REVIEW
Comparison of Methods for Extracting Three-Dimensional L\'evy Source Parameters in $\sqrt{s_{\text{NN}}} = 200\text{ GeV}$ Au+Au Collisions with EPOS3
Nuclear Theory · Submitted 5 Nov 2026 · 25 references
Pith's verdict: The FFT-equivalence result is clean, but the FFT fit as specified pins λ≈1 at q=0 while the plotted λ≈0.66 come from a different grid; the joint labels need fixing before the comparison is trustworthy.
T0 · AI review
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
Direct integration and phase-corrected 3D FFT of the pair-source distribution are the same discrete Fourier sum: equal to RMSE $< 10^{-14}$, identical L\'evy fits, and 1.3 seconds instead of four hours per bin.
The reading
What carries the argument
The load-bearing object is the pair-source distribution $D(\boldsymbol{\rho})$, the autocorrelation of the single-particle emission function, whose Fourier transform is the correlation function through the Koonin--Pratt relation $C_2(q) = 1 + \tilde{D}(q)/\tilde{D}(0)$. In the longitudinal comoving system the four-dimensional phase $q\cdot r$ reduces to $-\mathbf{q}\cdot\boldsymbol{\rho}$, with a three-dimensional reduced separation $(\rho_{\rm out}, \rho_{\rm side}, \rho_{\rm long})$ that folds temporal emission duration into the out direction; this reduction is what makes the problem a plain 3D spatial Fourier transform. The methodological core is the phase-corrected FFT: after accumulating $D(\boldsymbol{\rho})$ on a $301^3$ grid with random $\pm\boldsymbol{\rho}$ sign assignment to enforce reflection symmetry, the shift theorem's phase $\varphi = \mathbf{q}\cdot\boldsymbol{\rho}_0/\hbar c$ is rotated out of the complex FFT output before the real part is taken. Because the resulting discrete sum is identical to the Riemann sum of the Koonin--Pratt integral, the two numerical routes coincide to floating-point precision, and the FFT's $O(N^3 \log N)$ scaling replaces the integration's $O(N_\rho^3 N_q^3)$ cost. Fits then use the analytical L\'evy form $C_2(q) = 1 + \lambda\exp\left(-(q_{\rm out}^2 R_{\rm out}^2 + q_{\rm side}^2 R_{\rm side}^2 + q_{\rm long}^2 R_{\rm long}^2)^{\alpha/2}\right)$, whose autocorrelation-preserving property is what makes the L\'evy parametrization self-consistent.
Extended reading notes
Core claim
The paper's central claim is that for interaction-free femtoscopy of transport-model output, the two-particle correlation function can be computed directly from the ensemble-accumulated pair-source distribution $D(\boldsymbol{\rho})$ -- the relative-coordinate histogram of all accepted same-event pion pairs, summed over the whole centrality class -- and that the natural way to do it is a phase-corrected 3D FFT. Because the histogram's origin sits at the first bin edge rather than at $\boldsymbol{\rho}=0$, the Fourier shift theorem inserts a momentum-dependent phase $\varphi = \mathbf{q}\cdot\boldsymbol{\rho}_0/\hbar c$; rotating that phase out of the complex FFT output exactly recovers the physical transform. The identity that carries the argument is that this phase-corrected discrete Fourier transform is algebraically the same sum as the direct Riemann evaluation of the Koonin--Pratt integral, verified numerically at RMSE $< 10^{-14}$ in all eight $K_T$ bins with identical fitted parameters and a speedup from over four hours to about 1.3 seconds per bin. On that basis the paper claims the integration/FFT route is a faster and smoother alternative to CRAB for interaction-free studies, while acknowledging that the FFT shortcut disappears once Coulomb or strong final-state interactions enter the kernel, and that the uncertainties of the new method are strongly correlated because every bin comes from the same histogram. It also reports a systematic discrepancy between the two spaces -- momentum-space fits yield $\alpha$ values about 7--9% lower than event-averaged coordinate-space source fits, with consistent orderings in $R_{\rm side}$ and $R_{\rm long}$ -- and attributes the shift to either residual non-L\'evy features or the event-averaged versus ensemble-accumulated comparison, which it states it cannot yet disentangle.
Load-bearing premise
The comparison stands on treating the earlier coordinate-space fits to event-averaged pair sources as the benchmark for parameters that the new method fits from one source summed over all events; because the paper states that a mixture of L\'evy distributions with different scales fits with a lower effective $\alpha$, the 7--9% $\alpha$ gap it reports is only a real method discrepancy if that averaging artifact does not account for it.
Editorial extensions
If this is right
- For interaction-free femtoscopy of transport-model output, the correlation function becomes a seconds-scale computation per $K_T$ bin instead of a multi-hour one, so fine $K_T$ binning and 3D fits on large simulated datasets become routine.
- The FFT/integration route removes CRAB's event-pooling and pair-sampling fluctuations, giving smooth transverse-mass trends with much smaller, though strongly correlated and not statistically interpretable, uncertainties.
- The reported 7--9% lower $\alpha$ in momentum space is now a controlled discrepancy: with identical particles, cuts, and binning on both sides, its resolution must come from either a non-L\'evy source model or from the event-averaged versus ensemble-accumulated comparison, both of which the paper identifies as candidates.
- The FFT shortcut is conditional on the plane-wave kernel: once Coulomb or strong final-state interactions are included, the kernel is no longer a pure Fourier phase and the method reverts to direct integration at several hours per bin, while CRAB retains the ability to model those interactions.
Reading between the lines
- Because the two Fourier routes agree to $10^{-14}$, any future disagreement between femtoscopy methods applied to the same simulated source is attributable to source modeling or fitting choices, not to the integration machinery -- the numerical question is closed.
