Methods

{
  "name": "ParallelBVHNarrowPhase",
  "namespace": "multithreading::component::collision::detection::algorithm",
  "module": "MultiThreading",
  "include": "MultiThreading/component/collision/detection/algorithm/ParallelBVHNarrowPhase.h",
  "doc": "Parallel version of the narrow phase collision detection based on boundary volume hierarchy.",
  "inherits": [
    "BVHNarrowPhase",
    "TaskSchedulerUser"
  ],
  "templates": [],
  "data_fields": [],
  "links": [],
  "methods": [
    {
      "name": "init",
      "return_type": "void",
      "params": [],
      "is_virtual": true,
      "is_pure_virtual": false,
      "is_static": false,
      "access": "public"
    },
    {
      "name": "addCollisionPairs",
      "return_type": "void",
      "params": [
        {
          "name": "v",
          "type": "const int &"
        }
      ],
      "is_virtual": false,
      "is_pure_virtual": false,
      "is_static": false,
      "access": "public"
    }
  ],
  "description": "The `ParallelBVHNarrowPhase` component is part of the SOFA framework's MultiThreading module and extends the functionality of narrow phase collision detection using Boundary Volume Hierarchies (BVH). It leverages parallel processing to enhance performance by distributing the workload across multiple threads.\n\n**Role and Purpose:**\nThis component is designed for efficient narrow-phase collision detection in scenarios where multiple pairs of models need to be tested for collisions simultaneously. By utilizing a task scheduler, it ensures that the collision detection tasks are distributed effectively among available CPU cores, leading to faster computation times.\n\n**Interactions with Other Components:*\nThe `ParallelBVHNarrowPhase` inherits from `BVHNarrowPhase`, indicating that it builds upon traditional BVH-based narrow phase algorithms. It also implements `TaskSchedulerUser`, allowing it to interface with SOFA's task scheduling system for parallel execution.\n\n*Practical Usage and Data Fields:*\n- **Initialization:** The component initializes a thread pool during the `init` method, preparing it for concurrent tasks.\n- **Collision Pair Handling:** It processes collision pairs using the `addCollisionPairs` method. This function prepares output structures and creates CPU tasks to manage collision detection for each pair in parallel.\n- **Task Management:** Tasks are created as instances of `ParallelBVHNarrowPhasePairTask`, which execute the actual collision detection logic.",
  "maths": "The `ParallelBVHNarrowPhase` component in the SOFA framework is designed to enhance performance in narrow phase collision detection using Boundary Volume Hierarchies (BVH) through parallel processing. This component does not directly contribute to governing equations or operators such as mass matrix $M$, stiffness matrix $K$, internal force $f_{int}$, residual $R$, etc., which are central to FEM simulations. Instead, it focuses on the efficient management of collision detection tasks in a multithreaded environment.\n\n### Role and Purpose:\n\nThe `ParallelBVHNarrowPhase` component is specifically aimed at improving the performance of narrow phase collision detection by distributing tasks across multiple CPU cores. This approach leads to faster computation times for scenarios involving numerous pairs of models requiring simultaneous collision testing.\n\n### Interactions with Other Components:\n\n1. **Inheritance from BVHNarrowPhase:** The `ParallelBVHNarrowPhase` inherits from the `BVHNarrowPhase`, meaning it builds upon traditional BVH-based narrow phase algorithms while adding parallel processing capabilities.\n2. **TaskSchedulerUser Interface:** By implementing the `TaskSchedulerUser` interface, this component interfaces with SOFA's task scheduling system to manage and distribute tasks across available CPU cores effectively.\n\n### Key Methods and Operations:\n\n1. **Initialization (`init` method):** The initialization of a thread pool is performed during the `init` method, setting up the component for concurrent execution of collision detection tasks.\n2. **Collision Pair Handling (`addCollisionPairs` method):** This function processes pairs of models for collision detection by preparing necessary output structures and creating CPU tasks to manage each pair's collision detection in parallel.\n3. **Task Management:** Each task is encapsulated within an instance of the `ParallelBVHNarrowPhasePairTask`, which executes the collision detection logic.\n\n### Numerical Methods or Discretization Choices:\n\nThe `ParallelBVHNarrowPhase` does not directly encode any specific numerical methods or discretization choices related to FEM. Instead, it focuses on parallelizing the narrow phase collision detection process, thereby improving computational efficiency without altering the underlying physics or mathematical formulation of the simulation.\n\n### Integration into Variational / Lagrangian Mechanics Framework:\n\nWhile `ParallelBVHNarrowPhase` does not contribute directly to variational or Lagrangian mechanics formulations, it plays a critical role in ensuring that collision detection tasks are handled efficiently within the broader context of FEM simulations. This efficient handling ensures that the physical consistency and numerical stability required by these formulations are maintained during real-time or high-fidelity simulations.\n\nIn summary, `ParallelBVHNarrowPhase` is a component that enhances performance through parallel processing, ensuring rapid and accurate collision detection in complex simulations without altering the core physics-based simulation pipeline.",
  "abstract": "The `ParallelBVHNarrowPhase` enhances narrow-phase collision detection using parallel processing based on Boundary Volume Hierarchies (BVH). It leverages SOFA's task scheduling system to distribute tasks across multiple CPU cores for improved performance.",
  "sheet": "# ParallelBVHNarrowPhase\n\n## Overview\nThe `ParallelBVHNarrowPhase` component is part of the MultiThreading module in the SOFA framework. It extends traditional BVH-based narrow phase collision detection by utilizing parallel processing to enhance computational efficiency. This component inherits from `BVHNarrowPhase`, indicating it builds upon established BVH algorithms, and implements `TaskSchedulerUser` for effective task distribution across multiple CPU cores.\n\n## Parameters and Data\nThe `ParallelBVHNarrowPhase` does not expose any significant data fields or parameters that control its behavior beyond those inherited from parent classes.\n\n## Dependencies and Connections\nThis component inherits from `BVHNarrowPhase`, indicating it builds upon traditional BVH-based narrow phase algorithms. It also implements the `TaskSchedulerUser` interface to manage tasks within SOFA's task scheduling system, ensuring efficient parallel execution of collision detection tasks.\n\n## Practical Notes\nThe `ParallelBVHNarrowPhase` component initializes a thread pool during its initialization (`init`) method and processes collision pairs using the `addCollisionPairs` method. Each collision pair is handled in parallel through CPU tasks created as instances of `ParallelBVHNarrowPhasePairTask`. This approach ensures rapid and accurate collision detection, but users should be aware that performance gains are most pronounced in scenarios with a high number of concurrent collision tests."
}