Paul Cosgrove
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Paul Cosgrove has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Eugene Shwageraus
University of Cambridge
9 shared publications - Martin Skretteberg
University of Cambridge
5 shared publications - Mikolaj Adam Kowalski
University of Cambridge
4 shared publications - Valeria Raffuzzi
University of Cambridge
3 shared publications E. Shwageraus
2 shared publications- Esmae J. Woods
University of Cambridge
2 shared publications
Latest Publications
- Open Access
- 17 September 2026
- Published by Frontiers Media SA in Frontiers in Nuclear Engineering
- Vol. 5, 1921592
Better physics, less noise: inline multi-physics reduces correlation in Monte Carlo reactor simulations
Recently, several authors have proposed that multi-physics calculations involving coupled neutronics and thermal-hydraulics are most efficiently performed using tight or ‘inline’ coupling. This consists of passing neutronic information to the thermal-hydraulics solver before the neutronics solution is fully converged, providing benefits in stability and runtime. In the context of Monte Carlo neutronics, this work considers the effects of correlation on the results of these calculations by varying both the frequency of data exchanges and the neutron population used each cycle. This is done using newly developed multi-physics capabilities in the Monte Carlo code SCONE. The results show that inline coupling throughout the entire Monte Carlo calculation is not only robust to the choice of neutron population, but also reduces the inter-generational correlation between neutrons, producing lower variance results than if the coupled physics fields were not updated. For the 3D pin problem considered, this implies about 2.89× fewer active particles can be simulated in a coupled calculation to achieve the same power errors as an uncoupled calculation using the correct temperature and density fields.
- 17 September 2026
- Published by Frontiers Media SA in Frontiers in Nuclear Engineering
- Vol. 5, 1921592
Open AccessBetter physics, less noise: inline multi-physics reduces correlation in Monte Carlo reactor simulations
Recently, several authors have proposed that multi-physics calculations involving coupled neutronics and thermal-hydraulics are most efficiently performed using tight or ‘inline’ coupling. This consists of passing neutronic information to the thermal-hydraulics solver before the neutronics solution is fully converged, providing benefits in stability and runtime. In the context of Monte Carlo neutronics, this work considers the effects of correlation on the results of these calculations by varying both the frequency of data exchanges and the neutron population used each cycle. This is done using newly developed multi-physics capabilities in the Monte Carlo code SCONE. The results show that inline coupling throughout the entire Monte Carlo calculation is not only robust to the choice of neutron population, but also reduces the inter-generational correlation between neutrons, producing lower variance results than if the coupled physics fields were not updated. For the 3D pin problem considered, this implies about 2.89× fewer active particles can be simulated in a coupled calculation to achieve the same power errors as an uncoupled calculation using the correct temperature and density fields.
- Open Access
- 11 September 2026
- Published by Elsevier in Journal of Computational Physics
- Vol. 568
Spectral convergence properties and optimal performance of Monte Carlo functional expansion tallies vs. the histogram approximation
- 11 September 2026
- Published by Elsevier in Journal of Computational Physics
- Vol. 568
Open AccessSpectral convergence properties and optimal performance of Monte Carlo functional expansion tallies vs. the histogram approximation
- Open Access
- 27 July 2026
- Published by Elsevier in SSRN Electronic Journal
The Random Ray Method with Space-Time Characteristics for Transient Neutral Particle Transport
This paper presents a novel simulation method for transient neutral particle transport. The proposed method builds on the mathematical method of characteristics to solve the time-dependent transport equation by tracing rays that simultaneously traverse the space and time domains. This new approach offers the ability to fully represent the angular component of particle flux in both space and time in a computationally efficient manner. This capability is an advancement over current state-of-the-art methods, which either abandon the angular time dependence and approximate the respective time derivative using isotropic scalar fluxes, or otherwise retain the full angular time dependence at the cost of being computationally impractical for realistic applications. The proposed method adopts the random ray method, which leverages recent advances in stochastic transport theory that enable minimal-bias truncation strategies of the space-time tracks that are not possible with traditional deterministic methods. The resulting truncated space-time characteristics random ray method is implemented in the SCONE neutron transport code and tested on the C5G7-TD community benchmark as well as a simplified radiation streaming problem. The results demonstrate high accuracy without any significant additional time or memory requirements as compared to lower-fidelity state-of-the-art methods and confirm that the new approach remains reliable for problems with strong directional dependence, for which the isotropic approximation fails.