- A closure test the paper does not run would settle its open discrepancy: fit $D(\boldsymbol{\rho})$ event by event and average the fitted parameters, then compare with the fit to the summed source; if the 7--9% $\alpha$ gap closes, the reported shift is a L\'evy-mixture averaging artifact rather than physics of the source.
- If the mixture effect does explain the shift, comparisons of model predictions with experimental L\'evy-$\alpha$ measurements should be matched on identical averaging footing, or the comparison inherits the same systematic bias.
- The same phase-corrected FFT construction transfers to any correlation whose kernel is a plane-wave phase, for example identical-boson correlations in other collision systems or on finer $K_T$ grids, wherever complete freeze-out coordinates are available from a model.
Original abstract
Bose--Einstein femtoscopy probes the spatiotemporal geometry of the particle-emitting source in heavy-ion collisions, where experimental data favor a L\'evy-stable source description. We compare four approaches for extracting 3D L\'evy source parameters using $\pi^+\pi^+$ pairs from EPOS3-simulated Au+Au collisions at $\sqrt{s_{\text{NN}}} = 200\text{ GeV}$: direct fits to the pair-source distribution $D(\boldsymbol{\rho})$, and fits to the two-particle correlation function $C(\boldsymbol{q})$ obtained via (i) the CRAB package, (ii) direct numerical integration of $D(\boldsymbol{\rho})$, and (iii) fast Fourier transformation (FFT). On an identical momentum grid, numerical integration and FFT are analytically identical and numerically equivalent ($\text{RMSE} < 10^{-14}$), while FFT reduces the computation time from hours to seconds. We extract the transverse-mass dependence of the L\'evy radii, L\'evy index $\alpha$, and correlation strength $\lambda$ in $0\text{--}10\,\%$ central collisions. All methods reproduce the same physical trends, but CRAB results fluctuate due to event pooling, and correlation-function fits deviate systematically from direct source fits, e.g., with $\sim 7\text{--}9\,\%$ lower $\alpha$. These deviations may reflect residual non-L\'evy features of the source, probed differently in coordinate and momentum space, and the comparison of event-averaged parameters with fits to the ensemble-accumulated source, a mixture of L\'evy distributions with lower effective $\alpha$.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
Bose--Einstein femtoscopy probes the spatiotemporal geometry of the particle-emitting source in heavy-ion collisions, where experimental data favor a L\'evy-stable source description. We compare four approaches for extracting 3D L\'evy source parameters using $\pi^+\pi^+$ pairs from EPOS3-simulated Au+Au collisions at $\sqrt{s_{\text{NN}}} = 200\text{ GeV}$: direct fits to the pair-source distribution $D(\boldsymbol{\rho})$, and fits to the two-particle correlation function $C(\boldsymbol{q})$ obtained via (i) the CRAB package, (ii) direct numerical integration of $D(\boldsymbol{\rho})$, and (iii) fast Fourier transformation (FFT). On an identical momentum grid, numerical integration and FFT are analytically identical and numerically equivalent ($\text{RMSE} < 10^{-14}$), while FFT reduces the computation time from hours to seconds. We extract the transverse-mass dependence of the L\'evy radii, L\'evy index $\alpha$, and correlation strength $\lambda$ in $0\text{--}10\,\%$ central collisions. All methods reproduce the same physical trends, but CRAB results fluctuate due to event pooling, and correlation-function fits deviate systematically from direct source fits, e.g., with $\sim 7\text{--}9\,\%$ lower $\alpha$. These deviations may reflect residual non-L\'evy features of the source, probed differently in coordinate and momentum space, and the comparison of event-averaged parameters with fits to the ensemble-accumulated source, a mixture of L\'evy distributions with lower effective $\alpha$.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Bose--Einstein femtoscopy probes the spatiotemporal geometry of the particle-emitting source in heavy-ion collisions, where experimental data favor a L\'evy-stable source description. | background 1.00 | 0.06 | 0.04 | asserted 0.98 |
| s1 | We compare four approaches for extracting 3D L\'evy source parameters using $\pi^+\pi^+$ pairs from EPOS3-simulated Au+Au collisions at $\sqrt{s_{\text{NN}}} = 200\text{ GeV}$: direct fits to the pair-source distribution $D(\boldsymbol{\rho})$, and fits to the two-particle correlation function $C(\boldsymbol{q})$ obtained via (i) the CRAB package, (ii) direct numerical integration of $D(\boldsymbol{\rho})$, and (iii) fast Fourier transformation (FFT). | method 0.62 | 0.95 | 0.05 | asserted 0.98 |
| s2 | On an identical momentum grid, numerical integration and FFT are analytically identical and numerically equivalent ($\text{RMSE} < 10^{-14}$), while FFT reduces the computation time from hours to seconds. | result 0.98 | 0.91 | 0.19 | asserted 1.00 |
| s3 | We extract the transverse-mass dependence of the L\'evy radii, L\'evy index $\alpha$, and correlation strength $\lambda$ in $0\text{--}10\,\%$ central collisions. | contribution 0.70 | 0.95 | 0.04 | asserted 0.98 |
| s4 | All methods reproduce the same physical trends, but CRAB results fluctuate due to event pooling, and correlation-function fits deviate systematically from direct source fits, e.g., with $\sim 7\text{--}9\,\%$ lower $\alpha$. | result 1.00 | 0.95 | 0.47 | asserted 1.00 |
| s5 | These deviations may reflect residual non-L\'evy features of the source, probed differently in coordinate and momentum space, and the comparison of event-averaged parameters with fits to the ensemble-accumulated source, a mixture of L\'evy distributions with lower effective $\alpha$. | interpretation 1.00 | 0.54 | 0.29 | hedged possibility 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
Figures
Figures reproduced from arXiv:2611.08069 by the authors.
References
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.