Research - Agnieszka Sorensen, PhD

You can find all my publications on InspireHEP.

My research focuses on uncovering the properties of dense, strongly-interacting matter. These properties can be summarized in the QCD equation of state (EOS). The EOS is key for describing behavior of nuclear matter in extreme environments, such as supernovae, neutron stars and their mergers, and relativistic collisions of heavy nuclei, and it can address the following open questions:

Does QCD matter undergo a first-order phase transition at finite baryon density?
How does the isospin dependence of the EOS evolve with density and temperature?
How do nuclear matter degrees of freedom and their properties change as the thermodynamic variables are varied?

I study these questions by using dynamical simulations of relativistic heavy-ion collisions, developing flexible EOS meta-models, and performing wide-ranging comparisons to experimental data. In my work, I use both high-performance computing and machine learning (ML) methods to enable these analyses.

Currently, my group is working on the following projects:

Observable discovery with AI/ML (part of Genesis Mission)
As part of the inaugural cohort for the U.S. Department of Energy's Genesis Mission, our team explores using machine learning (ML) to discover new EOS observables in heavy-ion collisions. In particular, we are looking for observables that, while sensitive to the EOS, are at the same time insensitive to other, often poorly-constrained inputs to the collision dynamics (such as, e.g., the in-medium modification of the cross sections). Successful completion of this project will not only help to put robust constraints on the EOS, but open a new era of using novel approaches and architectures to analyzing data and/or emulating events.

This ongoing work is pursued together with MSU/FRIB graduate students Daniel Barr and Joshua Belieu, and collaborators from Duke University (Steffen Bass, Lance Lampert, Andrea Kirschner), UC Davis (Daniel Cebra, Mathias Labonte, Kate Matsubayashi, Andrea Pellot Jimenez), and LLNL (Indra Chakraborty, Ron Soltz).
Modeling the momentum dependence of the nuclear interaction
Momentum dependence is a crucial component of the nuclear interaction, measured in experiments colliding protons in heavy nuclei, without which the EOS extracted from comparisons of transport simulations to experimental data is strongly biased to exhibit more repulsive behavior. However, modeling the momentum dependence in hadronic transport can be challenging due to the inherent complexity, and thus numerical cost, of the relevant expressions. Because of this, implementations often use drastic simplifying assumptions, which is also the case for the recent implementation of the momentum-dependent potential in the hadronic transport code SMASH. In this project, we aim to restore the expected behavior of the momentum-dependent potentials currently implemented in SMASH, as well as to explore generalization of the approach to full Lorentz-covariance.

This ongoing work is led by MSU/FRIB graduate student Daniel Barr.
Extracting flow observables from detectors with complicated geometries
The collective behavior of matter produced in heavy-ion collisions is sensitive to the dense nuclear matter equation of state. However, traditional flow observables -- whch quantify this behavior -- can be difficult to extract from data collected using detectors with complicated geometries. This is because common approaches of computing flow require establishing the reaction plane of the collision, which can be problematic outside without a full solid angle coverage. In this work, we explore using the cumulant method of extracting flow at very low collision energies, such as those explored at NSCL and FRIB. In particular, we aim to apply the developed method to data collected in NSCL experiments E15190, E14030.

This ongoing work is pursued together with MSU/FRIB graduate student Samuel Brown and his Ph.D. advisor Kyle Brown.
Modeling fluctuations in the initial state of heavy-ion collisions
Boltzmann-Uehling-Uhlenbeck (BUU) transport frameworks model the evolution of the system's distribution function \( f(\mathbf{x}, \mathbf{p}) \) with the method of test particles, i.e., by evolving thousands of samples of \( f(\mathbf{x}, \mathbf{p}) \). This oversampling allows one to obtain excellent estimates of the local densities and, therefore, mean-field potentials. However, oversampling can also unphysically suppress density fluctuations related to the finite number of nucleons composing the colliding nuclei. In this project, we implement a practical algorithm to account for these fluctuations in two BUU frameworks: SMASH and DJBUU.

This ongoing work is pursued together with collaborators: Sangyong Jeon (McGill University, Canada) as well as Youngman Kim and and Kyungil Kim (Institute for Basic Science, South Korea).

