,0000,,small ones and we are expected to run\Nuntil 2030. So you see that all of that
Dialogue: 0,0:06:45.27,0:06:50.15,Default,,0000,0000,0000,,is large-scale simply because we're trying\Nto make good use of the money we have.
Dialogue: 0,0:06:50.15,0:06:56.02,Default,,0000,0000,0000,,Here, you see one of these caverns that\Nare used by the experiments while it was
Dialogue: 0,0:06:56.02,0:07:01.49,Default,,0000,0000,0000,,empty. The experiment was then lowered\Nthrough this hole by the roof, piece by
Dialogue: 0,0:07:01.49,0:07:07.19,Default,,0000,0000,0000,,piece, and these things are humongous. To\Ngive you an impression of how big it is, I
Dialogue: 0,0:07:07.19,0:07:12.52,Default,,0000,0000,0000,,put Waldo in there, so your job for the\Nnext three slides is to find Waldo. You
Dialogue: 0,0:07:12.52,0:07:15.80,Default,,0000,0000,0000,,know, that gives you the scale. He's\Nfriendlily waving at you, so it should be
Dialogue: 0,0:07:15.80,0:07:21.99,Default,,0000,0000,0000,,easy to find him. So then we put a\Ndetector in there. Here it's pulled apart
Dialogue: 0,0:07:21.99,0:07:26.16,Default,,0000,0000,0000,,a little bit, so it looks nicer, you can\Nactually see something. You can for
Dialogue: 0,0:07:26.16,0:07:31.04,Default,,0000,0000,0000,,example see the beam pipe, so that's where\Nthe particles are flying through, and then
Dialogue: 0,0:07:31.04,0:07:34.88,Default,,0000,0000,0000,,they're coming from both directions and\Ncolliding in the center of the detector
Dialogue: 0,0:07:34.88,0:07:38.49,Default,,0000,0000,0000,,and then things happen we try to\Nunderstand what
Dialogue: 0,0:07:38.49,0:07:44.79,Default,,0000,0000,0000,,is happening. That's yet another view,\Nfrontal view on one of the detectors and
Dialogue: 0,0:07:44.79,0:07:51.06,Default,,0000,0000,0000,,now you have to imagine that, you know,\Nyou can't just open up Amazon and order an
Dialogue: 0,0:07:51.06,0:07:56.21,Default,,0000,0000,0000,,LHC experiment, right, that's not how it\Nworks. We do this stuff ourselves, like
Dialogue: 0,0:07:56.21,0:08:02.67,Default,,0000,0000,0000,,PhD students, postdocs, engineers. You\Nknow, that's all done by hand, just like
Dialogue: 0,0:08:02.67,0:08:06.94,Default,,0000,0000,0000,,the microscope you saw before. Of course\Nyou order the parts, but you know the
Dialogue: 0,0:08:06.94,0:08:11.06,Default,,0000,0000,0000,,design, the whole conception and actually\Nscrewing these things together, making
Dialogue: 0,0:08:11.06,0:08:16.97,Default,,0000,0000,0000,,sure that all fits, is all done by hand.\NAnd I find that just beautiful, I mean
Dialogue: 0,0:08:16.97,0:08:21.76,Default,,0000,0000,0000,,that's close to a miracle, right? That\Nnations, like people no matter what
Dialogue: 0,0:08:21.76,0:08:26.82,Default,,0000,0000,0000,,nation, people across the globe work\Ntogether to build such a huge thing and
Dialogue: 0,0:08:26.82,0:08:39.49,Default,,0000,0000,0000,,then you turn it on and it works. More or\Nless, but you get it to work. That's not
Dialogue: 0,0:08:39.49,0:08:44.31,Default,,0000,0000,0000,,my applause, that's your applause, because\Nyou make this possible. Really, but it's,
Dialogue: 0,0:08:44.31,0:08:49.69,Default,,0000,0000,0000,,it's huge this is for me one of the things\NI love most about CERN: That is this
Dialogue: 0,0:08:49.69,0:08:55.28,Default,,0000,0000,0000,,international thing that just works\Nsmoothly. Now the detectors are like a
Dialogue: 0,0:08:55.28,0:09:01.31,Default,,0000,0000,0000,,massive camera. We have lots of pixels and\Nwe take many, many pictures a second. We
Dialogue: 0,0:09:01.31,0:09:06.68,Default,,0000,0000,0000,,do this to identify particles and then\Nsort of estimate what has happened during
Dialogue: 0,0:09:06.68,0:09:15.47,Default,,0000,0000,0000,,the collision. Now, life at CERN is of\Ncourse an important ingredient for
Dialogue: 0,0:09:15.47,0:09:19.53,Default,,0000,0000,0000,,scientists as well, and if you live at\NCERN then actually it's just work at CERN
Dialogue: 0,0:09:19.53,0:09:23.98,Default,,0000,0000,0000,,and that's what it's about. But it's not\Nthat bad, so we hang out together in our
Dialogue: 0,0:09:23.98,0:09:30.04,Default,,0000,0000,0000,,control rooms, make sure that the\Nexperiments work correctly. We also, you
Dialogue: 0,0:09:30.04,0:09:33.72,Default,,0000,0000,0000,,know, study the forces.\N{\i1}laughter{\i0}
