Energy dependence of the deformed nuclear structure at small-x Authors: Heikki Mäntysaari, Pragya Singh Published: 2025 ABSTRACT AbstractWe quantify the effect of high-energy JIMWLK evolution on the deformed structure of heavy (Uranium) and intermediate (Ruthenium) nuclei. The soft gluon emissions in the high-energy evolution are found to drive the initially deformed nuclei towards a more spherical shape, although the evolution is slow, especially for the longest distance-scale quadrupole deformation. We confirm a linear relationship between the squared eccentricity εn2 and the deformation parameter βn2 in central collisions across the energy range covered by the RHIC and LHC measurements. The applied JIMWLK evolution is found to leave visible signatures in the eccentricity evolution that can be observed if the same nuclei can be collided at RHIC and at the LHC, or in rapidity-dependent flow measurements. Our results demonstrate the importance of including the Bjorken-x dependent nuclear geometry when comparing simulations of the Quark Gluon Plasma evolution with precise flow measurements at high collision energies. Introduction Understanding the high-energy structure of protons and nuclei is of fundamental interest. This structure is a result of complex interplay between different QCD phenomena. At the the confinement scale, strong interactions are responsible for forming a color-neutral bound state out of quarks and gluons. At high collision energies, this structure is dominated by a large number of soft gluons emitted throughout the high-energy QCD evolution, which itself can be understood by weak coupling techniques. A simple pointlike probe that can be used to probe the internal structure of protons and nuclei is a virtual photon, emitted by a lepton in Deep Inelastic Scattering (DIS). Thanks to the precise electron-proton DIS measurements performed at HERA [1], the partonic content of the proton, as well as the proton geometry and how it depends on Bjorken-x, is relatively well known. In the next decade, the Electron-Ion Collider in the US [2] will enable similar studies with nuclear targets, for the first time in collider kinematics. One particular advantage of the EIC is its capability to accelerate almost any nuclei, making it possible to also study specific nuclei with interesting geometric structures. Nuclear geometry is also a crucial input for the studies of Quark Gluon Plasma (QGP) in relativistic heavy ion collisions. This is because the hydrodynamical evolution of the Quark Gluon Plasma (QGP) will transform initial state pressure (density) anisotropies into observable final state correlations [3]. Detailed knowledge of the initial nuclear geometry is especially important when studying collisions among deformed nuclei [4–17], such as U+U, Ru+Ru or Zr+Zr collisions at RHIC [18, 19] or Xe+Xe collisions at the LHC [20]. These experiments provide unique opportunities to probe the properties of the Quark Gluon Plasma in systems with vastly different geometries. Additionally, they bridge the gap between the low-energy and high-energy nuclear physics communities. For example, it was shown in Ref. [21] that the U+U collisions measured at RHIC provide evidence for the hexadecapole deformation of the 238U nucleus. Further possibilities to probe the nuclear structure in high-energy nucleus-nucleus collisions have been discussed e.g. in Refs. [22–24]. On the other hand, in Ref. [25] it was demonstrated that at the future EIC, diffractive processes in electron-nucleus scattering can probe quadrupole, octupole and hexadecapole deformations of the heavy nuclei. Before the EIC is realized, similar studies are also possible in ultra-peripheral collisions [26] as demonstrated e.g. in Refs. [27, 28]. The purpose of this work is to quantify how the deformed nuclear geometry changes with the collision energy, or with the rapidity at which the measurement is performed. This evolution we calculate within the Color Glass Condensate (CGC) [29] framework which provides a convenient method to describe QCD in the high energy (high parton density) domain. Within the CGC, the energy (or equivalently rapidity) dependence of the nuclear structure is obtained by solving the JIMWLK [30–34] high-energy evolution equation. Furthermore, using eccentricities as a proxy for flow measurements, we quantify the expected effects of the JIMWLK evolution on the collectivity measurements in nucleus-nucleus collisions. We argue that a proper interpretation of the flow measurements probing the Quark Gluon Plasma properties requires a detailed understanding of the energy-dependent nuclear geometry. Energy dependent nuclear geometry At high energies where parton densities are very large, it is convenient to describe the nucleus as a dense color