My past projects include:

Locating the QCD critical point with finite-size scaling of proton cumulants
In this first analysis of experimental data to locate the QCD critical point in a range consistent with modern theoretical predictions, we study fluctuations of the number of protons produced in collisions of gold nuclei. By adjusting the portion of the collision we analyze (using proton fluctuations measured in different rapidity bin widths), we test how fluctuations scale with system size. At collision energies above \( \sqrt{s_{\rm{NN}}} = 7.7~\rm{GeV}\), we observe that the data follows scaling with a power law: something expected near a critical point. Our analysis indicates a critical point near a baryon chemical potential of \( \mu_B = 700 \pm 100~\rm{MeV} \). However, we also find that similar scaling behavior in hadronic transport simulations without a critical point. We show that this behavior can be traced to evaluating the finite-size scaling function along the freeze-out line, where scaling behavior can arise even in systems without criticality if the proton fluctuations are described by the Skellam distribution.

You can read this work here.
Speed of sound from neutron stars to heavy-ion collisions
In this project, we explored the compatibility of several neutron-star EOSs with a large peak in the speed of sound squared as a function of density, favored by observational data, with heavy-ion collision measurements. To convert from a neutron-star EOS to an EOS applicable in heavy-ion collisions, we explored a range of possible symmetry energy expansion coefficients and rejected coefficient sets which led to acausal, unstable, or unphysical behavior of nuclear matter. This in turn put a constraint on the high-density behavior of the symmetry energy, as can be seen in the figure. Using a flexible parametrization of density-dependent mean-field interactions, we then compared results using the converted EOSs against experimental data, and we found that a neutron-star EOS with a peak in the speed of sounds squared around twice the saturation density of nuclear matter is the most favored.

This work was published in Phys.Rev.C 109 (2024) 6, 065803. You can read it here.
Bayesian analysis of Beam Energy Scan FXT data
In this project, we used flexible density-dependent mean-field potentials in a Bayesian analysis of STAR measurements of golg-gold collisions at \( \sqrt{s_{\rm{NN}}} = \{3.0, ~4.5\}~\rm{GeV} \). The flexibility of the potentials allowed for a meaningful exploration of the behavior of the EOS at high baryon densities, leading to a Bayesian constraint that suggests a strong softening of the EOS -- and possibly even a first-order phase transition -- at densities between 3~and~4 times the nuclear saturation density (as indicated by the area with green vertical stripes in the figure). This nontrivial result at high densities was inaccessible in previous studies (see area with black horizontal stripes in the figure), which used potentials parametrized using only a single parameter: the incompressibility of nuclear matter at saturation, making it impossible to account for changes in the behavior of nuclear matter at high densities.

This work was published in Phys.Rev.C 108 (2023) 3, 034908. You can read it here.
Speed of sound from baryon cumulants in heavy-ion collisions
In this work, I connected cumulants of baryon number to the isothermal speed of sound squared \(c_T^2 \) and its derivative with respect to the baryon density. Using experimental data (red triangles in the figure) compared to a model (purple stars), I showed that the density-derivative of \(c_T^2 \) plateaus as collision energy is decreased (within the collider range of the BES, \(\sqrt{s_{\rm{NN}}} \geq 7.7~\rm{GeV}\)), but then dramatically increases at the HADES collision energy of \(\sqrt{s_{\rm{NN}}} = 2.4~\rm{GeV}\). This suggests a dramatic change in the behavior of the speed sound between these two energy regimes.

This work was published in Phys.Rev.Lett. 127 (2021) 4, 042303. You can read it here.
Flexible density-dependent EOS for hadronic transport
In this work I developed flexible, relativistically-covariant vector density functional (VDF) model of mean-field interactions. The model has been designed to allow for high flexibility of the EOS. For applications to hadronic transport simulations of heavy-ion collisions, the EOS at low to moderate densities is parametrized to reproduce the known properties of nuclear matter (such as the saturation density, binding energy, and the location of the nuclear liquid-gas critical point), while at higher density the EOS are fit to produce a postulated first-order phase transition. In the figure, I show coexistence (solid lines) and spinodal (dashed lines) regions for the nuclear-liquid gas phase transition (black) and several postulated phase transitions at high density (colorful lines) with varying location of the critical point and/or width of the coexistence region.

This work was published in Phys.Rev.C 104 (2021) 3, 034904. You can read it here.

You can find all my publications on InspireHEP.

Currently, my group is working on the following projects:

Observable discovery with AI/ML (part of Genesis Mission)

Modeling the momentum dependence of the nuclear interaction

Extracting flow observables from detectors with complicated geometries

Modeling fluctuations in the initial state of heavy-ion collisions

My past projects include:

Locating the QCD critical point with finite-size scaling of proton cumulants

Speed of sound from neutron stars to heavy-ion collisions

Bayesian analysis of Beam Energy Scan FXT data

Speed of sound from baryon cumulants in heavy-ion collisions

Flexible density-dependent EOS for hadronic transport