Dialogue: 0,0:09:33.72,0:09:38.74,Default,,0000,0000,0000,,We have scientific discourse, in the sun,\Nview on the Mont Blanc, with a good
Dialogue: 0,0:09:38.74,0:09:45.43,Default,,0000,0000,0000,,coffee. We have lectures and we are\Nlectured and of course, as you, we have
Dialogue: 0,0:09:45.43,0:09:54.57,Default,,0000,0000,0000,,more laptops than people. And, then we do\Nstuff and so this presentation is going to
Dialogue: 0,0:09:54.57,0:09:58.58,Default,,0000,0000,0000,,introduce you to some of the things we are\Ndoing, and more on the computing and the
Dialogue: 0,0:09:58.58,0:10:04.10,Default,,0000,0000,0000,,society side as I said. But because I have\Nso much to talk to about I decided that
Dialogue: 0,0:10:04.10,0:10:08.81,Default,,0000,0000,0000,,you just build your own talk, you tell me\Nwhat you want to hear. So let's do this,
Dialogue: 0,0:10:08.81,0:10:14.41,Default,,0000,0000,0000,,you can choose between A, physics, and B,\Nmodel simulation and data. You remember
Dialogue: 0,0:10:14.41,0:10:18.62,Default,,0000,0000,0000,,these books like from the old days when we\Nwere all young? It's that kind of thing,
Dialogue: 0,0:10:18.62,0:10:24.45,Default,,0000,0000,0000,,ok? You decide/design your own talk here.\NSo, by applause, do you want to hear about
Dialogue: 0,0:10:24.45,0:10:27.72,Default,,0000,0000,0000,,physics?\N{\i1}applause{\i0}
Dialogue: 0,0:10:27.72,0:10:35.73,Default,,0000,0000,0000,,Okay. Or the model simulation data part?\N{\i1}louder applause{\i0}
Dialogue: 0,0:10:35.73,0:10:45.10,Default,,0000,0000,0000,,Okay, there we go. So, this is what we\Nskip. Model simulation data it is. You're
Dialogue: 0,0:10:45.10,0:10:49.70,Default,,0000,0000,0000,,a strange crowd, first time I meet people\Nwho don't want to hear about physics... no
Dialogue: 0,0:10:49.70,0:10:51.45,Default,,0000,0000,0000,,I'm kidding.\N{\i1}laughter{\i0}
Dialogue: 0,0:10:51.45,0:10:53.80,Default,,0000,0000,0000,,Audience: {\i1}inaudible interjection{\i0}\N{\i1}laughter{\i0}
Dialogue: 0,0:10:53.80,0:11:00.08,Default,,0000,0000,0000,,So model simulation data it is. So our\Ntheory is actually incredibly precise.
Dialogue: 0,0:11:00.08,0:11:04.45,Default,,0000,0000,0000,,It's so precise that our basic job is\Nreally really boring, because we already
Dialogue: 0,0:11:04.45,0:11:10.51,Default,,0000,0000,0000,,understand everything. Whenever there is a\Ncollision, we know what's going to happen.
Dialogue: 0,0:11:10.51,0:11:15.43,Default,,0000,0000,0000,,Except for these very rare things. So we\Nare trying to find these very rare things
Dialogue: 0,0:11:15.43,0:11:19.58,Default,,0000,0000,0000,,out of this haystack of fairly boring\Nthings that we really understand well. And
Dialogue: 0,0:11:19.58,0:11:25.59,Default,,0000,0000,0000,,the weird things are, for example,\Nmonopoles, supersymmetry, or black holes.
Dialogue: 0,0:11:25.59,0:11:32.06,Default,,0000,0000,0000,,Now the theorists job is to tell us what\Nwe should be seeing in the detector, given
Dialogue: 0,0:11:32.06,0:11:42.35,Default,,0000,0000,0000,,some fancy physics. Then we use simulation\Nto see how our detector would respond to
Dialogue: 0,0:11:42.35,0:11:53.48,Default,,0000,0000,0000,,that. Now, of course the question is: We\Nare just counting, basically, when we do
Dialogue: 0,0:11:53.48,0:11:58.10,Default,,0000,0000,0000,,experiments and the question is: How often\Ndo we need to see something to say: "Well,
Dialogue: 0,0:11:58.10,0:12:03.31,Default,,0000,0000,0000,,that's not just the ordinary. That is\Nsomething new, that's something that could
Dialogue: 0,0:12:03.31,0:12:09.87,Default,,0000,0000,0000,,be explained by a weird theory. We use the\Ndetector simulation as I said to basically
Dialogue: 0,0:12:09.87,0:12:15.03,Default,,0000,0000,0000,,predict how much we expect to see things.\NWe use reconstruction software which
Dialogue: 0,0:12:15.03,0:12:20.68,Default,,0000,0000,0000,,tells us what has happened, or might have\Nhappened in the detector to count how
Dialogue: 0,0:12:20.68,0:12:25.40,Default,,0000,0000,0000,,often we saw something. And then we use\Nstatistics to compare these two and to say
Dialogue: 0,0:12:25.40,0:12:31.61,Default,,0000,0000,0000,,whether something is expected or not. Now,\Nthat's fairly abstract but it's fairly
Dialogue: 0,0:12:31.61,0:12:36.90,Default,,0000,0000,0000,,common, a fairly common approach. For\Nexample, if you look at climate versus
