Toward AI as Co-Scientists - Dr. Zongrui Pei

Publicado em

Vídeo em Inglês, publicado pelo canal USC Physics and Astronomy.

About this content: The summary, key points and topics below are generated automatically from the video transcript. They reflect the content, opinions, claims and predictions presented in the video and do not represent a position, validation or fact-check by IA Dots.

Video summary

Dr. Björn Schenke discusses the internal structure of protons and nuclei at high energies, focusing on quarks, gluons, and the phenomenon of gluon saturation. He explains theoretical frameworks like quantum chromodynamics, color glass condensate, and diffractive vector meson production as methods to probe nuclear structure and spatial distributions using colliders such as HERA, the LHC, and the upcoming Electron-Ion Collider (EIC).

Key points

  • Protons and nuclei consist not only of valence quarks but also a dense sea of antiquarks and gluons, with gluons carrying about 50% of the proton's momentum.
  • Quantum Chromodynamics (QCD) evolution equations, such as DGLAP and BFKL, describe how parton distributions change with resolution scale (Q squared) and longitudinal momentum fraction (x).
  • Gluon saturation occurs at high energies and very small x when gluon splitting and merging processes balance, preventing indefinite cross-section growth and characterized by a saturation scale QS.
  • Diffractive vector meson production processes provide sensitivity to both gluon saturation and the spatial distribution or gluonic radius of targets.
  • Nuclear targets increase gluon density proportional to A to the 1/3, making them more efficient than protons for reaching and studying the gluon saturation regime.
  • Upcoming facilities like the Electron-Ion Collider (EIC) and ultra-peripheral heavy ion collisions at the LHC offer complementary measurements to pin down gluon saturation and nuclear deformation.

Topics

Proton structureGluon saturationQuantum ChromodynamicsDeep inelastic scatteringDiffractive scatteringElectron Ion ColliderColor glass condensateNuclear structure

What you will learn

  • Gluons carry approximately 50% of a proton's momentum, demonstrating that protons contain much more structure than just three valence quarks.
  • The gluonic radius of a proton, measured via diffractive processes, is found to be significantly smaller than its electric charge radius (around 0.6 femtometers versus 0.84 femtometers).
  • Using heavier nuclear targets instead of protons lowers the cost and energy requirements needed to reach the gluon saturation regime because nucleon gluon distributions line up and add together.
  • Incoherent diffraction cross-sections serve as a measure of target fluctuations, including geometric substructures and nuclear deformations like quadrupole or hexadecapole shapes.

Questions and answers

What percentage of a proton's momentum is carried by gluons?

Gluons carry about 50% of the momentum of the proton.

What is the estimated gluonic radius of a proton based on the discussed diffractive measurements?

The extracted gluonic radius of the proton is about 0.6 femtometers.

Why are nuclear targets preferred over protons for studying gluon saturation?

Nuclear targets have more nucleons lined up behind one another, which increases the density of gluons and makes the gluon saturation regime easier and cheaper to access.

Video transcript

Transcript from available captions; it may contain errors.

