LinearDiffuser
(*args, **kwds)[source]¶Bases: landlab.core.model_component.Component
This component implements linear diffusion of a Landlab field.
Component assumes grid does not deform. If the boundary conditions on the
grid change after component instantiation, be sure to also call
updated_boundary_conditions()
to ensure these are reflected in the
component (especially if fixed_links are present).
The method keyword allows control of the way the solver works. Options other than ‘simple’ will make the component run slower, but give second order solutions that incorporate information from more distal nodes (i.e., the diagonals). Note that the option ‘resolve_on_patches’ can result in somewhat counterintuitive behaviour  this option tells the component to treat the diffusivity as a field with directionality to it (i.e., that the diffusivites are defined on links). Thus if all links have the same diffusivity value, with this flag active “effective” diffusivities at the nodes will be higher than this value (by a factor of root 2) as the diffusivity at each patch will be the mean vector sum of that at the bounding links.
The primary method of this class is run_one_step()
.
Construction:
LinearDiffuser(grid, linear_diffusivity=None, method='simple')
Parameters:  grid : ModelGrid
linear_diffusivity : float, array, or field name (m**2/time)
method : {‘simple’, ‘resolve_on_patches’, ‘on_diagonals’}
deposit : {True, False}


Examples
>>> from landlab import RasterModelGrid
>>> import numpy as np
>>> mg = RasterModelGrid((9, 9), 1.)
>>> z = mg.add_zeros('node', 'topographic__elevation')
>>> z.reshape((9, 9))[4, 4] = 1.
>>> mg.set_closed_boundaries_at_grid_edges(True, True, True, True)
>>> ld = LinearDiffuser(mg, linear_diffusivity=1.)
>>> for i in range(1):
... ld.run_one_step(1.)
>>> np.isclose(z[mg.core_nodes].sum(), 1.)
True
>>> mg2 = RasterModelGrid((5, 30), 1.)
>>> z2 = mg2.add_zeros('node', 'topographic__elevation')
>>> z2.reshape((5, 30))[2, 8] = 1.
>>> z2.reshape((5, 30))[2, 22] = 1.
>>> mg2.set_closed_boundaries_at_grid_edges(True, True, True, True)
>>> kd = mg2.node_x/mg2.node_x.mean()
>>> ld2 = LinearDiffuser(mg2, linear_diffusivity=kd)
>>> for i in range(10):
... ld2.run_one_step(0.1)
>>> z2[mg2.core_nodes].sum() == 2.
True
>>> z2.reshape((5, 30))[2, 8] > z2.reshape((5, 30))[2, 22]
True
An example using links: >>> mg1 = RasterModelGrid((10, 10), 100.) >>> mg2 = RasterModelGrid((10, 10), 100.) >>> z1 = mg1.add_zeros(‘node’, ‘topographic__elevation’) >>> z2 = mg2.add_zeros(‘node’, ‘topographic__elevation’) >>> dt = 1. >>> nt = 10 >>> kappa_links = mg2.add_ones(‘link’, ‘surface_water__discharge’) >>> kappa_links *= 10000. >>> dfn1 = LinearDiffuser(mg1, linear_diffusivity=10000.) >>> dfn2 = LinearDiffuser(mg2, linear_diffusivity=’surface_water__discharge’) >>> for i in range(nt): ... z1[mg1.core_nodes] += 1. ... z2[mg2.core_nodes] += 1. ... dfn1.run_one_step(dt) ... dfn2.run_one_step(dt) >>> np.allclose(z1, z2) True >>> z2.fill(0.) >>> dfn2 = LinearDiffuser(mg2, linear_diffusivity=’surface_water__discharge’, ... method=’resolve_on_patches’) >>> for i in range(nt): ... z2[mg2.core_nodes] += 1. ... dfn2.run_one_step(dt) >>> np.all(z2[mg2.core_nodes] < z1[mg2.core_nodes]) True
diffuse
(dt, **kwds)[source]¶See run_one_step()
.
run_one_step
(dt, **kwds)[source]¶Run the diffuser for one timestep, dt.
If the imposed timestep dt is longer than the CourantFriedrichsLewy condition for the diffusion, this timestep will be internally divided as the component runs, as needed.
Parameters:  dt : float (time)


time_step
¶Returns internal timestep size (as a property).
updated_boundary_conditions
()[source]¶Call if grid BCs are updated after component instantiation.
Sets fixed_grad_nodes, fixed_grad_anchors, & fixed_grad_offsets, such that:
value[fixed_grad_nodes] = value[fixed_grad_anchors] + offset
Examples
>>> from landlab import RasterModelGrid
>>> import numpy as np
>>> mg = RasterModelGrid((4, 5), 1.)
>>> z = mg.add_zeros('node', 'topographic__elevation')
>>> z[mg.core_nodes] = 1.
>>> ld = LinearDiffuser(mg, linear_diffusivity=1.)
>>> ld.fixed_grad_nodes.size == 0
True
>>> ld.fixed_grad_anchors.size == 0
True
>>> ld.fixed_grad_offsets.size == 0
True
>>> mg.at_link['topographic__slope'] = mg.calc_grad_at_link(
... 'topographic__elevation')
>>> mg.set_fixed_link_boundaries_at_grid_edges(True, True, True, True)
>>> ld.updated_boundary_conditions()
>>> ld.fixed_grad_nodes
array([ 1, 2, 3, 5, 9, 10, 14, 16, 17, 18])
>>> ld.fixed_grad_anchors
array([ 6, 7, 8, 6, 8, 11, 13, 11, 12, 13])
>>> ld.fixed_grad_offsets
array([1., 1., 1., 1., 1., 1., 1., 1., 1., 1.])
>>> np.allclose(z[ld.fixed_grad_nodes],
... z[ld.fixed_grad_anchors] + ld.fixed_grad_offsets)
True