- 27 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessThe Random Ray Method with Space-Time Characteristics for Transient Neutral Particle Transport
This paper presents a novel simulation method for transient neutral particle transport. The proposed method builds on the mathematical method of characteristics to solve the time-dependent transport equation by tracing rays that simultaneously traverse the space and time domains. This new approach offers the ability to fully represent the angular component of particle flux in both space and time in a computationally efficient manner. This capability is an advancement over current state-of-the-art methods, which either abandon the angular time dependence and approximate the respective time derivative using isotropic scalar fluxes, or otherwise retain the full angular time dependence at the cost of being computationally impractical for realistic applications. The proposed method adopts the random ray method, which leverages recent advances in stochastic transport theory that enable minimal-bias truncation strategies of the space-time tracks that are not possible with traditional deterministic methods. The resulting truncated space-time characteristics random ray method is implemented in the SCONE neutron transport code and tested on the C5G7-TD community benchmark as well as a simplified radiation streaming problem. The results demonstrate high accuracy without any significant additional time or memory requirements as compared to lower-fidelity state-of-the-art methods and confirm that the new approach remains reliable for problems with strong directional dependence, for which the isotropic approximation fails.
- Open Access
- 27 July 2026
- Published by Elsevier in SSRN Electronic Journal
Stochastic Fourier analysis of the random ray method
Different iterative solution algorithms for the neutron transport equation have distinct stability and convergence characteristics. Understanding these properties is important in order to robustly apply these algorithms to problems of engineering interest. In this work, Fourier analysis is applied to The Random Ray Method (TRRM) of neutron transport to derive its stability and asymptotic convergence rate in terms of system parameters. Importantly, the stochastic nature of TRRM is not neglected, and the impact of noise on the stability is accounted for; to our knowledge, this is the first time Fourier analysis has been successfully applied to stochastic radiation transport algorithms. The average TRRM spectral radius is identical to that of power iteration, but with a small correction term that is a function of system parameters, such as the ray length. TRRM is found to be unconditionally stable on average, although noise can produce transient growth in error modes. The analysis is also used to explain previously observed adverse behaviour of another TRRM implementation; this is deterministically equivalent to the common implementation but requires dramatically higher ray numbers and lengths to produce acceptable results. This highlights the additional predictive power of accounting for stochastic behaviour. The predictions are compared with measured results obtained using the neutron transport code SCONE and excellent agreement for both the mean and variance of the spectral radius is observed.
- 27 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessStochastic Fourier analysis of the random ray method
Different iterative solution algorithms for the neutron transport equation have distinct stability and convergence characteristics. Understanding these properties is important in order to robustly apply these algorithms to problems of engineering interest. In this work, Fourier analysis is applied to The Random Ray Method (TRRM) of neutron transport to derive its stability and asymptotic convergence rate in terms of system parameters. Importantly, the stochastic nature of TRRM is not neglected, and the impact of noise on the stability is accounted for; to our knowledge, this is the first time Fourier analysis has been successfully applied to stochastic radiation transport algorithms. The average TRRM spectral radius is identical to that of power iteration, but with a small correction term that is a function of system parameters, such as the ray length. TRRM is found to be unconditionally stable on average, although noise can produce transient growth in error modes. The analysis is also used to explain previously observed adverse behaviour of another TRRM implementation; this is deterministically equivalent to the common implementation but requires dramatically higher ray numbers and lengths to produce acceptable results. This highlights the additional predictive power of accounting for stochastic behaviour. The predictions are compared with measured results obtained using the neutron transport code SCONE and excellent agreement for both the mean and variance of the spectral radius is observed.