field. The nuclear structure can be probed by a simple, perturbatively understandable probe, which we take to be a small qq¯ dipole. As such, this corresponds to a DIS process in the dipole picture where the photon splits to a dipole long before the scattering. Unlike in an actual experiment, for the purposes of this work, we consider a fixed-size dipole and fix the orientation of the deformed nucleus in order to probe (or define) the energy-dependent nuclear density. The dipole-nucleus scattering we calculate within the CGC approach. The nuclear (deformed) high-energy structure at the initial condition of the small Bjorken-x evolution, taken to be x=x0, is described as in the IP-Glasma [35, 36] framework applied recently in a similar context e.g. in Refs. [22, 24, 25, 28, 37]. We first sample the nucleon positions (in 3d) from the Woods-Saxon distribution where the effective radius depends on the angle Θ measured from the symmetry axis of the nucleus: Here Yi0 are spherical harmonics, and we have oriented our coordinate system such that, in the case of quadrupole deformation β2>0, the long axis corresponds to Θ=0. The skin depth a is nucleus-dependent, and taken to be independent of x in this work. We note that in principle the JIMWLK evolution can also affect that, but varying a had negligible effect on the deformation parameters. The normalization constant ρ0 is set such that ∫d2bTA(b)=A. Given the sampled nucleon positions, we next proceed to evaluate the small-x color field of the nucleus by applying the impact parameter dependent generalized McLerran–Venugopalan (MV) model [38–40]. Within the MV model, the color charge ρa is assumed to be a local random Gaussian variable with a correlator Here g is the QCD coupling. The generalized MV model incorporates finite transverse extent by introducing transverse coordinate dependence into the MV model parameter μ(x)=∫dx-λA(x-,x) controlling the color charge density. The original MV model corresponding to an infinite nucleus is recovered when the color charge density g2μ(x)=constant. In the IP-Glasma framework [35, 36], the color charge density is related to the local saturation scale Qs(x) as cg2μ(x)=Qs(x), and we use the default value c=0.8 for the proportionality constant [41]. In order to sample the color charges, it is then necessary to determine the local saturation scale Qs2(x). In the IP-Glasma model, this is extracted from the IPsat parametrization [42, 43] (fitted to the HERA structure function data) as a function of the local thickness function T(x) as Qs2(x)∼T(x), which itself is obtained by summing over the individual nucleon thickness functions Tp,i. In particular, we have where is the nucleon thickness function. Here, B=4GeV-2 controls the nucleon size, and xi represents the transverse coordinates of the nucleons sampled from the Woods–Saxon distribution (1). Once the color charge distribution ρ(x-,x) is sampled, one can solve for the Wilson lines in the fundamental representation that describe the eikonal propagation of a quark within the target color field: Here, P- refers to the path ordering, ta are SU(3) generators in fundamental representation, and an infrared regulator m~=0.2GeV is included in order to regulate the Coulomb tails, and Y=lnx0x is the evolution rapidity. The dipole-nucleus scattering amplitude can then be written in terms of the Wilson lines as The average ⟨O⟩ refers to the average over possible target configurations, i.e., average over nucleon configurations and different color charge distributions. In practice, we use a lattice implementation following the discretization procedure outlined in Ref. [44]. The Bjorken-x dependence of the Wilson lines is obtained by numerically solving the JIMWLK evolution equation [45] (see also Ref. [46] for a recent development on including the JIMWLK evolution in the IP-Glasma). To regulate Coulomb tails in the evolution, the evolution kernel describing the gluon emission is modified at large distances following Refs. [47, 48] as Here K1 is a Bessel function and we use m=0.2 GeV as a regulator (which generically should be of the order of the QCD scale ΛQCD), and a fixed coupling αs=0.15 is used in the JIMWLK evolution. In this work, we do not aim to determine optimal values for the model parameters describing confinement scale physics (see however, related works [37, 48]), but use these phenomenologically motivated values. We also note that although m and m~ both suppress Coulomb tails, their particular values depend on the chosen parametrizations and as such there is no clear connection between the two. The conclusions of this work can be expected to be independent on the actual values chosen for these parameters. Generically running coupling effects should result in faster evolution at