underworld.systems module

This module contains routines relating to differential system.

Module Summary

functions:

underworld.systems.Solver This method simply returns a necessary solver for the provided system.

classes:

underworld.systems.AdvectionDiffusion This class provides functionality for a discrete representation of an advection-diffusion equation.
underworld.systems.HeatSolver Steady State Heat Equation Solver.
underworld.systems.SteadyStateDarcyFlow This class provides functionality for a discrete representation of the steady state darcy flow equation.
underworld.systems.SteadyStateHeat This class provides functionality for a discrete representation of the steady state heat equation.
underworld.systems.Stokes This class provides functionality for a discrete representation of the Stokes flow equations.
underworld.systems.StokesSolver The Block Stokes Schur Complement Solver: This solves the saddle-point system
underworld.systems.SwarmAdvector Objects of this class advect a swarm through time using the provided velocity field.
underworld.systems.TimeIntegration Abstract class for integrating numerical objects (fields, swarms, etc.) in time.

Module Details

functions:

underworld.systems.Solver(eqs, type='BSSCR', *args, **kwargs)[source]

This method simply returns a necessary solver for the provided system.

classes:

class underworld.systems.AdvectionDiffusion(phiField=None, velocityField=None, fn_diffusivity=None, fn_sourceTerm=None, method='SUPG', conditions=[], phiDotField=None, allow_non_q1=False, **kwargs)[source]

Bases: object

This class provides functionality for a discrete representation of an advection-diffusion equation.

\[\frac{\partial\phi}{\partial t} + {\bf u } \cdot \nabla \phi= \nabla { ( k \nabla \phi ) } + H\]

Two methods are available to integrate the scalar \(\phi\) through time: 1) SUPG - The Streamline Upwind Petrov Galerkin method.

paper ref
  1. SLCN - The Semi-Lagrangian Crank-Nicholson method. Implements the Spiegelman & Katz. Semi-lagrangian Advection / Crank Nicholson Diffusion algorithm

Currently the integration method can only be defined at class instanciation. SLADE is the preferred method for Q1, regular cartesian meshes. It is quicker, less diffusive and less restrictive on the timestep size. SUPG is the legacy method for and used for deformed meshes.

Parameters:
  • phiField (underworld.mesh.MeshVariable) – The concentration field, typically the temperature field
  • velocityField (underworld.mesh.MeshVariable) – The velocity field.
  • fn_diffusivity (underworld.function.Function) – A function that defines the diffusivity within the domain.
  • fn_sourceTerm (underworld.function.Function) – A function that defines the heating within the domain. Optional.
  • conditions (underworld.conditions.SystemCondition) – Numerical conditions to impose on the system. This should be supplied as the condition itself, or a list object containing the conditions.
  • [SUPG only] (phiDotField) – A MeshVariable that defines the initial time derivative of the phiField. Typically 0 at the beginning of a model, e.g. phiDotField.data[:]=0 When using a phiField loaded from disk one should also load the phiDotField to ensure the solving method has the time derivative information for a smooth restart. No dirichlet conditions are required for this field as the phiField degrees of freedom map exactly to this field’s dirichlet conditions, the value of which ought to be 0 for constant values of phi.
  • allow_non_q1 (Bool (default False)) – Allow the integration to perform over a non Q1 element mesh. (Under Q2 elements instabilities have been observed as the implementation is only for Q1 elements)

Notes

Constructor must be called by collectively all processes.

get_max_dt()[source]

Returns a timestep size for the current system.

Returns:The timestep size.
Return type:float
integrate(dt=0.0, **kwargs)[source]

Integrates the advection diffusion system through time, dt Must be called collectively by all processes.

Parameters:dt (float) – The timestep interval to use
class underworld.systems.HeatSolver(heatSLE, **kwargs)[source]

Bases: underworld._stgermain.StgCompoundComponent

Steady State Heat Equation Solver.

configure(solve_type='')[source]

Configure velocity/inner solver (A11 PETSc prefix).

solve_type can be one of:

  • mumps : MUMPS parallel direct solver.
  • superludist : SuperLU parallel direct solver.
  • superlu : SuperLU direct solver (serial only).
  • lu : LU direct solver (serial only).
solve(nonLinearIterate=None, nonLinearTolerance=0.01, nonLinearMaxIterations=500, callback_post_solve=None, **kwargs)[source]