Dialogue: 0,0:12:36.90,0:12:40.33,Default,,0000,0000,0000,,weather, right, I mean we always have\Ntemperature fluctuations because of
Dialogue: 0,0:12:40.33,0:12:46.48,Default,,0000,0000,0000,,weather, and the question is: Is that rise\Nin temperature because of a weather effect
Dialogue: 0,0:12:46.48,0:12:50.38,Default,,0000,0000,0000,,or because of a climate effect? Is that\Nlarge-scale or just a short-term
Dialogue: 0,0:12:50.38,0:12:55.61,Default,,0000,0000,0000,,fluctuation. So there, we have a very\Nsimilar problem and here what you do is
Dialogue: 0,0:12:55.61,0:13:00.88,Default,,0000,0000,0000,,you measure temperatures, and you want to\Ndetect abnormal variations, and you can
Dialogue: 0,0:13:00.88,0:13:06.42,Default,,0000,0000,0000,,improve that by measuring longer, like,\Nfor 300 years instead of 20 years. That
Dialogue: 0,0:13:06.42,0:13:11.93,Default,,0000,0000,0000,,gives you a better prediction what you\Nwould expect in the future. Also, larger
Dialogue: 0,0:13:11.93,0:13:14.17,Default,,0000,0000,0000,,deviations help, right?. If you look for\Nsomething that
Dialogue: 0,0:13:14.17,0:13:19.70,Default,,0000,0000,0000,,is just 0.1 degree, then you might not be\Nable to find it. If there is a deviation
Dialogue: 0,0:13:19.70,0:13:25.23,Default,,0000,0000,0000,,of 5 degrees, you will definitely find it.\NAnd for us it's very similar. So here we
Dialogue: 0,0:13:25.23,0:13:31.61,Default,,0000,0000,0000,,have a plot, one of the first Higgs\Ndiscovery plots, and you can see that we
Dialogue: 0,0:13:31.61,0:13:38.80,Default,,0000,0000,0000,,have many ingredients there. So, the black\Ndots are what we measure and they have
Dialogue: 0,0:13:38.80,0:13:43.83,Default,,0000,0000,0000,,certain uncertainty, because when we\Nmeasure, we count and we might have, you
Dialogue: 0,0:13:43.83,0:13:48.98,Default,,0000,0000,0000,,know, not seen something, or we might have\Nseen more than we we should have seen, so
Dialogue: 0,0:13:48.98,0:13:54.97,Default,,0000,0000,0000,,there's always an uncertainty. And then we\Nalso have theory, which tells us you
Dialogue: 0,0:13:54.97,0:14:00.08,Default,,0000,0000,0000,,should have seen so many and so for the\Nred part that's something that we know
Dialogue: 0,0:14:00.08,0:14:04.89,Default,,0000,0000,0000,,exists, it's nothing spectacular. It's\Nsimply what theory is telling us what we
Dialogue: 0,0:14:04.89,0:14:10.66,Default,,0000,0000,0000,,should be seeing. And you can see the data\Nfollows the red part fairly well. But then
Dialogue: 0,0:14:10.66,0:14:15.98,Default,,0000,0000,0000,,there is this other bump in our dots on\Nthe right-hand side or in the center and
Dialogue: 0,0:14:15.98,0:14:21.23,Default,,0000,0000,0000,,that does not make sense, unless you take\Nthe Higgs into account, right, which is
Dialogue: 0,0:14:21.23,0:14:26.89,Default,,0000,0000,0000,,the light blue part and so here you can\Nsee how this interplay between different
Dialogue: 0,0:14:26.89,0:14:38.28,Default,,0000,0000,0000,,sources of physics and statistics works\Nfor us. Now just as for the climate, more
Dialogue: 0,0:14:38.28,0:14:43.69,Default,,0000,0000,0000,,data helps. And there are two versions of\Nmore data more data: Either by having more
Dialogue: 0,0:14:43.69,0:14:48.08,Default,,0000,0000,0000,,collisions, which is why we are running\N24/7, or more data by combining different
Dialogue: 0,0:14:48.08,0:14:52.06,Default,,0000,0000,0000,,analyses which is what's happening here.\NSo here you see all these different
Dialogue: 0,0:14:52.06,0:14:56.99,Default,,0000,0000,0000,,analyses. If you combine them, of course\Nyou get a much stronger prediction of, in
Dialogue: 0,0:14:56.99,0:15:03.30,Default,,0000,0000,0000,,this case, the Higgs mass, then if you\Njust take any single one of them. You see
Dialogue: 0,0:15:03.30,0:15:08.54,Default,,0000,0000,0000,,how similar what we are doing is to, you\Nknow, any of the big data analyses out
Dialogue: 0,0:15:08.54,0:15:16.41,Default,,0000,0000,0000,,there. Okay, so that was that part. Now\Ncomes the obligatory part again,
Dialogue: 0,0:15:16.41,0:15:22.93,Default,,0000,0000,0000,,computering. When we were designing the\NLHC,not me, when people were designing the
Dialogue: 0,0:15:22.93,0:15:31.12,Default,,0000,0000,0000,,LHC, they needed to project computing\Npower from 1990 to 2000 2010 and so on.