So, let's start the colloquium. Today we have Dr. Björn Schenke from Brookhaven National Lab. I'm going to let Professor uh Pavel introduce >> I can give a short short introduction. Uh Björn is a distinguished scientist at the nuclear theory group at Brookhaven National Lab. And he originally got his PhD from the Goethe University in Frankfurt. Then he was in Canada for a while at McGill. And since then at Brookhaven where he is the acting director of the EIC Theory Institute. He's a fellow of the APS. He was the chair of the APS group on uh hadronic physics. PI of the topical collaboration on gluon saturation. And also last but not least a member of the second detector [laughter] working group. So, Björn, let's hear about nuclei and gluons. >> Okay. Thank you so much, Pavel. Thank you everyone and thanks for the hospitality. It's It's It's been great already. Um and it's a very nice campus. Thanks for the tour, Pavel, and very impressive. Uh for example, even the hawks know how to drive. [laughter] So, that was from today. Um but let's jump into the topic. Um so my my first question is sort of what's inside a proton? Okay? And so that's that's what I want to go through here and and the question mainly is what's inside a proton at very high energy and what's inside a nucleus at very high energy. But let's start at lower energy and and ask how do we find out? So, how do you find out? You do kind of like Rutherford did to find the atomic nucleus by shooting something, in this case alpha particles, uh on a gold foil and look at how they scatter, right? And then the patterns of how they are deflected are such that sometimes you have this really hard back scattering. And that of implied that, you know, we have uh some concentrated um um mass or charge in this case uh in in uh in in very concentrated spot and we don't have like a smooth charge distribution of the whole atom, right? So this was the discovery of the atomic nucleus. They did that in Manchester, but when you are at McGill, they tell you he came up with the idea at McGill and all of that happened there. They tell you that a lot uh but um the same principle you can use, right? Just by going to higher and higher energy, you'll be able to have higher and higher momentum transfers. And the higher the momentum transfers, the smaller so the resolution scale is that you that you can achieve. So um momentum transfers I'm going to always express by this Q squared, so that's typically a squared momentum transfer or the negative of a squared momentum transfer. So remember that that's one of the two things like there's Q squared and X and both you have to remember throughout the talk, but that's it, okay? So Q squared is sort of a momentum transfer measure and uh if it's higher, you have finer resolution. So a proton scattering uh or an experiment where an electron scatters off a proton will look something like this and so I will scatter off some constituents of that. So what's the answer? Well, the first answer was three quarks. Okay? So this sort of was established over the 1960s and 70s. So first of all, there was a quark model. There was independently uh one developed by a Gell-Mann and one by Zweig. And um the uh that proposed that we have sort of two up quarks and a down quark. And then SLAC uh experiments, so-called deep inelastic scattering basically just this hitting an electron onto a proton found that we have point-like constituents. And that this this graph here kind of illustrates this. I don't want to go through all the details, but basically on this axis is this Q squared. So that's how hard I I knocked the proton. And this is the scattering probability. And the data is here. These are these points. And this dashed line is sort of what you expect if the proton itself was a smooth kind of uh distribution of of charge without any hard scattering centers. And what I find is there is something here at the very fine resolution scale that gives me a large scattering probability. So there's something there at short distance scale right of which I can scatter off pretty hard. Uh so that's why the data is much larger than this expectation. So this was the indication of that I have sort of basically or almost point-like constituents. Then in the '70s there were other experiments where uh this was combined basically with results from neutrino scattering uh um with CERN's Gargamelle. If you combine these right you have sort of uh one where you couple to the charge, the electric charge like the electron scattering, [clears throat] and one where you couple um uh with with with uh like uh W bosons and and sort of the weak uh uh interaction. And so that allowed you to kind of distinguish what the charges are, namely electric charges are fractional of 1/3 and 2/3 for the d and u quark. And that was exactly what was predicted by the by the quark model. [clears throat] So here we are that says there are three quarks to first approximation. But there's more, right? So there's actually uh c quarks as well and gluons and that was sort of found also in the same kind of experiments where if you uh measure as the function of the momentum fraction carried by these constituents, you see that um at at lower momentum fraction you actually have more stuff and I show you a plot in the measurement of what um where that came from in in the next slide. Um so basically uh this reveals that we have more structure than just three three simple quarks in the in the proton, so illustrated by more things flying around. So there's quarks, there's antiquarks uh which you can for example, extract by by again neutrino anti-neutrino deep inelastic scattering, when you look at which allows you to separate sort of quark plus anti-quark and quark minus anti-quark um reactions demonstrating that there is an anti-quark sea, so that there is actually anti-quarks present when you scatter off a proton. Now you can sum all the momenta of those charged guys and you find that they don't really add up to the proton's momentum, so the rest must be carried by something else, which we understand now is are the gluons, which carry actually about 50% of the momentum of of the proton. So this is sort of a technical slide, but I just have it here for to repeat that these two numbers will come up again and again. So I'm looking at processes, as I said, where scatter an electron of a proton, right? So this is the electron with some incoming momentum, some outgoing momentum. It basically exchanges a photon with the proton. The photon may or may not break up depending on the process, typically in deep inelastic inelastic means it does break up. So it can produce all kinds of stuff summarized by X. And so as I said before, the photon momentum, that's the momentum that's transferred is is this little Q. If I square it, take a negative, that's my Q squared that I introduced before. I said before the transverse resolution scale is sort of the inverse of that, so that's this kind of formalizes this a little bit. And I can define this quantity, another thing that I can measure just by measuring all the outgoing and incoming momenta is this Bjorken X. And you can show that that's basically in some in some frame where the proton essentially has infinite momentum and I can neglect its mass is equivalent to the longitudinal momentum fraction that the part on that I scatter off carries. So let's say I have a quark in there or a gluon and it's carries about 10% of the uh of the proton's momentum and I scatter off that, then I find that this x B will be 0.1. It's measurable, and it tells me sort of, you know, what kind of parton do I hit. Um so yeah, so Q squared tells us how finely we look in in the transverse direction, and x B tells us which momentum scale and the longitudinal momentum I probe. So I have I want to illustrate this a little bit in the next few pictures. Just one more word on that we can calculate things. So this is quantum chromodynamics is sort of the the theory that covers how quarks and gluons interact. And they carry color charge. So just like an electron or a proton carry electric charge, these guys carry also color charge. So that means uh gluons, for example, because they carry color charge themselves, which