Solve the HeatEq system

Parameters:
  • nonLinearIterate (bool) – True will perform non linear iterations iterations, False (or 0) will not
  • nonLinearTolerance (float, Default=1.0e-2) – Relative tolerance criterion for the change in the velocity field
  • nonLinearMaxIterations (int, Default=500) – Maximum number of non linear iteration to perform
  • callback_post_sovle (func, Default=None) – Optional callback function to be performed at the end of a linear solve iteration. Commonly this will be used to perform operations between non linear iterations, for example, calibrating the solution or removing the system null space.
class underworld.systems.SteadyStateDarcyFlow(pressureField, fn_diffusivity, fn_bodyforce=None, voronoi_swarm=None, conditions=[], velocityField=None, swarmVarVelocity=None, _removeBCs=True, **kwargs)[source]

Bases: underworld._stgermain.StgCompoundComponent

This class provides functionality for a discrete representation of the steady state darcy flow equation.

The class uses a standard Galerkin finite element method to construct a system of linear equations which may then be solved using an object of the underworld.system.Solver class.

The underlying element types are determined by the supporting mesh used for the ‘pressureField’.

The strong form of the given boundary value problem, for \(f\), \(q\) and \(h\) given, is

\[\begin{split}\begin{align} q_i =& \kappa \, ( -u_{,i} + S_i ) & \\ q_{i,i} =& \: f & \text{ in } \Omega \\ u =& \: q & \text{ on } \Gamma_q \\ -q_i n_i =& \: h & \text{ on } \Gamma_h \\ \end{align}\end{split}\]

where,

  • \(\kappa\) is the diffusivity, \(u\) is the pressure,
  • \(S\) is a flow body-source, for example due to gravity,
  • \(f\) is a source term, \(q\) is the Dirichlet condition, and
  • \(h\) is a Neumann condition.

The problem boundary, \(\Gamma\), admits the decomposition \(\Gamma=\Gamma_q\cup\Gamma_h\) where \(\emptyset=\Gamma_q\cap\Gamma_h\). The equivalent weak form is:

\[-\int_{\Omega} w_{,i} \, q_i \, d \Omega = \int_{\Omega} w \, f \, d\Omega + \int_{\Gamma_h} w \, h \, d \Gamma\]

where we must find \(u\) which satisfies the above for all \(w\) in some variational space.

Parameters:
  • pressureField (underworld.mesh.MeshVariable) – The solution field for pressure.
  • fn_diffusivity (underworld.function.Function) – The function that defines the diffusivity across the domain.
  • fn_bodyforce (underworld.function.Function) – A function that defines the flow body-force across the domain, for example gravity. Must be a vector. Optional.
  • voronoi_swarm (underworld.swarm.Swarm) – A swarm with just one particle within each cell should be provided. This avoids the evaluation of the velocity on nodes and inaccuracies arising from diffusivity changes within cells. If a swarm is provided, voronoi type numerical integration is utilised. The provided swarm is used as the basis for the voronoi integration. If no voronoi_swarm is provided, Gauss integration is used.
  • conditions (underworld.conditions.SystemCondition) – Numerical conditions to impose on the system. This should be supplied as the condition itself, or a list object containing the conditions.
  • velocityField (underworld.mesh.MeshVariable) – Solution field for darcy flow velocity. Optional.
  • swarmVarVelocity (undeworld.swarm.SwarmVariable) – If a swarm variable is provided, the velocity calculated on the swarm will be stored. This is the most representative velocity data object, as the velocity calculation occurs on the swarm, away from mesh nodes. Optional.

Notes

Constructor must be called collectively by all processes.

fn_bodyforce

The heating function. You may change this function directly via this property.

fn_diffusivity

The diffusivity function. You may change this function directly via this property.

class underworld.systems.SteadyStateHeat(temperatureField, fn_diffusivity, fn_heating=0.0, voronoi_swarm=None, conditions=[], _removeBCs=True, **kwargs)[source]

Bases: underworld._stgermain.StgCompoundComponent

This class provides functionality for a discrete representation of the steady state heat equation.

The class uses a standard Galerkin finite element method to construct a system of linear equations which may then be solved using an object of the underworld.system.Solver class.

The underlying element types are determined by the supporting mesh used for the ‘temperatureField’.

