GitHub

Changelog

DispersiveShallowWater.jl follows the interpretation of semantic versioning (semver) used in the Julia ecosystem. Notable changes will be documented in this file for human readability.

Changes in the v0.10 lifecycle

Added

  • The HyperbolicSainteMarieEquations1D were added for different types of bathymetry (#288).
  • Add linear dispersion relation for the HyperbolicSerreGreenNaghdiEquations1D (#321).
  • still_water_surface and bathymetry can now be used as conversion functions, e.g., for plotting with plot(semi => sol, conversion = bathymetry) (#338).

Fixed

  • If the water height becomes negative, e.g., because the time step is too large, the SerreGreenNaghdiEquations1D with a FourierDerivativeOperator or a VarCoefDerivativeOperator now return NaNs instead of throwing a PosDefException. This allows adaptive time integration methods to reject the step, as for the other solvers (#334).
  • The AnalysisCallback now ignores unsupported analysis errors (with a warning) instead of reporting them as zero in errors. The errors returned by errors are now stored according to their names, which fixes wrong labels or a BoundsError if analysis_errors is passed in a different order or without the L2 and L∞ errors (#329).
  • Invalid arguments to Mesh1D, Solver, Semidiscretization, convergence_test, and the plot recipe of the AnalysisCallback now throw an ArgumentError with an explanation instead of an AssertionError. A Semidiscretization of the HyperbolicSainteMarieEquations1D with reflecting boundary conditions now throws an ArgumentError instead of a MethodError (#330).
  • Mesh1D accepts any integer type for the number of nodes (#330).
  • convergence_test no longer sorts the passed vector of resolutions in place and writes all its output to io (#330).
  • Plotting a solution over time at a fixed x (e.g., plot(semi => sol, x)) now shows all variables. Before, variables after the bathymetry were dropped, e.g., H for the HyperbolicSerreGreenNaghdiEquations1D with conversion = prim2prim (#327).
  • initial_condition_soliton for the HyperbolicSerreGreenNaghdiEquations1D now respects a nonzero still-water surface eta0. Before, the water height was off by eta0 (#326).
  • The keyword constructors of all equations now promote their parameters to a common type, so that, e.g., SerreGreenNaghdiEquations1D(gravity = 1) or HyperbolicSerreGreenNaghdiEquations1D(gravity = 9.81, lambda = 500) no longer throw a MethodError (#325).
  • source_terms_manufactured_reflecting for the SerreGreenNaghdiEquations1D with bathymetry_mild_slope and bathymetry_variable now uses the gravitational acceleration of the equations instead of a hard-coded value of 9.81 (#324).

Changes when updating to v0.10 from v0.9.x

Changed

  • There are no default boundary conditions anymore in Semidiscretization. Users now have to explicitly provide boundary conditions via the boundary_conditions keyword argument, which were periodic by default before (#286).
  • convergence_test now returns the complete convergence orders. To obtain the mean convergence rates, use DispersiveShallowWater.calc_mean_convergence on the convergence orders (#285).

Changes when updating to v0.9 from v0.8.x

Changed

  • The argument plot_initial in the plotting functions now always plots the initial condition at time t = first(tspan). To plot the analytical solution at the current time step, use plot_analytical instead (#277).
  • Changed the default_example() from a BBM-BBM setup to a physically more relevant Serre-Green-Naghdi setup (#276).

Changes in the v0.8 lifecycle

Added

  • Add function data_dingemans to load experimental data for the Dingemans experiment and expand the documentation about the experiment (#238).

Changes when updating to v0.8 from v0.7.x

Changed

  • The initial condition initial_condition_dingemans uses equation-specific offsets to account for phase shifts compared to the experimental data and a slightly different range for the initial wave train (#237).

Changes in the v0.7 lifecycle

Added

  • Improved documentation (#233).
  • Add linear dispersion relation for the KdVEquation1D (#230).
  • Support reflecting boundary conditions for HyperbolicSerreGreenNaghdiEquations1D (#228).
  • Add conversion functions prim2nondim and nondim2prim for non-dimensional variables for KdVEquation1D (#217).
  • Support reflecting boundary conditions for SerreGreenNaghdiEquations1D (flat bathymetry: #205, general: #223).
  • Add KdVEquation1D (#198).