distinguishes them from photons, right? They are charge carriers, but they don't or they are enforced carriers, but they don't carry charge themselves. If a gluon, for example, can split into two other gluons with lower momenta, or it can split into a quark antiquark pair, or a quark or antiquark can radiate a gluon. And it turns out the emission probability to for any of such process behaves in a way that we see it's enhanced for very small transverse momentum. That means emitting something collinearly uh is very probable. And also for small x emission, so emitting something very soft is also very probable. So uh so in the end, this leads to some logarithmic divergencies, right? So to to some resummation that's necessary, this resummation can be formalized uh in an evolution equation. And this you might have might have heard of. Um so there's two evolution equation, one for the evolution of transverse momentum scale that just tells us how does my distribution change if I go to larger and larger Q squared? That's the D GLAP named after the authors. And if I want to know what happens if I go to smaller and smaller x, smaller and smaller momentum fraction, that's covered by this BFKL, again named after the authors evolution equation. Um, what this means, for example, is um, originally I showed you this plot earlier, right? Remember this plot? So, this is the Q squared dependence of the scattering probability. And it's pretty flat. Maybe not entirely, but it's very flat, right? In Q squared. If if it was very flat, um, if it was completely flat, basically it tells us I have just have these point-like constituents. Um, but DGLAP evolution, if you actually solve the equation, finds that you get some Q squared dependence, but it's only logarithmic. So, that's a weak dependence okay? Um, but it is a violation of what was called Bjorken scaling, which would be would be the total Q squared independence right? Um, so if I measure something at fixed X uh, and then uh, look at how it depends on Q squared, right? That's done here. So, each curve is a different X and then a function of Q squared. I actually find there's some dependence. So, these are the very low X, uh, very small X ones, uh, that you see actually increase if I go to finite finite resolution scale, whereas if I go to large X, I find the decrease. And this is basically something that's predicted by QCD and you solve it with this evolution equation and it was measured, right? All these data points are the measurement. This seems to work really well. Um, and now I want to kind of tell you what this looks like in some illustrations. So, there were actually guys from MIT and from JLab that worked with some artists to try to visualize the whole thing that I was talking about. So, this is a function of Q squared. So, if I this sets to the resolution scale. So, if I have low Q squared, I kind of can resolve my proton at this scale, right? So, it's I see that there are kind of these these these these colored blobs, which are my valence quarks. Uh, and then if I go finite finite, I see finite finite structures right? Um, so basically it allows me then to to scatter off smaller and smaller distance scale objects if you will. And uh this describes the Q squared dependence. The X dependence is similar. So if I go to smaller and smaller X here to the left, I see the number of gluons is increasing. Uh something that was measured, something that also comes from the evolution equation now in in X. So both on the theory side and experiment we have observed this. And here's the experimental observation by experiments at at HERA. That's the Hadron Electron Ring Accelerator at at DESY in Hamburg that that used to run. So it's basically an electron proton collider. Uh and this shows you sort of uh something that's proportional to the abundance of quarks and gluons inside a proton. So you see there's these two distributions that have bumps of uh around X. That's the momentum fraction they carry of around, you know, a little under 1/3, right? So these are my valence quarks. So you see I have the down quark and I have up. And I told you before the proton we expect to be up up down, right? So it's actually more up than down. And they're kind of concentrated at larger momentum fraction. But you see these other two curves right? They increase as a function of decreasing X. And that's um the valence quark. Sorry, that's the C quarks down here and the gluons which I want to focus on. But you see also this, they scale down already by actually factor of twin. So they become much much more abundant as I go to small X. And this is kind of what's illustrated here. Uh as I go to smaller and smaller X, um I have this huge abundance of mainly gluons but also C quarks. [snorts] And this is also in illustrated here in this little movie that they made. Uh so basically if you scan through this in X in the momentum fraction of the of the partons, you see how the structure changes. This is a neat little animation that they made to to visualize how these parton distribution, here in this case at fixed resolution, depend on the momentum uh fraction that these part-ons carry. Okay. So now my talk has in the title gluon saturation. So what does that mean? So we've seen now that we go if we go to smaller and smaller x, we have these splittings and uh of gluons, so they become more and more and more. But eventually you would say well if this keeps growing uh I would violate unitarity. My cross section would grow indefinitely. Now this is cured by the fact that once the density is high enough the probability for merging two gluons into one also becomes large and this kind of balances the splitting process. So once we reach that point where these probabilities are comparable that's the time or that's the um that's the x if you will when uh gluon saturation sets in um and the evolution becomes non-linear here, right? So because now the probability for merging to gluons depends on the distribution of two gluon distributions right? Um and the gluon saturation is characterized by a transverse momentum scale, that's so so-called saturation scale QS that depends on where where m and x and basically all gluons that have momenta transverse momenta lower than this scale are in the saturated regime. So again this is a little technical but really it's not that hard. So basically I'm just saying I'm going to high energy. So I have a high energy collision. I build a very big expensive collider. Uh the I take uh this is the center of mass energy squared, this s. I take it to infinity. Uh the kinematics tells me it's proportional to this q squared divided by x. So I can reach infinite energy by either keeping x fixed and taking q squared to infinity. In this plot uh that's in on the vertical, right? And that's the so-called Bjorken limit. So what happens is I can actually increase q squared that means my resolution scale also gets finer and finer. This is why these blocks become smaller and smaller. Alternatively, I can keep Q squared fixed and go to very small x. That's down here. Right? And you see the resolution scale doesn't change, indicated by the blocks staying the same size. But then what happens eventually, if they become more and more, they will overlap. So I kind of saturate uh my my space here in this illustration. And that's actually where I hit the the uh saturation regime. So again, I'm just repeating myself here, right? So this line is the is where the saturation scale sits. Right? Uh once I cross it, I'm in the saturated regime. I'm in the regime where these non-linear effects are important. The question is how can I actually see that I am in that regime? Right? Can I experimentally access that and see the signatures of it? So generally you would think, well, to reach it I have to go to very high energy, namely have to go to very small x. I should go also to large nuclei because once I have more nucleons in there that all sit there together, I also have more gluons. So I already um boost or increase the the density by going to a large nucleus compared to a proton. And I want to measure at somewhat low transverse momentum, I call PT here, uh because, you know, I I I told you the gluons that have lower momentum