Dialogue: 0,0:15:31.12,0:15:34.14,Default,,0000,0000,0000,,And then they said: "Well, we need\Nmassive amount of computers" and for you
Dialogue: 0,0:15:34.14,0:15:38.42,Default,,0000,0000,0000,,there's now "Ughhh - everybody has it, we\Nhave it as well, we have our racks of
Dialogue: 0,0:15:38.42,0:15:44.24,Default,,0000,0000,0000,,computers". This is something that the big\Ncompanies usually don't show: You you know
Dialogue: 0,0:15:44.24,0:15:48.51,Default,,0000,0000,0000,,there is actually a ramp where the trucks\Narrive and they offload the things and
Dialogue: 0,0:15:48.51,0:15:53.82,Default,,0000,0000,0000,,then someone needs to screw them together\Nand then looks shiny. This is how we are
Dialogue: 0,0:15:53.82,0:16:00.87,Default,,0000,0000,0000,,spending our CPU time: We have about\N60,000 cores that are spinning all the
Dialogue: 0,0:16:00.87,0:16:06.68,Default,,0000,0000,0000,,time for us, and they are distributed\Naround the world. You can see that CERN,
Dialogue: 0,0:16:06.68,0:16:14.53,Default,,0000,0000,0000,,for example, is the red part there near\Nthe bottom. Yeah, so we make good use of
Dialogue: 0,0:16:14.53,0:16:20.83,Default,,0000,0000,0000,,that. We also monitor the efficiency, and\Nbecause 100 percent efficient is for
Dialogue: 0,0:16:20.83,0:16:29.30,Default,,0000,0000,0000,,beginners we are actually about 700\Npercent efficient. Don't ask why. They
Dialogue: 0,0:16:29.30,0:16:33.92,Default,,0000,0000,0000,,decided if you are multi-threading, then\Nwe, you know, we multiply your efficiency
Dialogue: 0,0:16:33.92,0:16:39.95,Default,,0000,0000,0000,,by the number of threads you have. Makes\Nno sense to me. We also have storage,
Dialogue: 0,0:16:39.95,0:16:44.93,Default,,0000,0000,0000,,currently we use about 0.7 exabytes. We\Nalso have available at one point seven
Dialogue: 0,0:16:44.93,0:16:49.13,Default,,0000,0000,0000,,exabytes, so that's good, we make use of\Nthe storage we have. Where it's, you know,
Dialogue: 0,0:16:49.13,0:16:55.53,Default,,0000,0000,0000,,tera- peta- exa-, so it's a lot, and here\Nyou can see on the right hand side you
Dialogue: 0,0:16:55.53,0:16:59.61,Default,,0000,0000,0000,,see, for example, the tape usage on the\Nbottom and you see this dip that was
Dialogue: 0,0:16:59.61,0:17:04.27,Default,,0000,0000,0000,,before we were starting the accelerator\Nagain, we needed to make some space so we
Dialogue: 0,0:17:04.27,0:17:09.09,Default,,0000,0000,0000,,monitor our hard disk usage all the time.\NHey, here comes the next decision point:
Dialogue: 0,0:17:09.09,0:17:13.63,Default,,0000,0000,0000,,So, do you want to hear about, 1,\Ndistributed computing or 2, measure
Dialogue: 0,0:17:13.63,0:17:17.84,Default,,0000,0000,0000,,effects of bugs. So, 1, distributed\Ncomputing
Dialogue: 0,0:17:17.84,0:17:26.47,Default,,0000,0000,0000,,{\i1}applause{\i0}\Nand 2, measure the effects of bugs
Dialogue: 0,0:17:26.47,0:17:35.56,Default,,0000,0000,0000,,{\i1}similar amount of applause{\i0}\NOkay, so that's my call, and I would say
Dialogue: 0,0:17:35.56,0:17:41.46,Default,,0000,0000,0000,,we do we do... Measure the effects of\Nbugs, because it's shorter.
Dialogue: 0,0:17:41.46,0:17:47.13,Default,,0000,0000,0000,,{\i1}laughter{\i0}\NSo this is one of the views you can, you
Dialogue: 0,0:17:47.13,0:17:50.74,Default,,0000,0000,0000,,know, electronic views you can get from a\Ndetector and you see how we trace the
Dialogue: 0,0:17:50.74,0:17:55.38,Default,,0000,0000,0000,,particles that fly through the detector.\NNow, that software right, that's the
Dialogue: 0,0:17:55.38,0:17:59.93,Default,,0000,0000,0000,,result of software, and you might not\Nbelieve it, if you have bugs in there, in
Dialogue: 0,0:17:59.93,0:18:00.81,Default,,0000,0000,0000,,that software.
Dialogue: 0,0:18:02.85,0:18:07.26,Default,,0000,0000,0000,,And you know, these bugs are sometimes\Nwrong coordinate transformations, so
Dialogue: 0,0:18:07.26,0:18:12.59,Default,,0000,0000,0000,,things don't go this way but that way,\Nit's kind of weird if you look at it, and
Dialogue: 0,0:18:12.59,0:18:17.47,Default,,0000,0000,0000,,the result is that our particles don't go\Nthrough the path that they should have
Dialogue: 0,0:18:17.47,0:18:25.19,Default,,0000,0000,0000,,been going, but we are attributing them a\Ndifferent path. Now, the the nice thing
Dialogue: 0,0:18:25.19,0:18:30.96,Default,,0000,0000,0000,,is that we are doing this a million times,\Nright? So all of that is smeared. We are
Dialogue: 0,0:18:30.96,0:18:35.73,Default,,0000,0000,0000,,not systematically doing this wrong it's\Njust, we are always doing it a little bit
Dialogue: 0,0:18:35.73,0:18:41.67,Default,,0000,0000,0000,,wrong. And so the net result is that if we\Nmeasure our particles, we will not measure
Dialogue: 0,0:18:41.67,0:18:46.86,Default,,0000,0000,0000,,the right thing but always a little bit\Nwobbly left wobbly right you know? Things
Dialogue: 0,0:18:46.86,0:18:53.81,Default,,0000,0000,0000,,are not as precise. That's simply an\Nuncertainty. So for us just like counting
Dialogue: 0,0:18:53.81,0:18:59.06,Default,,0000,0000,0000,,has an uncertainty and predictions have\Nan uncertainty, software bugs introduced