The strong form of the given boundary value problem, for \(f\), \(h\) and \(h\) given, is

\[\begin{split}\begin{align} q_i =& - \alpha \, u_{,i} & \\ q_{i,i} =& \: f & \text{ in } \Omega \\ u =& \: g & \text{ on } \Gamma_g \\ -q_i n_i =& \: h & \text{ on } \Gamma_h \\ \end{align}\end{split}\]

where, \(\alpha\) is the diffusivity, \(u\) is the temperature, \(f\) is a source term, \(g\) is the Dirichlet condition, and \(h\) is a Neumann condition. The problem boundary, \(\Gamma\), admits the decomposition \(\Gamma=\Gamma_g\cup\Gamma_h\) where \(\emptyset=\Gamma_g\cap\Gamma_h\). The equivalent weak form is:

\[-\int_{\Omega} w_{,i} \, q_i \, d \Omega = \int_{\Omega} w \, f \, d\Omega + \int_{\Gamma_h} w \, h \, d \Gamma\]

where we must find \(u\) which satisfies the above for all \(w\) in some variational space.

Parameters:
  • temperatureField (underworld.mesh.MeshVariable) – The solution field for temperature.
  • fn_diffusivity (underworld.function.Function) – The function that defines the diffusivity across the domain.
  • fn_heating (underworld.function.Function) – A function that defines the heating across the domain. Optional.
  • voronoi_swarm (underworld.swarm.Swarm) – If a voronoi_swarm is provided, voronoi type numerical integration is utilised. The provided swarm is used as the basis for the voronoi integration. If no voronoi_swarm is provided, Gauss integration is used.
  • conditions (underworld.conditions.SystemCondition) – Numerical conditions to impose on the system. This should be supplied as the condition itself, or a list object containing the conditions.

Notes

Constructor must be called collectively by all processes.

Example

Setup a basic thermal system:

>>> linearMesh = uw.mesh.FeMesh_Cartesian( elementType='Q1/dQ0', elementRes=(4,4), minCoord=(0.,0.), maxCoord=(1.,1.) )
>>> tField = uw.mesh.MeshVariable( linearMesh, 1 )
>>> topNodes = linearMesh.specialSets["MaxJ_VertexSet"]
>>> bottomNodes = linearMesh.specialSets["MinJ_VertexSet"]
>>> tbcs = uw.conditions.DirichletCondition(tField, topNodes + bottomNodes)
>>> tField.data[topNodes.data] = 0.0
>>> tField.data[bottomNodes.data] = 1.0
>>> tSystem = uw.systems.SteadyStateHeat(temperatureField=tField, fn_diffusivity=1.0, conditions=[tbcs])

Example with non diffusivity:

>>> k = tField + 1.0
>>> tSystem = uw.systems.SteadyStateHeat(temperatureField=tField, fn_diffusivity=k, conditions=[tbcs])
>>> solver = uw.systems.Solver(tSystem)
>>> solver.solve()
Traceback (most recent call last):
...
RuntimeError: Nonlinearity detected.
Diffusivity function depends on the temperature field provided to the system.
Please set the 'nonLinearIterate' solve parameter to 'True' or 'False' to continue.
>>> solver.solve(nonLinearIterate=True)
fn_diffusivity

The diffusivity function. You may change this function directly via this property.

fn_heating

The heating function. You may change this function directly via this property.

class underworld.systems.Stokes(velocityField, pressureField, fn_viscosity, fn_bodyforce=None, fn_one_on_lambda=None, fn_source=None, voronoi_swarm=None, conditions=[], _removeBCs=True, _fn_viscosity2=None, _fn_director=None, fn_stresshistory=None, _fn_stresshistory=None, _fn_v0=None, _fn_p0=None, _callback_post_solve=None, **kwargs)[source]

Bases: underworld._stgermain.StgCompoundComponent

This class provides functionality for a discrete representation of the Stokes flow equations.

Specifically, the class uses a mixed finite element method to construct a system of linear equations which may then be solved using an object of the underworld.system.Solver class.

The underlying element types are determined by the supporting mesh used for the ‘velocityField’ and ‘pressureField’ parameters.

The strong form of the given boundary value problem, for \(f\), \(g\) and \(h\) given, is

\[\begin{split}\begin{align} \sigma_{ij,j} + f_i =& \: 0 & \text{ in } \Omega \\ u_{k,k} + \frac{p}{\lambda} =& \: H & \text{ in } \Omega \\ u_i =& \: g_i & \text{ on } \Gamma_{g_i} \\ \sigma_{ij}n_j =& \: h_i & \text{ on } \Gamma_{h_i} \\ \end{align}\end{split}\]

where,

  • \(\sigma_{i,j}\) is the stress tensor
  • \(u_i\) is the velocity,
  • \(p\) is the pressure,
  • \(f_i\) is a body force,
  • \(\lambda\) is pseudo compressibility factor,
  • \(H\) is the compressible equation source term,
  • \(g_i\) are the velocity boundary conditions (DirichletCondition)
  • \(h_i\) are the traction boundary conditions (NeumannCondition).