than QS, they will be the ones that are in the saturated regime. So QS is then, once I'm probing this, the relevant scale. So examples for that uh are actually quite numerous. I can do this in in proton nucleus collisions, for example. So for example, if my nucleus is in the saturated regime, but the proton is not at the same energy, then I can compare what I get for particle production in the proton nucleus collision to a scaled proton proton collision, scaled by the number of nucleons in this nucleus. And if that is suppressed, I know or at least it's an indication of that um the nucleus is already saturated, whereas the proton is not yet, for example. I can have also suppression of with uh and deco-correlation of forward uh dihedron or dijet correlation. So, for example, if I produce two pions in a process or or two jets, which is just sort of collimated sprays of of of particles, um I can look at how does that production and this correlations of those two particles depend on changing X or changing the nucleus that I'm working with as a target, um and how does it depend on transverse momentum. And there's certain predictions, you know, uh quantitatively from the saturation uh calculations that would tell me how how this correlation, for example, is is weakened because I have these these dense gluon fields in the target with which I interact. There's also diffractive observables, for example, in electron-nucleus collisions. And and those involve at least the two-gluon exchange. So, diffractive means essentially that the target usually doesn't break up and and that um my exchange is a color-neutral exchange. So, if it's color-neutral, but a gluon always carries color, right? I need to at least have two gluons exchange to compensate for the color. And because I have two gluon exchanges at least, right? I'm sensitive to the gluon distribution square, right? So, I'm more sensitive to what does my gluon distribution of the target look like. I give you one review which kind of reviews several of these and more. Uh but for this talk I want to focus on diffraction because it's actually quite exciting. First of all, as I said, it should be pretty sensitive to gluon saturation, but it has another interesting feature that it's also sensitive to the nuclear structure, to the to the spatial distribution of the gluons inside my target, um which makes it particularly interesting. So, here comes a little bit of theory, but don't be scared. It's not that much. Uh I'm just going to kind of sketch roughly how you would compute something uh for um you know, explaining these or describing these different experiments on the theory side or on the phenomenology side. So the tool for that typically is the color glass condensate which is essentially an effective theory of quantum chromodynamics in this limit where I go to high energy by going to very small x and keeping q squared fixed, right? So that's that's that's what this is and there it turns out because the gluon fields at small x are highly occupied, there's a lot of gluons, right? And whenever I have a lot of stuff I can treat it classically, right? Quantum effects become less and less important the more stuff I have, right? So I have a ton of stuff in this, right? So when I drop this to the ground I don't need to consider quantum effects right for that. So that's something like this. So and these gluon fields then I can basically compute as a solution to a classical equation namely the classical Yang-Mills equation. This may look familiar. It's basically a Maxwell equation except that my charges are now not electric charges but color charges. But the structure is the same. So this f mu nu is the field strength tensor which contains these a mu's and their derivatives and then on the right hand side I have a current, right? So the current moves and produces fields just like in the Maxwell equation okay? So I have moving sources which are the center of my fast moving degrees of freedom. You can think of them maybe as the the large x quarks and gluons that are inside my proton or my nucleus and then the fields are the the smaller x degrees of freedoms which I have to treat dynamically in this picture. Now if I want to interact with a system like this which is described by this collection of gluon fields for example I would shoot a colored particle, right? Through it. So let's say I shoot a quark through that target and then because the gluon gluon fields are so highly occupied Each interaction with the gluon field, right? Typically you would say, okay, in my perturbative expansion, right? I have a coupling. That coupling constant is G. Ideally that is small because my scale QS is supposed to be large enough, larger than my lambda QCD scale. So the coupling hopefully is smaller than one. Uh and I could in principle just say, okay, one exchange is more important than two exchanges. That's the typical perturbation theory that we do, right? However, because these A fields are highly occupied, it turns out actually the occupation is roughly like one over G. That means every insertion here is order one because it goes like G * A. So four of them is just as important as two of them, and 20 of them just as important as 10 of them, etc., right? So I have to resum. And that's done in this in this in these calculations by basically writing down an effective vertex which describes sort of the resummed collective interaction of multiple gluon fields with with my probe, okay? That's also called a Wilson line. For the experts, that's like a path-ordered exponential, um where essentially you just have an exponential and then integrate over all the gluon field insertion along along the path of the probe. But you can think of it as an as a an effective vertex which kind of takes care of all the interactions with the gluon fields. And so from that thing, from this effective vertex, I can construct all my processes that I'm interested in. Here's just some examples again. So I could have, for example, a proton nucleus collision. So A I always call the nucleus, little P is a proton, that can produce some hadron, whatever it might be, and and the rest, which is X, right? And then basically if the proton coming in, one of the quarks may interact with the target indicated by this structure here. The effective vertex is what what does things to it. Uh and then a hadron is produced. That's the amplitude for that process. Another process would be the inclusive cross-section for electron nucleus scattering, right? So that would be using and I'm assuming you heard of the optical theorem. So basically you take just the forward scattering amplitude um and that would involve a photon coming in that the electron emits that splits into a quark anti-quark dipole which then each of them uh interacts with the target and then uh and then you know I have the other side of this. Um that would be inclusive scattering. So many of those I can compute and always my building blocks are essentially these effective vertices or Wilson lines. And I can even use uh these gluon fields that I compute uh as initial states for for heavy ion collisions. So this is really where I came from. So you can then kind of take two nuclei with with these gluon fields and compute what happens when they collide and you can compute the fields as they are after the collision and get an initial state for whatever happens in a in a nucleus nucleus collision. But I don't want to focus on that today. Uh I want to give you one more example namely the fraction that I said I want to focus on. Again this is the same kind of graph. Forget about the green stuff for a second, right? This is just uh labeling everything. But it's a similar process as before. We have an electron, it emits a photon and I should say and I didn't say it in the previous slide, right? Because gluons don't carry electric charge the photon cannot couple directly to them, right? So the leading order process where the uh I can how I can couple to gluons is for this photon uh to split into a quark and anti-quark who then interact with the gluon field, right? And essentially because of the high energy that I'm at I do an approximation where um all that happens to them is a color