Dialogue: 0,0:18:59.06,0:19:05.56,Default,,0000,0000,0000,,another source of uncertainties. And here\Nyou can see how we are tracking
Dialogue: 0,0:19:05.56,0:19:09.37,Default,,0000,0000,0000,,uncertainties for for all of our\Nanalyses. We are trying to understand the
Dialogue: 0,0:19:09.37,0:19:16.22,Default,,0000,0000,0000,,different forces of uncertainties. And\Nagain, bugs are only one of the sources
Dialogue: 0,0:19:16.22,0:19:22.88,Default,,0000,0000,0000,,here, so if we find the bug then we\Nreduce our uncertainty and we can find new
Dialogue: 0,0:19:22.88,0:19:27.76,Default,,0000,0000,0000,,physics earlier, instead of having to\Nwait and collect more data. So for us
Dialogue: 0,0:19:27.76,0:19:32.21,Default,,0000,0000,0000,,finding bugs is really key, we really\Nlove finding bugs because it brings
Dialogue: 0,0:19:32.21,0:19:36.71,Default,,0000,0000,0000,,physics closer. I thought that was\Ninteresting. It's kind of rare that you're
Dialogue: 0,0:19:36.71,0:19:42.14,Default,,0000,0000,0000,,in environment where you're able to\Nmeasure the effect of bugs. Okay, so now
Dialogue: 0,0:19:42.14,0:19:47.87,Default,,0000,0000,0000,,we are talking, we'll be talking about\Ndata. I talked, told you that we are
Dialogue: 0,0:19:47.87,0:19:52.69,Default,,0000,0000,0000,,trying to find particle traces in our\Ndata and the way we do this is by using
Dialogue: 0,0:19:52.69,0:19:56.70,Default,,0000,0000,0000,,reconstruction programs and there are\Nmultiple gigabytes of binaries in shared
Dialogue: 0,0:19:56.70,0:20:01.80,Default,,0000,0000,0000,,libraries and stuff. They're huge, they're\Nexperiment specific and they are curated
Dialogue: 0,0:20:01.80,0:20:06.27,Default,,0000,0000,0000,,by the experiments, open-source for some\Nof them, and we want them to be correct
Dialogue: 0,0:20:06.27,0:20:14.14,Default,,0000,0000,0000,,and efficient. The data format we use is\Nnot comma separated values, it's binary
Dialogue: 0,0:20:14.14,0:20:21.08,Default,,0000,0000,0000,,and for some strange reason it's our own\Ncustom binary format. The reason is that
Dialogue: 0,0:20:21.08,0:20:26.99,Default,,0000,0000,0000,,it's really targeted and the kind of\Ndata we are having. We have collisions
Dialogue: 0,0:20:26.99,0:20:32.23,Default,,0000,0000,0000,,that are independent, so we only need one\Nin memory at any time and we have nested
Dialogue: 0,0:20:32.23,0:20:38.59,Default,,0000,0000,0000,,collections which makes the regular table\Nlayout a non-starter. We actually generate
Dialogue: 0,0:20:38.59,0:20:44.43,Default,,0000,0000,0000,,them from C++ objects so from classes,\Nclass definitions, C++ class definitions
Dialogue: 0,0:20:44.43,0:20:51.32,Default,,0000,0000,0000,,and we can read them back into C++ but\Nalso into JavaScript or Scala. Database
Dialogue: 0,0:20:51.32,0:20:56.84,Default,,0000,0000,0000,,just didn't do it for us. They have the\Nwrong model of data axis, they don't
Dialogue: 0,0:20:56.84,0:21:02.94,Default,,0000,0000,0000,,scale, it's just not the kind of system\Nthat works for us. Also using a file
Dialogue: 0,0:21:02.94,0:21:09.39,Default,,0000,0000,0000,,system as a storage back-end might sound\Nreally very traditional and boring but it
Dialogue: 0,0:21:09.39,0:21:13.89,Default,,0000,0000,0000,,works amazingly well and seems to be\Nfuture proof as well, so that's just the
Dialogue: 0,0:21:13.89,0:21:20.36,Default,,0000,0000,0000,,way to go for us. There are many other\Nstructured data formats out there, many of
Dialogue: 0,0:21:20.36,0:21:26.00,Default,,0000,0000,0000,,those did not exist when we started root\Nour own data format. But they also miss
Dialogue: 0,0:21:26.00,0:21:30.25,Default,,0000,0000,0000,,many things. For example, we wanted to\Nmake sure that we have schema evolution
Dialogue: 0,0:21:30.25,0:21:33.97,Default,,0000,0000,0000,,support. We can change the class layout\Nand still read back all data. We don't
Dialogue: 0,0:21:33.97,0:21:38.75,Default,,0000,0000,0000,,want to throw away all data just because\Nwe're changing the class. Also we do not
Dialogue: 0,0:21:38.75,0:21:43.37,Default,,0000,0000,0000,,trust people. That is a, you know, as a\Ncomputer scientist or whatever you
Dialogue: 0,0:21:43.37,0:21:46.75,Default,,0000,0000,0000,,probably know what I'm talking about\Nright? If people have to write their own
Dialogue: 0,0:21:46.75,0:21:50.63,Default,,0000,0000,0000,,streaming algorithm, there will be bugs\Nand we will lose data.
Dialogue: 0,0:21:50.63,0:21:54.61,Default,,0000,0000,0000,,We really don't want to do this, so we\Nwere trying to automate this, based on the
Dialogue: 0,0:21:54.61,0:22:03.07,Default,,0000,0000,0000,,class definition. So, last decision point\Nfor the story. Do you want to hear about
Dialogue: 0,0:22:03.07,0:22:10.41,Default,,0000,0000,0000,,cling, our C++ interpreter or about Open\NData and Applied Science? Let's start with
Dialogue: 0,0:22:10.41,0:22:14.86,Default,,0000,0000,0000,,option 1, the C++ interpreter\N{\i1}applause{\i0}
Dialogue: 0,0:22:14.86,0:22:21.11,Default,,0000,0000,0000,,Okay and and Open Data and Applied\NScience?