large distance scales, and as such the running coupling corrections can be expected to enhance the energy-dependence of the deformed geometry obtained in this work. This is because in order to obtain the same evolution speed for an observable such as the total DIS cross section at high Q2, the typical running coupling values should be close to the chosen fixed coupling reference. However, the geometry evolution takes place at longer distance scales where the running coupling is typically larger. The initial x0 is chosen to correspond to RHIC kinematics in this work, in practice x0∼0.01. Results for the LHC are then obtained after the JIMWLK evolution. This choice for the initial x0 is not unique. However, the main focus of this work is to quantify the geometry evolution between the RHIC and LHC energy ranges, and as such, the point at which the JIMWLK evolution is initialized has only a small effect on our results. If the evolution was initialized at a higher x0 above the x range probed at RHIC, then there would be some geometry evolution before one reaches RHIC kinematics, but the energy evolution from RHIC to LHC would remain almost identical. Although the dipole-nucleus amplitude is a clean probe of the nuclear high-energy structure, and we can use it to characterize the local density, its impact parameter dependence is not directly observable.1 More importantly, in an actual DIS process, it is not possible to control the orientation of the deformed nucleus. As such, our strategy in this work is to first use the dipole-nucleus scattering to extract the JIMWLK-evolved nuclear geometry. Then we study eccentricities in nucleus-nucleus collisions. As eccentricities are strongly correlated with the measurable final state particle flow harmonics [3], this provides us a method to determine the sensitivity of the flow measurements on the energy-dependent nuclear geometry. As we will demonstrate in this work, there is a clean correlation between the eccentricities and the extracted deformation parameters, demonstrating that eccentricity (flow) measurements can probe the x-dependent deformed nuclear structure. Eccentricities directly reflect the shape asymmetries in the initial-state geometry of heavy-ion collisions [13, 50–53]. We adopt the standard approach to characterize the event geometry using eccentricities, defined as: where ϕx is the azimuthal angle of the vector x and Tττ is the energy density. The average ⟨·⟩ here corresponds to the calculation of the root-mean-square. Note that here ϕx can be defined with respect to an arbitrary axis, as εn is an intrinsic property of the system. The averaging ⟨O⟩ again refers to nucleon position and color charge averages. The energy density is obtained as follows. Immediately after the collision (at τ=0+) in the forward light cone the non-zero components of the gauge fields read Here A and B refer to the two colliding nuclei, and the Wilson lines implicitly depend on evolution rapidity. As we consider midrapidity kinematics, the Wilson lines for both nuclei evolved to the same rapidity. Discretized lattice representation of the initial conditions (10) and (11) is used as an input for the classical Yang–Mills (CYM) equations of motion solved on the lattice up to a time of τ=0.2fm/c [54, 55]. At this point, we evaluate the energy–momentum tensor Tμν, and obtain the energy density as Here y=0 corresponds to the midrapidity kinematics. We consider collisions at zero impact parameter b=0 (corresponding to most central collisions) to minimize geometric asymmetries, enabling nuclear deformation to dominate the resulting observables. We, however, perform random 3d rotations for the nuclei before the collision (i.e. the long axes of the nuclei are not parallel but point in random directions), as experimentally it is not possible to align the colliding nuclei (see however Ref. [56] in the context of polarized light nuclei). We calculate eccentricities at fixed Bjorken-x, corresponding to midrapidity kinematics and particle production at fixed transverse momentum. This is in contrast to IP-Glasma, where one typically uses a dynamically determined b-dependent x defined as x=Qs(x,x)/s [10], where s is the nucleon-nucleon center-of-mass energy [10]. As the saturation depends on the local thickness function, in this setup the Wilson lines are evaluated at the evolution rapidity Y that depends on transverse coordinate x and has a small effect on the apparent geometry. Instead of this, we choose to evaluate Wilson lines at a fixed x when constructing the initial condition for the JIMWLK evolution (using the useFluctuatingx 0 option in IP-Glasma [41]). Furthermore, the fixed-x setup allows us to more directly quantify the effects of the JIMWLK evolution on the nuclear shape. The difference