rotation, right? Because this is sort of a shock wave of gluons because of the high energies I'm working with they I have Lorentz contractions. It's basically a pancake and all that happens during the interaction is a color rotation. I don't have time to change their transverse position, for example. Uh which is why it's also useful to work in coordinate space. So for example, I would have this effective vertex or Wilson line at position transverse position X and for the anti-quark I have the Hermitian conjugate at position Y and that's what gives me my interaction. And then in this process, you know, what could happen is that these make individual particles, but in this process it's the process where these to get together then form uh some vector meson. For example, a J psi. So that's just a specific particle that they can form, right? Um and uh this is the process of diffractive vector meson production. And I will focus on that a little bit more. Just one more thing on this theory section then you can all relax. Um so I talked about that right that I have these Wilson lines in this process, for example, one at position X, one at Y. This forms together a dipole, right? Because there's two of them. And it's the simplest example of a Yes. >> In previous slide you didn't call it diffraction. Is it some wavelength limited? >> It's diffraction mainly because it's actually going to be uh uh a color color singlet exchange. So and then it's sort of similar to diffraction in the sense that I get a diffractive um well I have a connection with a Fourier transform between my uh final state momentum distribution, right? Or spectrum and sort of the the the spatial structure that I'm scattering off, right? So it's kind of like uh in in optical diffraction where I'm um um sensitive to the the size of the slit or so that I go through and then I get a pattern, right? An angular distribution. Here it's it's in a similar way where I kind of I'm sensitive to the shape of my target or how big it is, for example, right? And the um and the uh the the spectrum that I measured of the transverse momentum transfer. So I'll get to that. So you actually see kind of like a diffractive pattern with dips sometimes depending on your target. So especially if you use a larger target then it's easier to access these these dips even. Yeah. So it's sort of the analogy to to what the optical diffraction is. So this is kind of where this this name comes from. Um I get to the diffractive part a little bit in more a little bit more detail as well. This is more general, right? So this is this happens to appear already on the amplitude level uh in the diffractive process. As I said, it's in the diffractive case it's a color color neutral exchange. So that means I take the trace already on the level of the amplitude. But I'm going maybe into too much detail here. All I want to say is sort of this is sort of my main building block, the simplest building block. There can be other combinations of Wilson lines depending on the process. Um but what this already nicely illustrates is sort of the interaction strength. So this quantity, this D, which is 1 minus this Wilson line correlator is illustrated here. And you see at small dipole sizes it's very small. Which you can understand because it's basically describing a quark anti-quark and how far apart they are, right? And it's a color singlet. So if this if they are basically on top of each other there's no color charge there. So there cannot be any interaction. So this has to go to zero. Once I take them further apart this can grow and there can be some interaction with the target. And how strong the interaction is kind of depends on how strong the gluon fields are, right? Which are what these Wilson lines characterize. And you see um it goes up and without saturation which basically would keep keep going up. And so the fact that this goes to one is actually already a saturation effect. And where it turns over, and this is a little arbitrary which point you choose, but let's say at I choose this point here and say this is sort of the transition between weak and strong scattering, right? So between where this is small to where this is large. I can define this as my saturation scale, which I introduced earlier, but now you have sort of a more quantitative definition. So a small saturation scale would be further to the to the right, right? So I reach strong scattering later, whereas a large scattering a large saturation scale needs to strong scattering already for smaller dipoles, right? So this is basically saying the QS is the inverse dipole size where the transition to strong scattering occurs. And I have two asides because I don't want to go too much into complicated details, but I wanted to mention to you that we don't just do classical calculations usually, so we can go beyond that. So for example, the first correction to the classical calculation would be including the most important quantum effects. The most important ones are, if you remember from the beginning, we have these emissions, right, that are enhanced when X is is small, right? So we had this 1/X term in the QCD emission probabilities. And so these are log lead to logarithmic divergencies. And so the most important thing as I go to small X will be re-summing those, right? So re-summing those in these calculations, I cannot just do the BFKL that I mentioned before, which is the one that's valid in the dilute regime, but I have to take into account also then the non-linear effects. And that's done by the so-called Jimwalk evolution equation named after these guys. Basically again, an evolution in X, but now taking care of these non-linear effects where the gluons can split and also merge. And you can go beyond that even. You can do next-to-leading order effects that are beyond this leading effects. So for example, we did this for the production of dijets. So you have again, I'm repeating all these things very often, I would have a photon splitting into QQ bar, which then interacts with at target, and then it produces here the quark and the antiquark coming out, which then can shower and become a jet. Again, this collimated stream of particles. And then I could measure this, right? I could measure these jets with these two momenta, and I can calculate it at leading order just like this. It's very simple, just this one graph. But at next leading order, I have all kinds of additional possibilities where I can emit uh another gluon. I can have virtual corrections. I can have vertex corrections. And we have to take all of these into account. These are not even all of them. Uh but if you do that, you get, you know, a corrected result that uh and it actually turns out quantitatively to be important um for, you know, for phenomenology and comparing to experiments eventually. But this was just an aside. Uh I want to go back to the simpler case where we do our classical color calculation for now. Um so as I said, I want to work in coordinate space because it's it's kind of nice here um because I will also get to that we will send be sensitive to the spatial distribution. You could do this all in In principle, you could do calculations in in momentum space. But here let's do coordinate space. So B here is the point where I hit the target, right? This is sort of my impact parameter, one calls it. Then I have delta, which is my momentum transfer, and these actually turn out to be Fourier conjugates because my amplitude uh so schematically it looks like this. Uh and R is the size of my dipole, right? Which is what I had in this plot. So if they're close together, I have weak interaction. If they're far apart, I get stronger interaction. That's what that is. And Z is the momentum fraction the quark carries from the photon. And so I indicated these important parts. So I have first the pro It's basically a factorized approach. I have first the probability of my photon splitting to this QQ bar pair. Then I have the interaction with the target, which is done by this dipole amplitude, this D. And then I have the probability for these QQ bar forming some vector meson, like the J psi for example. That's all there is really, right? And then I integrate uh overall these