Dialogue: 0,0:22:21.11,0:22:29.68,Default,,0000,0000,0000,,{\i1}more applause than before{\i0}\NYeah. I'm heading there. You miss a fish.
Dialogue: 0,0:22:29.68,0:22:35.30,Default,,0000,0000,0000,,You can look at the slides later. Okay, so\Nthere we go. Really? No. The slide number
Dialogue: 0,0:22:35.30,0:22:41.14,Default,,0000,0000,0000,,is wrong. Oh a bug! So, Open Data and\NApplied Science. Okay, you really wanted
Dialogue: 0,0:22:41.14,0:22:47.70,Default,,0000,0000,0000,,to know about our budget, I understand\Nthat. So we get from you about 1 billion
Dialogue: 0,0:22:47.70,0:22:50.72,Default,,0000,0000,0000,,year and the currency doesn't really\Nmatter anymore at this, at this point of
Dialogue: 0,0:22:50.72,0:22:54.20,Default,,0000,0000,0000,,time.\N{\i1}laughter{\i0}
Dialogue: 0,0:22:54.20,0:23:01.23,Default,,0000,0000,0000,,And that is a lot of money. And you know?\NWe try to do really wonderful things, I
Dialogue: 0,0:23:01.23,0:23:04.94,Default,,0000,0000,0000,,mean we really enjoy our job, we love it.\NIt's fantastic to work in such an
Dialogue: 0,0:23:04.94,0:23:09.25,Default,,0000,0000,0000,,environment. And thank you very much for\Nmaking that possible. Really, I mean it.
Dialogue: 0,0:23:11.11,0:23:16.69,Default,,0000,0000,0000,,But it also means that you decided as\Nsociety to enable something like CERN.
Dialogue: 0,0:23:17.47,0:23:22.14,Default,,0000,0000,0000,,Which I think really deserves my applause\Nand yours probably as well. I think it's a
Dialogue: 0,0:23:22.14,0:23:24.42,Default,,0000,0000,0000,,great decision to do something like this.
Dialogue: 0,0:23:24.42,0:23:30.21,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:23:31.32,0:23:35.69,Default,,0000,0000,0000,,So we realize this, right? We realized\Nthat we are basically, that we can do what
Dialogue: 0,0:23:35.69,0:23:40.21,Default,,0000,0000,0000,,we do because of you, and we are trying to\Nreact to that by giving back what we do.
Dialogue: 0,0:23:40.21,0:23:47.46,Default,,0000,0000,0000,,Software, research results, hardware and\Ndata. So the way we share research results
Dialogue: 0,0:23:47.46,0:23:52.60,Default,,0000,0000,0000,,is through open access. We have it,\Nfinally. It took us a long time to fight
Dialogue: 0,0:23:52.60,0:23:57.57,Default,,0000,0000,0000,,with publishers and, you know, the\Nestablishment, but now we have it. We
Dialogue: 0,0:23:57.57,0:23:59.22,Default,,0000,0000,0000,,also, yes thank you.
Dialogue: 0,0:23:59.22,0:24:03.40,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:24:03.40,0:24:07.52,Default,,0000,0000,0000,,We also put a lot of effort in\Ncommunicating our results and what we are
Dialogue: 0,0:24:07.52,0:24:12.68,Default,,0000,0000,0000,,doing. And if you're in the region, it's\Ndefinitely worth a visit. I mean the URL
Dialogue: 0,0:24:12.68,0:24:17.59,Default,,0000,0000,0000,,is really easy to remember, it's\Nvisit.cern, and you know, works. And you
Dialogue: 0,0:24:17.59,0:24:22.27,Default,,0000,0000,0000,,should go there by April, actually, if you\Ncan because then you can ask people how to
Dialogue: 0,0:24:22.27,0:24:27.58,Default,,0000,0000,0000,,get on the ground, because the accelerator\Nis off at the moment. We also do applied
Dialogue: 0,0:24:27.58,0:24:32.32,Default,,0000,0000,0000,,research, for example we have this super\Ncool experiment where we try to study how
Dialogue: 0,0:24:32.32,0:24:39.63,Default,,0000,0000,0000,,clouds form, based on cosmic rays. So the\Nthe influence of cosmic rays and cloud
Dialogue: 0,0:24:39.63,0:24:45.77,Default,,0000,0000,0000,,formation. Which is a key element in the\Nuncertainty of climate models. We are
Dialogue: 0,0:24:45.77,0:24:50.44,Default,,0000,0000,0000,,trying to, to think about, you know, how\Nto make energy from nuclear waste. So
Dialogue: 0,0:24:50.44,0:24:54.83,Default,,0000,0000,0000,,getting rid of nuclear waste while making\Nenergy from it. And we are trying to
Dialogue: 0,0:24:54.83,0:25:02.07,Default,,0000,0000,0000,,repurpose detectors that we have and you\Nknow develop. We have something called
Dialogue: 0,0:25:02.07,0:25:08.33,Default,,0000,0000,0000,,open hardware, for example White Rabbit:\Ndeterministic ethernet, we have Open Data,
Dialogue: 0,0:25:08.33,0:25:12.79,Default,,0000,0000,0000,,and we have the LHC@home and some other\Nprograms, where either you can donate
Dialogue: 0,0:25:12.79,0:25:21.25,Default,,0000,0000,0000,,compute power or your brain and help us\Nget better results. We explicitly try to
Dialogue: 0,0:25:21.25,0:25:25.75,Default,,0000,0000,0000,,use open source as much as possible, and\Nalso feed back, whenever we see issues.
