Econforgeinterp

class HARK.econforgeinterp.DecayInterp(interp, limit_fun, limit_grad=None, extrap_method='decay_prop')

Bases: MetricObject

A class of interpolators that use a limiting function for extrapolation.

See HARK/examples/Interpolation/DecayInterp.ipynb for examples of how to use this class.

distance_criteria = ['interp']

extrap_decay_hark(x, closest_x)

“decay_hark” extrapolation method. Takes into account the rate at which the interpolator and limiting function are approaching at the edge of the grid for combining them.

Parameters:

• x (inputs that require extrapolation.)

• closest_x (for each of the inputs that require extrapolation, contains) – the closest point that falls inside the grid.

extrap_decay_prop(x, closest_x)

“decay_prop” extrapolation method. Combines the interpolator’s default extrapolation and the limiting function with weights that depend on how far from the grid the values are.

Parameters:

• x (inputs that require extrapolation.)

• closest_x (for each of the inputs that require extrapolation, contains) – the closest point that falls inside the grid.

extrap_paste(x, closest_x)

“paste” extrapolation method. Uses the limiting function for extrapolation, but with a vertical shift that “pastes” it to the interpolator at the edge of the grid.

Parameters:

• x (inputs that require extrapolation.)

• closest_x (for each of the inputs that require extrapolation, contains) – the closest point that falls inside the grid.

class HARK.econforgeinterp.LinearFast(f_val, grids, extrap_mode='linear', prebuilt_grid=None)

Bases: MetricObject

A class that constructs and holds all the necessary elements to call a multilinear interpolator from econforge.interpolator in a way that resembles the basic interpolators in HARK.interpolation.

distance_criteria = ['f_val', 'grid_list']

gradient(*args)

Evaluates gradient of the interpolator.

Parameters:

args ([numpy.array]) – List of arrays. The i-th entry contains the i-th coordinate of all the points to be evaluated. All entries must have the same shape.

Returns:

List of the derivatives of the function with respect to each input, evaluated at the given points. E.g. if the interpolator represents 3D function f, f.gradient(x,y,z) will return [df/dx(x,y,z), df/dy(x,y,z), df/dz(x,y,z)]. Each element has the shape of items in args.

Return type:

[numpy.array]