coordinate space uh coordinates and I get in the end like uh um momentum dependent like delta dependent uh cross section. And the fact that I have this Fourier transform here, if I measure something as a function of this delta, this this momentum transfer, I you see through this connection I'll be sensitive to the spatial structure of the target. And so this is um uh showing the cross sections. So T is just uh essentially delta I mean it's minus delta squared. So that's just the Mandelstam variable if you heard of those. And uh again I repeat the diffractive process means I have a color neutral exchange. So I have at least two gluons. Uh experimentally that means actually I have a kind of a empty space in the detector. I call it rapidity gap here. Don't need to worry about what that means, but it's basically there's some space in the detector in the longitudinal direction where there's nothing because I'm not actually pulling color strings between what I produce and my target because I had a color neutral exchange to begin with. So there's nothing to pull a string and then that can break and produce stuff. That doesn't happen. That's why there's some empty space. Um now what can happen though is that either my target does not break up. That's this so-called coherent diffraction. And in that case basically it's an elastic process. My incoming state looks the same as my outgoing state. So that means I actually average over target configurations, like how how the different color fields look like in the target already on the amplitude level. And then I square. Right? You You maybe know if I compute a a cross section it's typically the amplitude squared, right? But here it matters how I do these averages. And the averages here are done on the level of the amplitude because the incoming outgoing set are the same. For the total diffractive cross section I do the opposite, right? Because the outgoing state may look different because the proton can break up or the nucleus. And so the total number will be just this first term here. So that's the amplitude squared and then averaged over all the color configurations I can have in the target. And the incoherent diffraction is just the difference between the total and the coherent. So that's the probability or the cross-section for the case where the proton definitely breaks up, okay? Whereas this first term includes all of it right? So the the interesting thing now is um this obviously is sensitive to the average structure or the average size, for example, of the target, where this you may recognize as a variance, right? So it's the average of the square minus the square of the average. That's a variance that measures fluctuations. And so this is sensitive to any kinds of fluctuations of this scattering amplitude, including geometric ones. And this I want to um sort of capitalize on by looking at how do my results look different if I assume different degrees of fluctuations of my target. So for example, my my what I call round proton, it's not perfectly round because there are some color charge fluctuations in there, but it's basically just like a Gaussian and then I have some fluctuating color charges sampled from that. Um that would look like one example of it looks like this. If I introduce some additional substructure, if I say, oh, maybe my gluons are concentrated around those valence quarks that emitted them in the first place, right? I would think maybe I have three hotspots of gluons. You could have many more hotspots of gluons. You could have all kinds of things. This is a model, right? This is just to see what effect would this have on my cross-sections. But certainly you can see that if you have something like this and in another event these blobs are at a different position and so on, this would lead to more fluctuations, right? Then than this one where I just have this sort of overall always round-looking thing with a little fluctuations. And indeed that's what you find. So let's look at some data and some results from these calculations. So now I am showing the fractive cross sections that are actually measured at HERA. So that was this thing in Hamburg where they collided electrons and protons. H1 is one of the experiments and the data points are these. So the solid points are the coherent one, the one where the proton does not break up. And the open points are the incoherent ones where the proton does break up. You see this is sort of actually like roughly a Gaussian shape. Well, this is a log scale, so it doesn't look super Gaussian, but this is plotted as a function of this delta square, basically this momentum transfer. And you see you can see I can now choose my parameters, right, the size of this proton etc. such that this fits it pretty well. Uh in both cases actually with the flag more fluctuating case and the less fluctuating case, I can get, you know, decent fits for both of them, at least close uh because, you know, this only measures the average shape. So if the fluctuating case has the same average size and shape, I expect this to be the same. But now look at the incoherent one, the one I said is sensitive to fluctuations. If I use this one that fluctuates only a little, I get this dashed line way below these data points, right? If I introduce additional fluctuations, uh I can tune the parameters, mainly the size of the proton and the size of each hotspot such that I can describe this very very well, right? And you see it's basically an order of magnitude or more difference. So it's a it's a huge effect. And so this tells you a few things. So first of all, let's look at the coherent one right? So I told you basically this this delta or T dependence is the Fourier transform of the spatial distribution. So if this And you know the Fourier transform of a Gaussian, right? If you have a broad Gaussian, the Fourier transform is a narrower Gaussian, right? And vice versa. So this means if this drops very uh slowly and is pretty broad in momentum space, it means it's actually much it's a small target uh in coordinate space. And And then if it was a larger target, it would drop more more rapidly. So this allows me to measure actually the size of the proton. But it's not the charge radius, right? Because I'm coupling to gluons, it's like a gluonic radius. So it's telling me where are the gluons located inside the proton. And what we extract just from this is about a radius of 0.6 femtometers. Okay, for this uh gluonic radius. This is sort of similar or at least close to what other measurements at JLab have measured um by a different process. They also kind of measure mass radius. I would say, well, mostly the stuff is gluons, so the mass radius should be similar to this. It's just my hand wavy argument. But what I want to emphasize is that it's definitely smaller than a charge radius. Right? You can measure this in other ways. It's like one example is or well one one I think this is sort of established as around 0.84 fermi femtometers. Uh so it's significantly smaller, right? So if this is telling you something about how the gluons distribute in the proton versus how are the charges electric charges distributed. Uh this I just want to quickly go over. We can do like a whole Bayesian analysis where we take this data and we do uh really kind of more complicated fits and then we can look at the correlations between different parameters. These include sort of the size of the proton, the size of these hotspots and other things. And you can do this very sophisticated and you can get, you know, pretty good fits uh to to the data and then, you know, learn about for example on the one hand the size of the target of the proton and also the degree of fluctuations by looking at the incoherent I don't want to dwell on this. I just say this can be done in a pretty sophisticated way. Uh I want to show you can also compute the W dependence. So here what we do is I talked about this gym work evolution briefly in my aside, right? So you include actually the leading quantum effects and that tell you the energy dependence of this cross section. So this is now basically you take this cross section that I've been talking about, the coherent one, and you integrate it over T, right? And