Dialogue: 0,0:25:27.70,0:25:33.62,Default,,0000,0000,0000,,But we also create open source. For\Nexample, we create Geant, which is a
Dialogue: 0,0:25:33.62,0:25:37.83,Default,,0000,0000,0000,,program that allows you to simulate how\Nparticles fly through a matter, for
Dialogue: 0,0:25:37.83,0:25:44.61,Default,,0000,0000,0000,,example used by the NASA. We have Indico,\Nwhich allows us to schedule meetings,
Dialogue: 0,0:25:44.61,0:25:48.94,Default,,0000,0000,0000,,upload slides, you know, these kind of\Nthings. Across the globe, lots of people,
Dialogue: 0,0:25:48.94,0:25:52.97,Default,,0000,0000,0000,,with access protection, all these kind of\Nthings. And it's open source. We have
Dialogue: 0,0:25:52.97,0:25:58.92,Default,,0000,0000,0000,,DaviX, the dimension we love HTTP. That's\Nthe next machine of Tim Berners-Lee. And
Dialogue: 0,0:25:58.92,0:26:03.14,Default,,0000,0000,0000,,that's his futile effort in trying to\Nprevent the cleaning personnel from
Dialogue: 0,0:26:03.14,0:26:07.53,Default,,0000,0000,0000,,switching it off. They don't speak\NEnglish, they did not back then at least.
Dialogue: 0,0:26:09.34,0:26:15.50,Default,,0000,0000,0000,,So we use we used DaviX to transfer files\Nover HTTP, with a high bandwidth. Or we
Dialogue: 0,0:26:15.50,0:26:21.24,Default,,0000,0000,0000,,have CVM-FS, which allows us to distribute\Nour binaries across the globe, and not
Dialogue: 0,0:26:21.24,0:26:26.57,Default,,0000,0000,0000,,rely on admins downloading stuff and\Nmaking sure it actually runs, and these
Dialogue: 0,0:26:26.57,0:26:31.58,Default,,0000,0000,0000,,kind of things. That is a lifesaver, it's\Nreally fantastic, it's a great tool. But
Dialogue: 0,0:26:31.58,0:26:37.73,Default,,0000,0000,0000,,nobody knows it. And we have ROOT, but\Nthat's coming up. So now, the last
Dialogue: 0,0:26:37.73,0:26:42.53,Default,,0000,0000,0000,,official part of this, of this\Npresentation, how do we do data analysis?
Dialogue: 0,0:26:42.53,0:26:44.95,Default,,0000,0000,0000,,Not like that.\N{\i1}laughter{\i0}
Dialogue: 0,0:26:44.95,0:26:52.21,Default,,0000,0000,0000,,{\i1}applause{\i0}\NWe use, we use C++ and actually physicists
Dialogue: 0,0:26:52.21,0:26:58.14,Default,,0000,0000,0000,,need to write their own analysis in C++.\NWe have very few people who have an actual
Dialogue: 0,0:26:58.14,0:27:03.88,Default,,0000,0000,0000,,education in programming. so that's sort\Nof a clash. As I said, we need to keep one
Dialogue: 0,0:27:04.61,0:27:08.46,Default,,0000,0000,0000,,collision in memory. And for what, you\Nknow, what matters to us is throughput. We
Dialogue: 0,0:27:08.46,0:27:13.34,Default,,0000,0000,0000,,want to have, we want to analyze as many\Ncollisions as possible per second. What we
Dialogue: 0,0:27:13.34,0:27:17.39,Default,,0000,0000,0000,,can do, is specialize our data format to\Nmatch the analysis, because we don't want
Dialogue: 0,0:27:17.39,0:27:23.42,Default,,0000,0000,0000,,to waste I/O cycles, if we can, you know,\Nif we can make use of the CPU better. ROOT
Dialogue: 0,0:27:23.42,0:27:29.11,Default,,0000,0000,0000,,allows us to do this since twenty years.\NIt's really the workhorse for the analysis
Dialogue: 0,0:27:29.11,0:27:35.20,Default,,0000,0000,0000,,in high energy physics. And it's also an\Ninterface to complex software. We have
Dialogue: 0,0:27:35.20,0:27:40.95,Default,,0000,0000,0000,,serialization facilities, we have the\Nstatistical tools, that people need, and
Dialogue: 0,0:27:40.95,0:27:44.48,Default,,0000,0000,0000,,we have graphics, because once you have\Ndone your analysis you need to communicate
Dialogue: 0,0:27:44.48,0:27:48.50,Default,,0000,0000,0000,,that to your peers and convince people,\Nand publish, and so on, so that's part of
Dialogue: 0,0:27:48.50,0:27:54.17,Default,,0000,0000,0000,,the game. All of that is open source, and,\Nof course, all of that is not just used by
Dialogue: 0,0:27:54.17,0:28:03.37,Default,,0000,0000,0000,,high energy physics. So, to conclude: We\Nare here, because you make it possible.
Dialogue: 0,0:28:03.37,0:28:05.22,Default,,0000,0000,0000,,Thank you very much. It's fantastic to\Nhave you.