Changes when updating to v0.7 from v0.6.x

Changed

  • The default parameters of the SvaerdKalischEquations1D changed from alpha = 0.0, beta = 0.2308939393939394, gamma = 0.04034343434343434 to alpha = 0.0, beta = 1/3, gamma = 0.0 (#196).

Changes in the v0.6 lifecycle

Added

  • Support source terms for SerreGreenNaghdiEquations1D (#180, #186).
  • Add initial support for ForwardDiff.jl for HyperbolicSerreGreenNaghdiEquations1D and DispersiveShallowWater.jacobian (#185).

Changes when updating to v0.6 from v0.5.x

Changed

  • The keyword argument and function gravity_constant have been changed to gravity (#174).

Changes in the v0.5 lifecycle

Added

  • Add LinearDispersionRelation and documentation about dispersion (#168).
  • Reflecting boundary conditions are added for the Svärd-Kalisch equations with alpha = gamma = 0 (#166).
  • Fix a bug in the upwind discretization of the SvaerdKalischEquations1D.
  • Use OrdinaryDiffEqTsit5.jl and OrdinaryDiffEqLowStorageRK.jl instead of OrdinaryDiffEq.jl in all examples to reduce latency (#163).
  • Allow Fourier and periodic rational derivative operators for BBMBBMEquations1D and SvaerdKalischEquations1D (#154).
  • Add BBMEquation1D (#150).

Changes when updating to v0.5 from v0.4.x

Changed

  • The BBMBBMVariableEquations1D were removed and BBMBBMEquations1D now supports a bathymetry_type to choose between a flat and a variable bathymetry (#147).
  • The default of bathymetry_type for the SerreGreenNaghdiEquations1D changed from bathymetry_flat to bathymetry_variable (#147).
  • bathymetry_type is now a keyword argument for all equations instead of a positional argument (#147).
  • The initial_condition_dingemans for the SerreGreenNaghdiEquations1D and HyperbolicSerreGreenNaghdiEquations1D was changed a bit to be more consistent with the other equations (#147).

Changes in the v0.4 lifecycle

Added

  • The SerreGreenNaghdiEquations1D were added for different types of bathymetry (#127, #135).
  • The HyperbolicSerreGreenNaghdiEquations1D were added for different types of bathymetry (#139).
  • The abstract interface AbstractShallowWaterEquations was added to unify several systems such as the SerreGreenNaghdiEquations1D, the BBMBBMEquations1D, and the SvaerdKalischEquations1D (#127).
  • A new conversion function prim2phys was introduced, defaulting to prim2prim. prim2phys is the default conversion function for plotting.

Changes when updating to v0.4 from v0.3.x

Changed

  • Use ArrayPartition from RecursiveArrayTools.jl to store the solution of the ODEProblem (#118).

Changes in the v0.3 lifecycle

Added

  • Add possibility to pass vector of Ns to convergence_test (#113).
  • Performance improvements by using factorized matrices for linear systems solves (#108, #112, #114).
  • Reflecting boundary conditions are added for the BBM-BBM equations (#104, #109).
  • Fix for the BBMBBMVariableEquations1D, where the still water surface was neglected leading to a bug in the Dingemans setup (#91).

Changes when updating to v0.3 from v0.2.x

Changed

  • Add keyword argument start_from when plotting AnalysisCallback (#87).
  • Manufactured solution for Svärd-Kalisch equations uses a variable bathymetry (#84).

Changes in the v0.2 lifecycle

Added

  • Add SummaryCallback (#75).

Changes when updating to v0.2 from v0.1.x

Changed

  • The code from the master thesis of Joshua Lampert was separated (#69).
  • Add support for source terms (#65).
  • A higher order interpolation is used when plotting the solution at a value x outside the grid (#64).