Pythia 8/Angantyr: Heavy-Ion Event Generator

• Pythia 8/Angantyr is a heavy-ion extension of the PYTHIA 8 event generator that extrapolates high-energy proton–proton dynamics to model pA and AA collisions.
• The framework integrates Glauber eikonal logic and wounded nucleon models to classify nucleon encounters and dynamically build event substructures.
• MPI and color reconnection mechanisms are tuned within Angantyr to accurately reproduce multiplicity distributions and transverse momentum spectra without invoking a QGP medium.

Pythia 8/Angantyr is the heavy-ion extension of the Pythia 8 event generator: a framework for building complete exclusive hadronic final states in high-energy proton–nucleus and nucleus–nucleus collisions by direct extrapolation of high-energy proton–proton dynamics, inspired by the Fritiof model and the wounded nucleon picture (Bierlich et al., 2018). Within PYTHIA 8.3, Angantyr is part of the heavy-ion collision machinery, and under certain circumstances multiple parton-level objects can be used for separate subcollisions and then combined for hadronization (Bierlich et al., 2022). In most of its phenomenological applications, it is used as a non-collective baseline: it retains standard PYTHIA ingredients such as hard and soft interactions, showers, hadronization, multiple-parton interactions, decays, and, depending on configuration, color reconnection, but it does not include an explicit QGP, hydrodynamic evolution, or jet quenching (Zuman et al., 2023).

1. Origins, scope, and position inside PYTHIA 8

The founding Angantyr model was introduced as a way to generate complete exclusive hadronic final states in high-energy nucleus collisions while staying as close as possible to the successful PYTHIA 8 description of $pp$ collisions (Bierlich et al., 2018). Its stated role is to bridge a large part of the gap between heavy-ion and high-energy physics phenomenology by extending PYTHIA’s perturbative hard processes, multi-parton interactions, showers, beam remnants, and Lund string hadronization to $pA$ and $AA$ systems without tuning to heavy-ion data (Bierlich et al., 2018).

The PYTHIA 8.3 manual places Angantyr inside the built-in heavy-ion collision machinery and describes it as a framework implemented within the code rather than as a separate generator (Bierlich et al., 2022). The manual gives a summary-level description rather than a full derivation: heavy-ion events are built from multiple subcollisions, multiple parton-level objects can be used for separate subcollisions, and these are then combined for hadronization (Bierlich et al., 2022). The same manual states an applicability domain for ion-ion collisions with ion geometries well described with a Woods–Saxon potential, for non-deformed nuclei with $A>16$ and $\sqrt{s_\mathrm{NN}}>10$ GeV (Bierlich et al., 2022). This suggests that the detailed model logic is meant to be read together with the dedicated Angantyr literature rather than from the manual alone.

2. Event construction and formal structure

At the level of nuclear geometry, Angantyr follows Glauber-style eikonal logic. For nuclei $A$ and $B$, with projectile nucleons at $\mathbf b_\mu$, target nucleons at $\mathbf b_\nu$, and overall nuclear impact parameter $\mathbf b$, the nucleus–nucleus $pA$0-matrix is written as

$pA$1

and, neglecting the real part of the amplitude, $pA$2 (Bierlich et al., 2018). For a fixed state, the absorptive probability in impact-parameter space is

$pA$3

which is the basis for deciding whether a nucleon–nucleon encounter is absorptive (Bierlich et al., 2018).

The model combines this geometry with Good–Walker diffraction and nucleon-state fluctuations. In the fluctuation model used in Angantyr, the nucleon interaction radius $pA$4 is sampled from

$pA$5

and diffractive excitation is generated by fluctuations in the scattering amplitude rather than by a separate external prescription (Bierlich et al., 2018). This fluctuation machinery matters because Angantyr does not only count encounters; it classifies them into absorptive, diffractive, and elastic classes, and it uses those classes when assembling the full event.

The wounded-nucleon logic appears explicitly through

$pA$6

with $pA$7 and $pA$8 the numbers of wounded projectile and target nucleons (Bierlich et al., 2018). Angantyr does not fit $pA$9 as an external emission function; it constructs the emission dynamically from PYTHIA subevents (Bierlich et al., 2018). This is one of its distinctive features relative to simpler participant models.

The event-building workflow is ordered and asymmetric. Candidate nucleon–nucleon interactions are sorted in increasing local impact parameter. If neither nucleon has interacted before, the encounter is a primary absorptive interaction and is generated as an ordinary PYTHIA non-diffractive minimum-bias event. If one nucleon has already interacted, the encounter is secondary absorptive and is modeled as a modified high-mass single-diffractive-like PYTHIA event. Diffractive interactions are then added, and the resulting partonic subevents are stacked into a single event and hadronized with the standard Lund string model (Bierlich et al., 2018). Within the PYTHIA 8.3 manual, this same architecture appears in condensed form as the use of multiple parton-level objects for separate subcollisions that are later combined for hadronization (Bierlich et al., 2022).