Dialogue: 0,0:28:05.22,0:28:10.86,Default,,0000,0000,0000,,{\i1}applause{\i0}\NWe want to share and we have great people
Dialogue: 0,0:28:10.86,0:28:17.08,Default,,0000,0000,0000,,for science outreach, but we have nobody\Nfor software outreach, basically. So maybe
Dialogue: 0,0:28:17.08,0:28:24.57,Default,,0000,0000,0000,,it's worth a look to see what what CERN is\Nproducing software-wise. Scientific
Dialogue: 0,0:28:24.57,0:28:29.94,Default,,0000,0000,0000,,computing is nothing new, it existed since\Na long time, but we had to start fairly
Dialogue: 0,0:28:29.94,0:28:35.49,Default,,0000,0000,0000,,early on a large scale. So when we were\Nbuilding it up, we had to take... we were
Dialogue: 0,0:28:35.49,0:28:39.96,Default,,0000,0000,0000,,trying to take pieces that existed and did\Nnot found find much. So now we ended up
Dialogue: 0,0:28:39.96,0:28:45.18,Default,,0000,0000,0000,,with C++ data serialization, efficient\Ncomputing for non computer scientists
Dialogue: 0,0:28:45.18,0:28:49.66,Default,,0000,0000,0000,,even... In the part that I skipped and,\Nyou know, one of the alternate tracks, you
Dialogue: 0,0:28:49.66,0:28:54.29,Default,,0000,0000,0000,,would have seen that we have a Python\Nbinding as well for the whole software
Dialogue: 0,0:28:54.29,0:28:59.97,Default,,0000,0000,0000,,stack in C++. And for us, what matters\Nmost is scale. Now we are seeing that we
Dialogue: 0,0:28:59.97,0:29:04.31,Default,,0000,0000,0000,,are not the only ones. There are many more\Nnatural sciences arriving at a similar
Dialogue: 0,0:29:04.31,0:29:09.12,Default,,0000,0000,0000,,challenge of having to analyze large\Namounts of data. Now I promised to you
Dialogue: 0,0:29:09.12,0:29:12.48,Default,,0000,0000,0000,,that I'll be bold and I'll try to make a\Nfew statements of what will happen with
Dialogue: 0,0:29:12.48,0:29:16.75,Default,,0000,0000,0000,,data analysis, not just in science.\NBecause what we see is that we actually
Dialogue: 0,0:29:16.75,0:29:22.61,Default,,0000,0000,0000,,educate the people who will do data\Nanalysis, not just in science. What we see
Dialogue: 0,0:29:22.61,0:29:30.99,Default,,0000,0000,0000,,is that in the past, data volume mattered\Nmost. So more data meant more power. Now
Dialogue: 0,0:29:30.99,0:29:35.93,Default,,0000,0000,0000,,that's not the complete truth anymore.\NIt's a lot about finding correlations. So
Dialogue: 0,0:29:35.93,0:29:40.88,Default,,0000,0000,0000,,even with the amount of data not growing\Nanymore, because it's already humongous,
Dialogue: 0,0:29:40.88,0:29:46.32,Default,,0000,0000,0000,,we try to squeeze more knowledge out of\Nit. And for that, I/O becomes important
Dialogue: 0,0:29:46.32,0:29:53.90,Default,,0000,0000,0000,,and CPU limitations is the crucial factor.\NWe see that multivariate techniques are
Dialogue: 0,0:29:53.90,0:29:59.03,Default,,0000,0000,0000,,still rising and they will just be part of\Nthe toolchain of the statistical tools;
Dialogue: 0,0:29:59.85,0:30:06.68,Default,,0000,0000,0000,,except for generative parts, which, I\Nbelieve, will change the way we model.
Dialogue: 0,0:30:10.23,0:30:16.36,Default,,0000,0000,0000,,Now, based on what I just described, this\Nis not a big surprise anymore. As we need
Dialogue: 0,0:30:16.36,0:30:21.21,Default,,0000,0000,0000,,throughput, we need to have a language for\Nthe core analysis part, that is close to
Dialogue: 0,0:30:21.21,0:30:26.97,Default,,0000,0000,0000,,metal, so something like C++.\NOn the other hand writing analyses is
Dialogue: 0,0:30:26.97,0:30:31.79,Default,,0000,0000,0000,,still complex, so you need a higher-level\Nlanguage and for that people could, for
Dialogue: 0,0:30:31.79,0:30:35.93,Default,,0000,0000,0000,,example, use Python. So, now language\Nbinding becomes relevant all of a sudden.
Dialogue: 0,0:30:35.93,0:30:42.01,Default,,0000,0000,0000,,It's much more important in the future.\NAnd we need to tailor I/O to the actual
Dialogue: 0,0:30:42.01,0:30:48.91,Default,,0000,0000,0000,,analysis to not waste CPU cycles. So\Nthroughput is the king and, in my point of
Dialogue: 0,0:30:48.91,0:30:54.33,Default,,0000,0000,0000,,view, also in the future we will see much\Nmore effort in increasing the throughput.
Dialogue: 0,0:30:55.60,0:31:03.12,Default,,0000,0000,0000,,Okay, so that was it. In case you want to\Ndiscuss anything with me, like "That's
Dialogue: 0,0:31:03.12,0:31:07.97,Default,,0000,0000,0000,,just wrong!", that's fine. I'm probably\Nhave several bugs in there. I'm still here
Dialogue: 0,0:31:07.97,0:31:12.91,Default,,0000,0000,0000,,until tomorrow. I don't know where yet,\Nso I'll wander around and you can contact
Dialogue: 0,0:31:12.91,0:31:16.82,Default,,0000,0000,0000,,me by email or Twitter. Thank you very\Nmuch for your attention. Thank you.
Dialogue: 0,0:31:16.82,0:31:20.52,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:31:20.52,0:31:27.99,Default,,0000,0000,0000,,{\i1}music{\i0}
Dialogue: 0,0:31:27.99,0:31:45.00,Default,,0000,0000,0000,,subtitles created by c3subtitles.de\Nin the year 2017. Join, and help us!