that's what I plot now. And then plot it at different collision energies. And you see you can do these experiments at HERA, that's the H1 and ZEUS experiments, but also at LHC, and I will tell you how LHC does this in a little bit. But it's uh essentially again a photon hitting a proton that comes either from an electron or it comes from another nucleus at LHC, which I will get to. But you see this cross section grows with energy, right? The the higher the energy of the collision uh the larger this cross section. And this evolution which are basically these the all these curves here. Uh so this is the color glass condensate calculation uh with these fluctuations in black. They all kind of similar describes this pretty well after you tune a bunch of parameters that regulate the evolution speed, but you can find some consistency. Now uh let me I should have put this earlier, but I want to now go to the point where I want to use nuclear targets. Uh nuclear targets are good because uh they're bigger, there's more gluons in there, so I think I will reach the saturation regime more quickly. And that's something the EIC can do, the Electron Ion Collider, which is a high luminosity machine, uh much higher luminosity than this HERA machine, uh and it's it works for polarized electron proton and ion collisions in this this is the energy regime. And what's special about it? Well, first of all, it will be the only collider in the foreseeable future, the only new collider. Uh and it will collide electrons and protons and light nuclei that are all polarized, but it can also go to very heavy nuclei, right? From deuteron to uranium essentially. So then it has a wide kinematic coverage, and the science is among other things is gluon saturation, but it can also do imaging, like 3D imaging of my nuclei. It can do studies on proton spin and how it's distributed among its constituents. And it also can deal with, you know, understanding confinement, the origin of mass, etc. But gluon saturation is one of them. Uh and besides it will have a state-of-the-art general purpose collider detector. So many new advances. Here's a picture of it. Uh I can show you it is going to be located at Brookhaven Lab. This is New York City. This is Long Island. Uh this is where it's going to be. And it basically is a replacement of the Relativistic Heavy Ion Collider. So we will keep one of the rings of RHIC uh and will add uh electron rings, right? Two of them. I guess we keep the yellow ring of of of RHIC and then the rest will be electrons. And so this will allow us to do electron ion collisions uh in the future. And I said this, but this is an illustration of why nuclei are better to reach gluon saturation. This is called sometimes nuclear So the saturation scale is proportional to A to the 1/3. A is the mass number of the nucleus. Why is it 1/3? 1/3 Well, basically the volume scales like A, right? Uh but the radius then scales like A to the 1/3. And so because I'm hitting this thing that's moving very fast along this way, right? I have essentially A to the 1/3 nucleons behind one another. They add together. Uh they add their gluon distributions together, if you will. And so I increase the density of gluons uh by uh lining up all these nucleons behind one another and then squeezing them into this shock wave. And the larger QS, the more gluons are in the saturated regime, so it's easier to access saturation. And you see so for the same X, right? Uh I will have a smaller QS for protons. Essentially these lines tell you how large QS is. And I have a larger QS for a gold nucleus at the same X, essentially at the same energy. So it's cheaper because increasing the energy of a collider is very expensive, but just putting in a nucleus, is not for free, but it's much, much cheaper, especially at Rick, where we already have this capability of accelerating heavy nuclei. Okay, so I told you LHC can measure this too, and we can do something called ultra peripheral heavy ion collisions. So that's essentially heavy ion collisions that don't happen. So the the heavy ions don't collide, but what collides is the photon field of one of these ions, right? The charges that are moving, they will have a field around them. And the photon field of one of them will collide with the other one or vice versa. So we have a photon nucleus collision at LHC, and we also had them at Rick. And they have the the feature that these photons, they're basically quasi real photons, so their virtuality is essentially zero, but we can reach smaller x than at the EIC because the energies are so high. And so what do we get for saturation effects? Let's look at this. This is again my coherent diffractive cross section, now integrated over t again. I plot it now it's this function of this y. It's not so important what this is. It's basically a longitudinal momentum measure. And I can calculate what happens if essentially there's no no significant nuclear effect. So basically I just take the result I get in protons, and only change I do is sort of take into account that the shape is different of my target, right? It's a bigger target. Now I have lead nuclei, right? And you you saw that the t spectrum changes when I change the target. If we make it bigger, it drops more quickly. So this effect is taken into account in this curve, but nothing else. And if I include additional saturation effects, so the fact that my density doesn't grow as fast as in a proton because I'm already in the saturated regime because the nucleus is so dense, I get the green curve. You see that it's closer to the data, but it's still not describing the entire spectrum. So this is basically our calculation. And just to show you there's a lot more calculations on the market, okay? And none of them actually describe all the the whole range of data. Like some are better here in this region, but then they're not as good down here, etc. And some of them show the error bands, some of them don't. There's some nuclear parton distribution calculations, some shadowing calculations, etc. So there's many models built on different frameworks. And the saturation model is just one of them. And they all kind of don't work exactly great. One thing to show you is now this is again a T spectrum. So this is transverse momentum spectrum now in a gamma lead collision, again J psi production. If I don't include saturation effects, I get this dashed or dotted curve. If I include them, I get this. So I get a somewhat of an improvement of the data. And one interesting thing about, you know, related to the title of the talk is I'm actually if I now do a Fourier transform back to coordinate space, I can see that when I start from a standard nuclear shape, this is like a Woods-Saxon distribution that's usually used to describe nuclear density distributions, the dot dotted curve. If I do this full calculation, compute the cross section and then Fourier transform back, I can basically extract what the nucleus looks like at different X different longitudinal momentum fractions. And you see these effects of saturation are especially important in the very dense region, which is what you would expect, right? So you have a suppression from saturation effects. The other thing you see, it grows with X decreasing because I have these additional gluon emissions, right? So they kind of spread out, sometimes called re-bob diffusion or something like that. So the nuclear shape at high energy is something I can access in this, and it's affected by the presence of gluon saturation. Now we actually found quite a discrepancy between and you this was already indicated in this plot basically, but this not working. But now looking at the energy dependence again, you remember this plot from the proton result. We can also do this now in gamma lead collisions, but if you take our fit to the proton data and we make a prediction for the lead targets, we get this magenta curve. We don't doesn't really work. We have much larger suppression, right? If we fit directly to gamma lead, we can get the good fit, but then the proton data doesn't work, okay? So there's some tension between the data from LHC and from these ultra peripheral collisions and the old data from HERA or also