[Paper Review] Lampray: Multi-group long characteristics ray tracing for adaptive mesh radiation hydrodynamics

Troels Frostholm, Troels Haugbølle+1 more|arXiv (Cornell University)|Sep 14, 2018

Computational Fluid Dynamics and Aerodynamics1 references4 citations

TL;DR

Lampray introduces a multi-group long characteristics ray tracing method for adaptive mesh radiation hydrodynamics in the RAMSES code, enabling accurate treatment of radiative transfer with shadowing, colliding beams, and non-equilibrium chemistry. It integrates H-He-C-O photochemistry with 36 species and 240 reactions, self-shielding for H₂ and CO, and dust temperature/ice formation, achieving efficient parallelization via ray-domain decomposition and hash-table lookups.

ABSTRACT

We present Lampray: a multi-group long characteristics ray tracing method for adaptive mesh radiation hydrodynamics in the Ramses code. It avoids diffusion, captures shadows, and treats colliding beams correctly, and therefore complements existing moment-based ray tracing in Ramses. Lampray includes different options for interpolation between ray and cell domain, and use either integral, Fourier, or an implicit method for hydrogen ionization to solve the radiative transfer. The opacity can either be tabulated or computed through a coupling to the general non-equilibrium astro-chemistry framework Krome. We use an H-He-C-O network with 36 species and 240 reactions to track the photo-chemistry in the interstellar medium across 6 and 10 orders of magnitude in temperature and density. Self-shielding prescriptions for H$_2$ and CO are used together with a new model for the diffuse interstellar UV-field. We also track the dust temperature, formation of H$_2$ on grains, and H$_2$O and CO ices in detail. Lampray is tested against standard benchmarks for molecular cloud and star formation simulations, including the formation of a Strömgren sphere, the expansion of an ionization front, the photo-evaporation of a dense clump, and the H-He-C-O chemistry in a static photo-dissociation front. Efficient parallelisation is achieved with a separate domain decomposition for rays where points along a ray reside in the same memory space, and data movement from cell- to ray-domain is done with a direct hash-table lookup algorithm. Point sources are treated without splitting rays, and therefore the method currently only scales to a few point sources, while diffuse radiation has excellent scaling.

Motivation & Objective

• To develop a high-fidelity radiative transfer method that captures shadows, colliding beams, and non-diffusive transport in adaptive mesh refinement (AMR) simulations.
• To couple detailed non-equilibrium astrochemistry (H-He-C-O network with 36 species and 240 reactions) to radiation transfer for accurate modeling of interstellar medium conditions.
• To include self-shielding for H₂ and CO, diffuse interstellar UV field modeling, dust temperature evolution, and ice formation on grains.
• To enable efficient parallelization of ray tracing via domain decomposition in ray space and direct hash-table lookups for cell-ray interpolation.
• To provide a scalable solution for radiation-hydrodynamics simulations with point sources and diffuse radiation fields, particularly for star formation and protostellar environments.

Proposed method

• Uses multi-group long characteristics ray tracing to compute radiation transport along discrete rays, avoiding flux-limited diffusion and preserving angular resolution.
• Employs integral, Fourier, or implicit c2ray methods for hydrogen ionization, with options for interpolation between ray and cell domains.
• Integrates with the KROME astrochemistry framework to solve time-dependent chemical networks across 6–10 orders of magnitude in density and temperature.
• Incorporates self-shielding for H₂ and CO using a modified Lyα-based prescription, and models the diffuse interstellar UV field with updated parameters.
• Tracks dust temperature evolution and includes surface chemistry for H₂ formation on grains and H₂O/CO ice formation/destruction via freeze-out and desorption processes.
• Parallelizes ray tracing via separate domain decomposition for rays, with data movement from cell to ray domain using a direct hash-table lookup algorithm.

Experimental results

Research questions

• RQ1 Can long characteristics ray tracing in adaptive mesh refinement accurately capture shadowing and colliding beams in radiation-hydrodynamics simulations?
• RQ2 How does non-equilibrium H-He-C-O chemistry, including self-shielding and dust processes, affect the structure and evolution of protostellar and molecular cloud environments?
• RQ3 What is the performance scaling of ray tracing with point sources versus diffuse radiation fields in a parallel AMR framework?
• RQ4 How do self-shielding and updated UV field models impact the accuracy of ionization and molecular chemistry in the interstellar medium?
• RQ5 To what extent does the inclusion of dust temperature and ice formation improve the physical realism of radiation-hydrodynamics simulations?

Key findings

• Lampray successfully captures shadowing and colliding beams in radiation-hydrodynamics, avoiding the unphysical smearing of flux-limited diffusion.
• The method accurately reproduces benchmark tests including Strömgren sphere formation, ionization front expansion, and photo-evaporation of dense clumps.
• The H-He-C-O network with 36 species and 240 reactions enables self-consistent modeling of non-equilibrium chemistry across 6–10 orders of magnitude in density and temperature.
• Self-shielding for H₂ and CO is effectively modeled, improving accuracy in low-optical-depth regions and reducing over-ionization compared to standard methods.
• Dust temperature evolution and ice formation/destruction processes are consistently tracked, with freeze-out and desorption rates computed via temperature- and density-dependent rates.
• The ray-domain decomposition with hash-table lookups enables efficient parallelization, with excellent scaling for diffuse radiation, though point source handling limits scaling to a few sources.