The problem boundary, \(\Gamma\), admits the decompositions \(\Gamma=\Gamma_{g_i}\cup\Gamma_{h_i}\) where \(\emptyset=\Gamma_{g_i}\cap\Gamma_{h_i}\). The equivalent weak form is:

\[\int_{\Omega} w_{(i,j)} \sigma_{ij} \, d \Omega = \int_{\Omega} w_i \, f_i \, d\Omega + \sum_{j=1}^{n_{sd}} \int_{\Gamma_{h_j}} w_i \, h_i \, d \Gamma\]

where we must find \(u\) which satisfies the above for all \(w\) in some variational space.

Parameters:
  • velocityField (underworld.mesh.MeshVariable) – Variable used to record system velocity.
  • pressureField (underworld.mesh.MeshVariable) – Variable used to record system pressure.
  • fn_viscosity (underworld.function.Function) – Function which reports a viscosity value. Function must return scalar float values.
  • fn_bodyforce (underworld.function.Function, Default = None) – Function which reports a body force for the system. Function must return float values of identical dimensionality to the provided velocity variable.
  • fn_one_on_lambda (underworld.function.Function, Default = None) – Pseudo-compressibility factor. Note that non-zero values are incompatible with the ‘penalty’ stokes solver. Ensure a ‘penalty’ equal to 0 is used if this function is non-zero. By default this is the case.
  • fn_source (underworld.function.Function, Default = None) – Mass source term. Check fn_one_on_lambda for usage caveats.
  • fn_stresshistory (underworld.function.Function, Default = None) – Function which defines the stress history term used for viscoelasticity. Function is a vector of size 3 (dim=2) or 6 (dim=3) representing a symetric tensor.
  • voronoi_swarm (underworld.swarm.Swarm) – If a voronoi_swarm is provided, voronoi type numerical integration is utilised. The provided swarm is used as the basis for the voronoi integration. If no voronoi_swarm is provided, Gauss integration is used.
  • conditions (underworld.conditions.SystemCondition) – Numerical conditions to impose on the system. This should be supplied as the condition itself, or a list object containing the conditions.

Notes

Constructor must be called by collectively all processes.

eqResiduals

Returns the stokes flow equations’ residuals from the latest solve. Residual calculations use the matrices and vectors of the discretised problem. The residuals correspond to the momentum equation and the continuity equation.

Returns:r1 is the momentum equation residual r2 is the continuity equation residual
Return type:(r1, r2) - 2 tuple of doubles

Notes

This method must be called collectively by all processes.

fn_bodyforce

The body force function. You may change this function directly via this property.

fn_one_on_lambda

A bulk viscosity parameter

fn_source

The volumetric source term function. You may change this function directly via this property.

fn_viscosity

The viscosity function. You may change this function directly via this property.

stokes_callback

Return the callback function used by this system

velocity_rms()[source]

Calculates RMS velocity as follows

\[v_{rms} = \sqrt{ \frac{ \int_V (\mathbf{v}.\mathbf{v}) \, \mathrm{d}V } {\int_V \, \mathrm{d}V} }\]
class underworld.systems.StokesSolver(stokesSLE, **kwargs)[source]

Bases: underworld._stgermain.StgCompoundComponent

The Block Stokes Schur Complement Solver: This solves the saddle-point system

\[\begin{split}\begin{bmatrix} K & G \\ G^T & C \end{bmatrix} \begin{bmatrix} u \\ p \end{bmatrix} = \begin{bmatrix}f \\ h \end{bmatrix}\end{split}\]

via a Schur complement method.

We first solve:

(1)\[S p= G^T K^{-1} f - h,\]

where \(S = G^T K^{-1} G-C\)

Then we backsolve for the velocity:

(2)\[K u = f - G p.\]

The effect of \(K^{-1}\) in (1) is obtained via a KSPSolve in PETSc. This has the prefix ‘A11’ (often called the ‘inner’ solve)

The solve in (1) for the pressure has prefix ‘scr’.

Assuming the returned solver is called ‘solver’, it is possible to configure these solves individually via the solver.options.A11 and solver.options.scr dictionaries.

Try help(solver.options.A11) for some details.