LHC data of the proton targets. Now we can do a combined fit. The combined fit with any adjustments to the model gives us this yellow curve. So you see also there's tension, right? To kind of overestimate what's what's here at the larger energies. If you introduce an additional normalization factor everything, you can make it work, but it's a little bit I mean, you could argue this factor kind of takes into account uncertainties of the model. Like there's some uncertainties in this wave function where make the vector meson etc. But it's it's it's not a great thing because it actually it's a pretty large K factor unique. Alternatively, it could be that the data is a little bit underestimated by the fact that, you know, they have to identify if it is a diffractive event in the first place, right? And then LHC it's challenging because there could be effects where you have a diffractive event, but then because of electromagnetic dissociation because, you know, another photon might be exchanged they something may happen and you you actually fill your detector with stuff and the effect does no longer look diffractive. So there's a paper by here by these guys that said, well, I could correct for potential misidentification, then that will move my data from these open circles to the to the closed ones. And then there's actually no more K factor needed and we have resolved the tension between gamma P and gamma lead collisions. But this is something that would still be have to be confirmed by the experimental collaborations, right? I'm not going to say they're wrong. It's I'm just saying if you if this is true, our attention is gone. And the last thing I want to say, uh and I'm almost done, is we can also uh study deformed nuclei. So you might have heard uranium nuclei, for example. They have a large quadrupole deformation. So they look kind of like a cigar rather than a sphere. And that can be uh parameterized in this extended Woods-Saxon distribution. And what I just want to indicate here, if it's deformed, you automatically get a large increase of the fluctuations, right? Because you already get the fluctuations by rotating your target, right? And if it's spherical, it doesn't matter, right? It gives you no fluctuation, but if it's deformed and it rotates, right? You may hit uh hit it this way or this way, it will look different. And that indeed is what you see when you do the calculation with such a target. In particular, you look at the incoherent fluctuation sensitive cross section. In this regime, uh I'm plotting a non-deformed nucleus, then the blue is sort of the standard value for the quadrupole deformation of a uranium nucleus, and 0.5 is just something larger, just for illustration. And you see there's a pretty large effect, right? This is a log plot. So this is a a huge effect even in the range where and here you see this diffractive pattern uh where the coherent cross section kind of dips down. So this might be accessible uh for example at the EIC with uranium targets uh to actually measure are we sensitive to the nuclear deformation. So opens up measurements of nuclear structure at high energy, right? How the gluons are distributed in that. And we will learn, you know, whether that somewhat washes out more or that it's compatible with what we predict. We predict essentially that the deformation lasts at these small x values, right? We have 10 to the minus 3 here in x. Uh coherent is also modified. I won't go into it, but I just want to say you can scan through uh the t values. And if you include variations in the quadrupole, hexa, uh decapole, other octupole deformations, you can actually see that if you have the lower ones, they correspond to changes on longer length scales. So they at low T, low transverse momentum, whereas you when you increase these changes correspond to higher resolution changes. And you see the variations then happen at larger T and even larger T for these. So you can scan essentially through through these length scales of deformations of your nucleus by varying or looking differentially in the transverse momentum transfer. So this is just summarizing this that all of this you could access in one plot, right? You have deformation sensitivity, maybe some nucleon clustering, short range correlations, and then you can even go to the sub sub nucleon fluctuations that I talked about which would be at the larger T. There's many more observables. I don't have time, but I just want to say we can compute all kinds of different observables besides the diffraction. And I think to unequivocally declare the discovery of gluon saturation, we will have to have a combined picture of all of these and a comprehensive comprehensive framework, you know, to bring the theory up to the the latest next leading order calculations, etc. and put it all together and compared to data and do a global analysis. And this is what the so-called search collaboration is meant to do and is doing. And this leads me to my summary. Gluon saturation has been predicted by theory, has not been fully pinned down. That's what we're working on. We want to look at large nuclei at high energy. That's the most that's the best place to look. I focused on diffraction which I hopefully showed you is sensitive to gluon saturation, but also to the nuclear structure at high energy or small X. The broader challenges to well, many have many complementary measurements which are tied together hopefully by precision theory and the EIC, the Electron Ion Collider and the Large Hadron Collider at CERN are going to be complementary uh experiments to meet that challenge. Thank you very much. [applause] AND I LEFT NO TIME FOR QUESTIONS. I'M SORRY. [laughter] >> NO, PLEASE. >> Any questions for the speaker? No? I I actually actually have one. >> Sure. >> Uh let me turn on the lights. So I'm I'm an optical person. >> Okay. >> So it this to me feels like you're using something to probe something. It's kind of like what we do with the optical field, but we have to worry about sometimes to if we want to look at linear effect, then we want to keep our light fairly weak so we don't change the system >> [clears throat] >> Mhm. >> and its intrinsic property. I wonder if you have a equivalent thing that you have to worry about that you you don't want to probe it with >> Yeah, I mean it's it's already sort of in this plot, right, where you say um if I don't want to break it, right, uh I have to stay at low T and at low momentum transfer, right, because that's where you see the that's where the cross section dominates where it doesn't break up, so where I don't kind of mess up the target. You see if I'm interested in uh larger T, I have a problem because then the cross section for where the target breaks up actually dominates. >> Okay. >> it it it's an experimental challenge then to to to separate those. But yeah, indeed. So if I if I hit it very hard, I I have a process at large T, uh it's very hard to measure this uh coherent uh cross section where my target stays intact. Um so for example, if you're interested in where these dips are, right, which you saw for the nuclei, so for uranium I showed you, you can actually see sort of something equivalent to a diffractive pattern, right, to have like a diffractive dip. If I'm interested in where this is, uh I have to deal with this background of where actually do break up the target. So I guess that's sort of related uh to to your point of yeah. >> so the >> incoherent part you are basically already probing the effect of your whatever excitation that you're setting. >> basically yeah exactly so that's that's the one where I have kind of have to sum over all the possible outgoing states yeah. >> cool. >> Yeah. >> Pretty neat. >> [laughter] >> Yeah. >> Anything else? No? Okay let's thank Dr. Shaker. >> Okay thank you. >> [applause]

Assistir ao vídeo original no YouTube

Assuntos: Agentes de IA, AGI. Publicado no YouTube em 2026-11-13.

Seleção automática por assunto e idioma. O vídeo pertence ao canal de origem.

Comentários

Postagens mais visitadas deste blog

How to build a JARVIS AI Voice Agent for FREE | Full Livekit Tutorial (2026)

How Does AI Actually Work? 🤖 | AI, Machine Learning & Generative AI Explained