3. MPI, color reconnection, and the treatment of stacked subcollisions

A central technical issue in Pythia 8/Angantyr is how MPI and color reconnection behave once many nucleon–nucleon subcollisions are stacked into a single nuclear event. In the original heavy-ion use of Angantyr, color reconnections were previously only allowed within separate nucleon sub-collisions. The 2023 update to the QCD-based color reconnection model introduced an impact-parameter constraint on the allowed reconnection range, with the explicit goal of enabling more realistic color reconnections across different sub-collisions in Angantyr while respecting the short-ranged character of QCD color interactions (Lönnblad et al., 2023).

That update had immediate tuning consequences. In $AA$0 collisions, the new impact-parameter constraint changed the final state enough that the multi-parton interaction parameters in PYTHIA had to be retuned to reproduce minimum-bias data. In $AA$1 collisions, the introduction of global color reconnections reduced multiplicity, so Angantyr parameters had to be modified while keeping the $AA$2 tune fixed. With that done, the model could be extrapolated to $AA$3 without further parameter tuning while retaining a reasonable description of the basic multiplicity distributions (Lönnblad et al., 2023).

Independent tune scans in heavy-ion applications also established a more specific statement: in this framework, CR affects observables only when MPI is present. In Pb–Pb at $AA$4 TeV, the comparison of MPI+CR, CR off, MPI off, and MPI off + CR off showed that the MPI+CR setup best described multiplicity and $AA$5, while CR had essentially no effect if MPI was disabled (Singh et al., 2021). The same MPI dependence was found in Au+Au studies of the light-nuclei yield ratio $AA$6: turning CR on or off produced no effect if MPI was off (Zuman et al., 2023). Taken together, these results fix MPI as the activity-generating mechanism and CR as a reshuffling mechanism whose observable consequences are contingent on that activity.

4. Baseline character and heavy-ion phenomenology

Angantyr’s founding claim was that a direct extrapolation of high-energy $AA$7 collisions could already give a good description of general final-state properties such as multiplicity and transverse-momentum distributions in both $AA$8 and $AA$9 collisions (Bierlich et al., 2018). Later applications retained that baseline interpretation. In Au+Au studies of neutron density fluctuations and neutron–proton correlations, the framework is explicitly described as a good baseline for studying collisions in the absence of a QGP system, given its lack of flow and jet quenching (Zuman et al., 2023). In charged-multiplicity fluctuation studies, it is again presented as a baseline reference for $A>16$0–$A>16$1 multiplicity fluctuations in a framework that does not include a deconfined QGP medium or collective final-state effects (Bhowmick et al., 2024).

That baseline status does not imply trivial phenomenology. In Pb–Pb at $A>16$2 TeV, a dedicated study with PYTHIA8.235, $A>16$3, and MPI-based color reconnection reported that the charged-particle multiplicity and mean-$A>16$4 distributions are well explained with proper tuning, that MPI+CR is the preferred configuration, and that the same setup generates spectral hardening and baryon/meson enhancement around $A>16$5 GeV/$A>16$6 without hydrodynamics (Singh et al., 2021). The authors interpret these as collectivity-like signatures generated by microscopic event-structure mechanisms rather than by an explicit medium (Singh et al., 2021). This does not negate the baseline interpretation; it sharpens it by showing which observables are not uniquely diagnostic of a thermalized medium.

System-specific studies further map where Angantyr succeeds and where it does not. In Xe–Xe at $A>16$7 TeV, Angantyr was used as a non-medium reference for centrality- and spherocity-dependent identified-particle production; more central events were found to be more isotropic, more peripheral events more jetty-like, and identified-particle spectra showed low-$A>16$8 isotropic dominance and high-$A>16$9 jetty dominance (Singh, 2022). In O+O predictions at $\sqrt{s_\mathrm{NN}}>10$0 TeV, the same framework was characterized as a pp-based nuclear event generator with no collective effects between constituent sub-collisions, and the resulting identified-hadron spectra, $\sqrt{s_\mathrm{NN}}>10$1, and $\sqrt{s_\mathrm{NN}}>10$2 and $\sqrt{s_\mathrm{NN}}>10$3 ratios were systematically softer and less collective than those of EPOS4 and AMPT-SM; the abstract summarizes this as substantially weaker flow effects in Pythia 8 (Bashir et al., 12 May 2025). In identified-particle balance functions for Pb–Pb at 2.76 TeV, PYTHIA 8.3 + Angantyr with the Monash 2013 tune described peripheral collisions reasonably well but did not quantitatively reproduce central Pb–Pb data, leading the authors to argue for a dedicated heavy-ion tuning of the Angantyr framework (Gupta et al., 21 Apr 2026). A plausible implication is that Angantyr is strongest as a no-medium baseline in peripheral and moderately dense systems, while central heavy-ion observables that are strongly shaped by medium evolution remain outside its default reach.

5. Fluctuations, correlations, and hadronic-phase extensions

One large research line uses Pythia 8/Angantyr to establish fluctuation baselines. For net-proton fluctuations, the model has been used in Au–Au at 200 GeV and Pb–Pb at 2.76 TeV as a baseline where the formation of a thermalized medium is not assumed (Behera et al., 2019). In that study, the net-proton number is

$\sqrt{s_\mathrm{NN}}>10$4

the cumulants $\sqrt{s_\mathrm{NN}}>10$5–$\sqrt{s_\mathrm{NN}}>10$6 are evaluated explicitly, and radial flow is added only through an afterburner; the resulting conclusion is that radial flow has substantial impact on