between the two setups is formally higher order in αs, and as the evolution is only logarithmic in x, only moderate effects are expected. However, we note that the density profiles extracted in Sect. 3 can be described by the Woods–Saxon distribution (1) only when the Wilson lines are evaluated at fixed Bjorken-x. Results We calculate the forward elastic dipole-nucleus scattering amplitude (7) for a fixed size |r|=0.1 fm dipole (integrated over dipole orientation) as a function of impact parameter b at different azimuthal angles θ. The angle θ is measured from the long symmetry axis of the nucleus that is chosen to lie on the transverse plane. As per the optical theorem, this is directly proportional to the cross-section for the fixed-size dipole-target scattering. In the case of a small dipole considered in this work, the dipole amplitude is not highly sensitive to saturation effects and can be considered to be directly proportional to the local transverse density. This corresponds to assuming TA(b)∼Qs2(b) consistently with Eq. (3), which is also a typical estimate used to model the impact parameter dependence of the nuclear high-energy structure, see e.g. Refs. [42, 57]. At each θ, we fit a Woods–Saxon distribution (1) integrated over the longitudinal coordinate (with a necessary normalization factor) to the impact parameter dependence of the scattering amplitude to obtain the effective radius R(θ). Finally, the deformation parameters βi are obtained by fitting the decomposition (2) to the extracted radii. Uranium We begin our analysis by considering the 238U nucleus, which has a well-understood low-energy structure with a highly deformed geometry [58, 59], and is also the heaviest nucleus used in collider experiments. We construct Uranium nuclei using the Woods–Saxon parameters RWS=6.81 fm and a=0.55 fm, with the deformated structure described by β2,WS=0.28,β3,WS=0 and β4,WS=0.093 from Refs. [58, 59]. Here the subscript WS refers to the fact that these parameters enter the Woods–Saxon distribution (1). As discussed in Refs. [60, 61], the multipole moments βlm,WS of the Wood–Saxon density (in this work we only consider m=0) do not exactly correspond to the deformed geometry of the constructed nucleus βlm, as the former represents the entire nuclear volume, while the latter describes surface deformation. Additionally, the shape parameters βlm,WS can mix at higher orders, meaning that, for example, β4,WS can contribute to the generation of β2, reflecting the interplay of harmonic components. We perform a JIMWLK evolution over Y=4.8 units of rapidity, corresponding to evolution from x=0.01 to x=0.01·e-4.8≈8·10-5, which covers the energy range from RHIC to LHC kinematics. Although there are currently no plans to perform U+U collisions at the LHC, this analysis allows us to quantify the importance of the geometry evolution between the two energy scales for any heavy nucleus. Furthermore, we emphasize that the small-x evolution over a comparable range is also probed by the rapidity dependence of e.g. flow measurements. This is because in such processes one probes the nuclear structure at x∼⟨pT⟩/se±y, where ⟨pT⟩ is the typical transverse momentum of the parton produced at rapidity y. Similar evolutions have been visualized for individual proton and nuclear configurations in Refs. [22, 48, 62]. The effective nuclear radius as a function of angle θ (with respect to the axis aligned along the long axis of U) is shown in Fig. 1. The results are shown both at the initial condition (Y=0), and at the maximum evolution rapidity after Y=4.8 units of rapidity evolution. The nuclear radius is seen to grow towards small-x. When moving from RHIC to LHC energies, this growth is approximately 5%. This is comparable to the growth obtained in Ref. [37] and required to describe the ALICE data for the exclusive J/ψ photoproduction in ultra peripheral collisions [63]. Angle-dependent radius for the Uranium at the initial condition Y=0 and after Y=4.8 units of rapidity evolution. The radius parametrized as Eq. (2) fitted to this data is also shown We fit the function (2) to the calculated radii and extract the deformation parameters βn. This fit is also shown in Fig. 1 for comparison. The fit quality is found the be excellent at all rapidities. We choose to include coefficients up to β4. Although lower harmonics βn,WS can mix and in principle generate n>4 harmonics in the density profile, these higher-order coefficients turn out to be so small that they can not be reliably extracted numerically. In this work we choose to include the minimum number of deformation parameters required to obtain a good and stable fit to the calculated R(θ) data. The fit stability is determined by requiring that the obtained βi coefficients are not strongly