Modules

Emit a solver-ready Mesh2D_t from an adaptively refined, 2:1-balanced quad-forest (AMR Stage 4b). This closes the adaptive-refinement loop: with the Stage-1 indicator flagging elements, Stage 2b mutating the forest, and Stage 4a balancing it, EmitMesh produces the nonconforming mesh - leaf geometry, conforming-side connectivity, and a mortar table for the 2:1 hanging faces - that the existing mortar solver machinery already handles.

Adaptive-mesh-refinement controller for 3-D DG models, the direct analogue of SELF_AMRController_2D: the driver that closes the AMR loop around a live, time-stepping model. One call to Adapt performs one adaptation epoch between time steps:

Device backend of the column operators (see SELF_ColumnOperators_3D_t). Every operator reads and writes the device copies (interior_gpu) of its fields and the device copy of the covariant basis, and moves no data between host and device. The caller uploads the inputs (UpdateDevice) and downloads the outputs (UpdateHost) when it needs them on the host.

Column operators on extruded meshes (Mesh3D%ExtrudedMesh).

Entropy-conserving linear scalar advection equation in 2-D.

Entropy-conserving DGSEM base class for 2-D conservation laws.

Partition a 3-D mesh into element sets and extract one sub-mesh per set.

Entropy-conserving compressible Euler equations in 3-D with potential temperature formulation (FastEddy / atmospheric LES equation set).

Linear Euler 2D with a Perfectly Matched Layer (PML) absorbing region.

This module defines a class that can be used to solve the Linear Euler equations in 3-D. The Linear Euler Equations, here, are the Euler equations linearized about a motionless background state. The sound speed, the background density, and the relaxation rate are carried as solution variables (static in time, possibly spatially varying) so that heterogeneous media and sponge/damping layers are supported, mirroring the 2-D model.

Geometry-aware two-point vector and split-form divergence for 3-D.

Mesh-level h-refinement operations for 3-D hexahedral meshes, the direct analogue of SELF_MeshRefinement_2D.

GPU backend of MultiSetModel3D. The interface exchange gathers each set's face trace from solution%boundary_gpu, and the default FillInterfaceBoundary writes solution%extBoundary_gpu with a device kernel. Overrides of FillInterfaceBoundary must likewise write the device copy (iface%ghost_gpu holds the ghost traces there), and must not call solution%UpdateDevice inside the step loop: it resets the halo exchange's static-variable state.

A parent model that runs one DG model per element set and couples the sets at their shared faces.

Forest-of-octrees data structure for adaptive h-refinement of 3-D hexahedral meshes, the direct analogue of SELF_QuadTreeMesh_2D. Each base-mesh element is the root of an octree; refinement replaces a leaf with eight children (one per reference octant, SELF child ordering = CGNS corner order), and coarsening merges a family of eight leaf siblings back into their parent. The forest tracks refinement levels and the active leaf set; the physical geometry of any leaf is produced by repeated exact isoparametric subdivision of its root (SELF_RefinementPrimitives_3D).

Modal (Legendre spectral-decay) refinement indicator for 2-D spectral element solutions.

Element-local primitives for h-refinement of 2-D quadrilateral spectral element meshes (AMR Stage 2). These routines are deliberately free of the Mesh2D container, MPI, and GPU dependencies so that the numerically and topologically delicate pieces - isoparametric child geometry and refined-mesh connectivity - are pure, deterministic, and unit-testable in isolation. The mesh-level driver (SELF_MeshRefinement_2D) assembles a Mesh2D_t from them.

GPU backend of the element-set interface exchange. When every set with interface faces supplies a device trace (SetTraceRef%trace_gpu), the face traces are gathered on the device and copied to the host in one transfer; the host exchange then runs as on the CPU, and the ghost traces are copied back to the device for FillSetInterfaceBoundary_gpu. Without device traces the host path runs unchanged.

Interface map and trace exchange between element sets.

Solution transfer between an element and its four h-refinement children (AMR Stage 3).

\file SELF_SupportRoutines.f90 Contains the \ref SELF_SupportRoutines module \defgroup SELF_SupportRoutines SELF_SupportRoutines This module defines a set of general purpose routines.

Old-leaf -> new-leaf solution transfer plan for one adaptation epoch of the 3-D oct-forest: the driver layer of AMR Stage 3 that connects the element-local transfer operators (SELF_SolutionTransfer_3D) to the forest mutation (SELF_OctreeMesh_3D).