Gyroscopic alignment of a fluid occurs when flow structures align with the rotation axis. This often gives rise to highly spatially anisotropic columnar structures that in combination with complex domain boundaries pose challenges for efficient numerical discretizations and computations. We define gyroscopic polynomials to be three-dimensional polynomials expressed in a coordinate system that conforms to rotational alignment. We remap the original domain with radius-dependent boundaries onto a right cylindrical or annular domain to create the computational domain in this coordinate system. We find the volume element expressed in gyroscopic coordinates leads naturally to a hierarchy of orthonormal bases. We build the bases out of Jacobi polynomials in the vertical and generalized Jacobi polynomials in the radial. Because these coordinates explicitly conform to flow structures found in rapidly rotating systems the bases represent fields with a relatively small number of modes. We develop the operator structure for one-dimensional semi-classical orthogonal polynomials as a building block for differential operators in the full three-dimensional cylindrical and annular domains. The differential operators of generalized Jacobi polynomials generate a sparse linear system for discretization of differential operators acting on the gyroscopic bases. This enables efficient simulation of systems with strong gyroscopic alignment.
翻译:流体的陀螺对准是指流动结构与旋转轴对其的现象。这往往会产生高度空间各向异性的柱状结构,与复杂的空间边界相结合,给高效数值离散化和计算带来了挑战。我们定义陀螺多项式为在与旋转对称一致的坐标系中表示的三维多项式。我们将具有半径依赖边界的原始区域重新映射到圆柱形或环形域,以在此坐标系中创建计算域。我们发现,陀螺坐标表达的体积元素自然地导致一系列正交基。我们使用垂直方向的Jacobi多项式和径向方向的广义Jacobi多项式构建基。因为这些坐标明确符合快速旋转系统中发现的流动结构,所以这些基代表着具有相对较少模式的场。我们以一维半经典正交多项式的运算符结构为基础,构建完整三维圆柱形和环形域内的微分运算符。广义Jacobi多项式的微分运算符生成了一个用于向作用于陀螺基的微分运算符离散化的稀疏线性系统。这使得能够有效模拟具有强陀螺对准的系统。