modified if the fit is performed over a limited angle of θ. We extract the deformation parameters βn separately at each evolution rapidity Y. The rapidity dependence of the extracted parameters is shown in Fig. 2. The JIMWLK evolution is found to result in a smoother nucleus with increasing evolution rapidity (decreasing x), which manifests itself as β2 and β4 approaching zero. The β3 is approximately compatible with zero at all rapidities, i.e., it is not generated by non-zero β2,4,WS at the initial condition or by the JIMWLK evolution. This geometry evolution is, however, only modest in the studied evolution range. For example, β2 decreases only by ∼1.3% from RHIC to LHC energies. For the β4, which is sensitive to shorter distance-scale structure where the JIMWLK evolution kernel (8) is less suppressed, the evolution is somewhat faster, with β4 decreasing by ∼10% over the same energy range. This weak evolution is compatible with the previous studies of Uranium, Neon and Oxygen nuclei in Refs. [22, 25], showing that the non-trivial shape is not significantly altered by the small-x evolution. Physically, this is because confinement effects suppress the evolution at distance scales much larger than the nucleon size (or 1/m in Eq. (8)), which is the scale at which deformations in the nuclear geometry become relevant. On the other hand, in Ref. [64] it has been shown that at the nucleon substructure level, the nucleus becomes much smoother as a result of JIMLWK evolution, which has a numerically large effect on the incoherent J/ψ production in γ+Pb scattering. Similarly, a significant geometry evolution for individual protons with non-trivial initial geometry has been reported in Refs. [47, 48, 65]. This suggests that the small-x evolution has a moderate effect on the nuclear geometry at the largest length scales, but can have a more pronounced effect at the length scales comparable to and shorter than the nucleon radius. The obtained values of β2 (left) and β4 (right) deformation parameters as a function of rapidity for the Uranium. The grey line shows the corresponding value at the initial condition in the Woods-Saxon distribution. The result for β3 (not shown) is small and approximatively compatible with zero Next, we determine the relation between the deformed geometry of the Uranium nucleus and the initial state eccentricities in U+U collisions. As discussed in Sect. 2, we consider U+U collisions at central impact parameter b=0, but perform random 3d rotations for both nuclei. As such, this in principle corresponds to actual heavy ion collisions that can be studied in collider experiments (although in U+U collisions the relationship between the impact parameter and different centrality estimators can be complex [52]), unlike the dipole-nucleus scattering amplitude studied above. The correlations between the squared eccentricities εn2 and deformation parameters βn2, extracted by considering a small dipole-nucleus scattering as discussed above, are shown in Fig. 3 for the second and fourth order coefficients. For the third-order coefficient, the results are compatible with zero. Interestingly, although eccentricities correspond to the case where the Uranium nuclei have random orientations, the JIMWLK evolution has numerically a comparable effect on both eccentricities and on deformation parameters. Eccentricity as a function of nuclear deformation parameter in U+U collisions. The line shows the best fit of the function (14). The numerical data corresponds to the rapidity evolution from Y=0 to Y=4.8 We find the linear relation suggested in Ref. [12], to be very accurate in central collisions. Although εn is computed from the full transverse profile, the anisotropy is dominated by peripheral regions where gradient variations are largest (driving also the development of the measured anisotropic flow). Consequently, εn remains sensitive to nuclear shape deformations that affect mainly the edge of the nucleus. The fitted coefficients a and b are shown in Table 1. The an coefficients are found to be very small, as expected, as in the βn→0 limit non-zero contribution to the (mean square) eccentricity originates from random nucleon positions and as such scales as 1/A. The correlation between β4 and ε4 is found to be stronger than in the n=2 case. These correlation results demonstrate that the Bjorken-x dependence of the deformed nuclear geometry is visible at the level of eccentricities, and as such expected to influence precise flow measurements. This also suggests that the rapidity dependence of the flow measurements can be sensitive to the small-x evolution. Table 1Fitted coefficients quantifying the correlation between the squared eccentricities and deformation parameters as defined in