- Open Access
- 22 June 2026
- Published by Taylor & Francis in Nuclear Technology
- p. 1-34
Functional Expansion Tallies for Kinetic Monte Carlo Variance Reduction
- 22 June 2026
- Published by Taylor & Francis in Nuclear Technology
- p. 1-34
Open AccessFunctional Expansion Tallies for Kinetic Monte Carlo Variance Reduction
- Open Access
- 9 February 2026
- Published by Elsevier in SSRN Electronic Journal
Spectral Convergence Properties and Optimal Performance of Monte Carlo Functional Expansion Tallies vs. the Histogram Approximation
We investigate asymptotic convergence properties of two methods for tallying responses in Monte Carlo particle transport. The first method is the histogram approximation, in which particle scores generated during a Monte Carlo simulation are projected onto a histogram representation of the domain. The second method employs functional expansion tallies, where scores are projected onto orthogonal basis functions, enabling a continuous representation of the response across the domain. For functional expansion tallies, previous work analysed the behaviour of expansion coefficients using an integration by parts framework, establishing upper bounds on the decay rates. We recast the spectral analysis in a more general Sobolev framework, obtaining both upper and lower bounds on coefficient decay under necessary and sufficient regularity conditions, applicable consistently across the bases considered. From these bounds, we derive bands for asymptotic convergence rates of truncation errors in the $L^2$ and $L^\infty$ norms. Moreover, we introduce a new optimality analysis that determines optimal configurations for both methods, namely the number of histogram bins or expansion orders that minimise total estimation error. The analysis yields explicit expressions for these optimal parameters, providing a practical and general guideline for configuring stochastic spectral estimators. We assess several orthogonal bases with respect to theoretical convergence and practical viability, and validate the theoretical framework on simple and representative numerical problems. The results are consistent with theoretical predictions and demonstrate that, depending on the smoothness of the target distribution, functional expansion tallies can yield more efficient estimates of the target response than the histogram approximation.- 9 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSpectral Convergence Properties and Optimal Performance of Monte Carlo Functional Expansion Tallies vs. the Histogram Approximation
We investigate asymptotic convergence properties of two methods for tallying responses in Monte Carlo particle transport. The first method is the histogram approximation, in which particle scores generated during a Monte Carlo simulation are projected onto a histogram representation of the domain. The second method employs functional expansion tallies, where scores are projected onto orthogonal basis functions, enabling a continuous representation of the response across the domain. For functional expansion tallies, previous work analysed the behaviour of expansion coefficients using an integration by parts framework, establishing upper bounds on the decay rates. We recast the spectral analysis in a more general Sobolev framework, obtaining both upper and lower bounds on coefficient decay under necessary and sufficient regularity conditions, applicable consistently across the bases considered. From these bounds, we derive bands for asymptotic convergence rates of truncation errors in the $L^2$ and $L^\infty$ norms. Moreover, we introduce a new optimality analysis that determines optimal configurations for both methods, namely the number of histogram bins or expansion orders that minimise total estimation error. The analysis yields explicit expressions for these optimal parameters, providing a practical and general guideline for configuring stochastic spectral estimators. We assess several orthogonal bases with respect to theoretical convergence and practical viability, and validate the theoretical framework on simple and representative numerical problems. The results are consistent with theoretical predictions and demonstrate that, depending on the smoothness of the target distribution, functional expansion tallies can yield more efficient estimates of the target response than the histogram approximation. - Open Access
- 16 December 2025
- Published by Taylor & Francis in Nuclear Science and Engineering
- p. 1-22
Fourier Analysis of Fission Matrix Convergence Acceleration
- 16 December 2025
- Published by Taylor & Francis in Nuclear Science and Engineering
- p. 1-22
Open AccessFourier Analysis of Fission Matrix Convergence Acceleration
- Open Access
- 27 November 2025
- Published by Elsevier in Annals of Nuclear Energy
- Vol. 227
Status of the SCONE Monte Carlo neutron transport code
- 27 November 2025
- Published by Elsevier in Annals of Nuclear Energy
- Vol. 227
Open AccessStatus of the SCONE Monte Carlo neutron transport code
- Open Access
- 26 November 2025
- Published by Elsevier in Progress in Nuclear Energy
- Vol. 192
Negative fluxes and cell-miss errors in the random ray method
- 26 November 2025
- Published by Elsevier in Progress in Nuclear Energy
- Vol. 192
Open AccessNegative fluxes and cell-miss errors in the random ray method
- Open Access
- 20 November 2025
- Published by Taylor & Francis in Nuclear Science and Engineering
- p. 1-26
Selection-Driven Evolutionary Population Control for Kinetic Monte Carlo Variance Reduction
- 20 November 2025
- Published by Taylor & Francis in Nuclear Science and Engineering
- p. 1-26
Open AccessSelection-Driven Evolutionary Population Control for Kinetic Monte Carlo Variance Reduction