Common configurations are provided via the set_inner_method() method.

help(solver.set_inner_method) for more.

For more advanced configurations use the solver.options.A11/scr dictionaries directly.

help(solver.options) to see more.

set_inner_method(solve_type='mg')[source]

Configure velocity/inner solver (A11 PETSc prefix).

Available options below. Note that associated solver software (for example mumps) must be installed.

  • mg : Geometric multigrid (default).
  • nomg : Disables multigrid.
  • lu : LU direct solver (serial only).
  • mumps : MUMPS parallel direct solver.
  • superludist : SuperLU parallel direct solver.
  • superlu : SuperLU direct solver (serial only).
set_mg_levels(levels)[source]

Set the number of multigrid levels manually. It is set automatically by default.

set_penalty(penalty)[source]

By setting the penalty, the Augmented Lagrangian Method is used as the solve. This method is not recommended for normal use as there is additional memory and cpu overhead. This method can often help improve convergence issues for problems with large viscosity contrasts that are having trouble converging.

A penalty of roughly 0.1 of the maximum viscosity contrast is not a bad place to start as a rule of thumb. (check notes/paper)

solve(nonLinearIterate=None, nonLinearTolerance=0.01, nonLinearKillNonConvergent=False, nonLinearMinIterations=1, nonLinearMaxIterations=500, callback_post_solve=None, print_stats=False, reinitialise=True, **kwargs)[source]

Solve the stokes system

Parameters:
  • nonLinearIterate (bool) – True will perform non linear iterations iterations, False (or 0) will not
  • nonLinearTolerance (float, Default=1.0e-2) – Relative tolerance criterion for the change in the velocity field
  • nonLinearMaxIterations (int, Default=500) – Maximum number of non linear iteration to perform
  • callback_post_sovle (func, Default=None) – Optional callback function to be performed at the end of a linear solve iteration. Commonly this will be used to perform operations between non linear iterations, for example, calibrating the pressure solution or removing the system null space.
  • print_stats (bool, Default=False) – Print out solver iteration and timing counts per solver
  • reinitialise (bool, Default=True,) – Rebuild the system discretisation storage (location matrix/petsc mats & vecs) and repopulate, if available, the stokes voronio swarm before the system is solved.
class underworld.systems.SwarmAdvector(velocityField, swarm, order=2, **kwargs)[source]

Bases: underworld.systems._timeintegration.TimeIntegration

Objects of this class advect a swarm through time using the provided velocity field.

Parameters:
integrate(dt, update_owners=True)[source]

Integrate the associated swarm in time, by dt, using the velocityfield that is associated with this class

Parameters:
  • dt (double) – The timestep to use in the intergration
  • update_owners (bool) – If this is set to False, particle ownership (which element owns a particular particle) is not updated after advection. This is often necessary when both the mesh and particles are advecting simutaneously.

Example

>>> import underworld as uw
>>> import numpy as np
>>> from underworld import function as fn
>>> dim=2;
>>> elementMesh = uw.mesh.FeMesh_Cartesian(elementType="Q1/dQ0", elementRes=(9,9), minCoord=(-1.,-1.), maxCoord=(1.,1.))
>>> velocityField = uw.mesh.MeshVariable( mesh=elementMesh, nodeDofCount=dim )
>>> swarm = uw.swarm.Swarm(mesh=elementMesh)
>>> particle = np.zeros((1,2))
>>> particle[0] = [0.2,-0.2]
>>> swarm.add_particles_with_coordinates(particle)
array([0], dtype=int32)
>>> velocityField.data[:]=[1.0,1.0]
>>> swarmAdvector = uw.systems.SwarmAdvector(velocityField=velocityField, swarm=swarm, order=2)
>>> dt=swarmAdvector.get_max_dt()
>>> swarmAdvector.integrate(dt)
>>> np.allclose(swarm.particleCoordinates.data[0], [ 0.27856742, -0.12143258], rtol=1e-4)
True
class underworld.systems.TimeIntegration(order, **kwargs)[source]

Bases: underworld._stgermain.StgCompoundComponent

Abstract class for integrating numerical objects (fields, swarms, etc.) in time.

The integration algorithm is a modified Runge Kutta method that only evaluates midpoint information varying in space - using only the present timestep solution. The order of the integration used can be 1,2,4

Parameters:order (int {1,2,4}) – Defines the numerical order ‘in space’ of the Runge Kutta like integration scheme.
dt

Time integrator timestep size.

time

Time integrator time value.