Eq. (14)n=2n=4an0.008±0.0090.006±0.002bn0.291±0.1471.124±0.316 Fitted coefficients quantifying the correlation between the squared eccentricities and deformation parameters as defined in Eq. (14) We also note that the non-linear dynamics included in the JIMWLK evolution can be expected to play a more important role at the level of eccentricities than in the case of small dipole-nucleus scattering studied above. In order to define the geometry, we focused on a scattering process with a dilute probe, but such a dilute limit is not taken when eccentricities are calculated. As such, it is non-trivial that the linear relationship (14) between the eccentricities and deformation parameters holds also when non-linear saturation effects are taken into account. Ruthenium In addition to the heaviest nuclei, there is also a growing interest to probe intermediate-A nuclei [16, 66, 67]. In order to study geometry evolution in the case of intermediate nuclei, we consider Ruthenium in this work. One additional motivation for this study is the fact that the JIMWLK evolution can be expected to have a pronounced effect on the geometry of smaller light nuclei. Additionally, on a phenomenological level, a precise knowledge of the deformed structure of 96Ru and 96Zr is essential due to their involvement in high-energy collisions at RHIC, aimed at probing local strong parity violation. Deviations from unity in the ratio of observables for 96Zr+96Zr and 96Ru+96Ru collisions are predominantly attributed to differences in their radial profiles and intrinsic deformations [18]. Precise characterization of these deformations is crucial for interpreting experimental data and isolating potential parity-violating effects. There is only quadrupole deformation in Ruthenium, i.e. β2,WS=0.158, and β3,WS=β4,WS=0 [60]. The other parameters for the Woods–Saxon distribution are R=5.085 fm and a=0.46 fm. We only achieve a stable fit to the R(θ) distribution when including the β2 coefficient and setting the higher-order coefficients to zero. An excellent description of the R(θ) distribution with only one free deformation parameter β2 is illustrated in Fig. 4. Consequently, we only extract β2 for Ruthenium and conclude that the other deformation parameters are negligible. We note that although there are large fluctuations in the calculated R(θ) data close to θ=0 and θ=π/2, the obtained β2 or radius R are not significantly affected if that part is not included in the fit. Angle-dependent radius for the Ruthenium at the initial condition Y = 0 and after Y = 4.8 units of rapidity evolution Similarly, as in the case of the Uranium, the Ruthenium radius increases by approximately 5% when evolving from RHIC to LHC energies. The rapidity dependence of the β2 deformation parameter is shown in Fig. 5. The JIMWLK evolution is again found to drive the nuclear geometry more towards a spherical shape. The β2 coefficient now decreases by ∼5% in the considered energy range, which is a much larger effect than what was seen in the case of Uranium. This stronger x dependence in β2 we attribute to the smaller size of the Ruthenium nucleus which is more effectively modified by the emission of soft gluons throughout the JIMWLK evolution. The values of β2 as a function of rapidity, derived from the fit to data using the Ruthenium setup The correlation between the quadrupole deformation β2 and the eccentricity ε2 is shown in Fig. 6. Again, we perform random rotations for the Ru nuclei before the collision and consider only the central b=0 events. Similarly, as in the case of the Uranium, we find a perfect linear correlation between ε22 and β22, with the coefficients shown in Table 2. This confirms that the linear relation (14) is valid for both intermediate and heavy nuclei with deformed geometries in central collisions. We note that the (mean square) eccentricity cannot become negative, and physically one requires an≥0. Our numerical result is consistent with this condition, indicating that the result extrapolated to β2→0 limit is consistent with zero. Compared to Uranium, a stronger correlation between β2 and ε2 is found. This again suggests that a precise description of the x dependent geometry is crucial when confronting simulations with precise collider data. Eccentricity in Ru+Ru collisions as a function of the nuclear deformation parameter. The line represents the best fit to the Ruthenium data. The numerical data covers the evolution rapidity domain from Y=0 to Y=4.8 Table 2Coefficients for Ruthenium for n=2n=2an-0.01±0.008bn1.376±0.343 Coefficients for Ruthenium for n=2 Conclusions We have calculated the small-x evolution of the deformed nuclear geometry, focusing on both heavy (Uranium) and intermediate (Ruthenium) nuclei. The energy dependence of the nuclear geometry is obtained by numerically solving the JIMWLK evolution equation. The JIMWLK evolution is found to drive the nuclear shape towards a larger and more rotationally symmetric form, decreasing the deformation parameters βn as defined in Eq. (2). This evolution is, however, found to be rather slow, having only a few percent effect for the dominant quadrupole deformation parameter β2 for intermediate and heavy nuclei, when evolving from RHIC to LHC energies. The evolution is stronger in the case of smaller nuclei or higher-order deformations. This can be understood to result from the fact that the confinement scale effects effectively included in the JIMWLK kernel limit the relevant length scales in the evolution to not significantly exceed the nucleon radius. The exact values for the deformation parameters and eccentricities at the LHC energy depend on the free parameters describing the initial state and regularization of Coulomb tails. These parameters can in principle be determined from fits to experimental data [37, 48, 68]. However, currently, most fits come with no uncertainty estimates and cannot simultaneously describe e.g. the DIS structure function and vector meson production data [48]. As such, exact values for e.g. eccentricities or deformation parameters at the LHC energy extracted in this work can not be considered to be precise estimates. Instead, the main observation is that there is a small but non-negligible (comparable to the experimental precision) geometry evolution that needs to be included when describing bulk observables across the available energy range. In this work, a fixed coupling evolution is used, and consequently, our results can be seen to correspond to a lower limit for the effect of the JIMWLK evolution on the deformed geometry. This is because including running coupling effects would enhance the evolution at long-distance scales relevant for deformation, see also discussion in Ref. [48]. We furthermore demonstrated that, for both Uranium and Ruthenium, the linear relation εn2=a+bβn2 connecting the eccentricities and deformation parameters, is valid in central collisions. In particular this relation is found to be valid even when the deformation parameters are extracted in the dilute limit, but no such limit suppressing non-linear effects is taken when eccentricities are evaluated. This demonstrates that the deformed energy-dependent geometry leaves visible signatures in central nucleus-nucleus collisions even when nuclei undergo a random rotation before the collision as in an actual experiment. Given the high precision of the flow measurements performed at RHIC and at the LHC, a precise understanding of the energy-dependent geometry is crucial to understand the accurate flow data, especially in the context of light ion collisions [66]. This has been recently demonstrated e.g. in Ref. [69]. In addition to the center-of-mass energy dependence, comparable signatures of the JIMWLK evolution in the nuclear geometry can be expected to be visible in rapidity-dependent flow observables. In the future, it would be interesting to both include the running coupling effects and to determine the signatures of the JIMWLK evolved geometry at the level of actual flow measurements in e.g. experimentally accessible multiplicity bins. Acknowledgements This work was supported by the Academy of Finland, the Centre of Excellence in Quark Matter (project 346324), and projects 338263, 346567 and 359902, and by the European Research Council (ERC, grant agreements No. ERC-2023-101123801 GlueSatLight and No. ERC-2018-ADG-835105 YoctoLHC). The content of this article does not reflect the official opinion of the European Union, and responsibility for the information and views expressed therein lies entirely with the authors. Computing resources from CSC–IT Center for Science in Espoo, Finland and from the Finnish Computing Competence Infrastructure (persistent identifier urn:nbn:fi:research -infras-2016072533) were used in this work. Data Availability Statement This manuscript has no associated data. [Authors’ comment: Numerical results can be reproduced using publicly available code [41].] Code Availability Statement Code/software will be made available on reasonable request. [Authors’ comment: Results were obtained using a publicly available code [41].] Footnotes We however note that that the impact parameter dependence is directly related to the momentum transfer dependence in exclusive processes [26, 49]. References Associated Data Data Availability Statement This manuscript has no associated data. [Authors’ comment: Numerical results can be reproduced using publicly available code [41].] Code/software will be made available on reasonable request. [Authors’ comment: Results were obtained using a publicly available code [41].]