Sparse Inversion with Iteratively Re-Weighted Least-Squares#

Least-squares inversion produces smooth models which may not be an accurate representation of the true model. Here we demonstrate the basics of inverting for sparse and/or blocky models. Here, we used the iteratively reweighted least-squares approach. For this tutorial, we focus on the following:

Defining the forward problem

Defining the inverse problem (data misfit, regularization, optimization)

Defining the paramters for the IRLS algorithm

Specifying directives for the inversion

Recovering a set of model parameters which explains the observations

import numpy as np
import matplotlib.pyplot as plt

from discretize import TensorMesh

from SimPEG import (
    simulation,
    maps,
    data_misfit,
    directives,
    optimization,
    regularization,
    inverse_problem,
    inversion,
)

# sphinx_gallery_thumbnail_number = 3

Defining the Model and Mapping#

Here we generate a synthetic model and a mappig which goes from the model space to the row space of our linear operator.

nParam = 100  # Number of model paramters

# A 1D mesh is used to define the row-space of the linear operator.
mesh = TensorMesh([nParam])

# Creating the true model
true_model = np.zeros(mesh.nC)
true_model[mesh.cell_centers_x > 0.3] = 1.0
true_model[mesh.cell_centers_x > 0.45] = -0.5
true_model[mesh.cell_centers_x > 0.6] = 0

# Mapping from the model space to the row space of the linear operator
model_map = maps.IdentityMap(mesh)

# Plotting the true model
fig = plt.figure(figsize=(8, 5))
ax = fig.add_subplot(111)
ax.plot(mesh.cell_centers_x, true_model, "b-")
ax.set_ylim([-2, 2])

(-2.0, 2.0)

Defining the Linear Operator#

Here we define the linear operator with dimensions (nData, nParam). In practive, you may have a problem-specific linear operator which you would like to construct or load here.

# Number of data observations (rows)
nData = 20

# Create the linear operator for the tutorial. The columns of the linear operator
# represents a set of decaying and oscillating functions.
jk = np.linspace(1.0, 60.0, nData)
p = -0.25
q = 0.25


def g(k):
    return np.exp(p * jk[k] * mesh.cell_centers_x) * np.cos(
        np.pi * q * jk[k] * mesh.cell_centers_x
    )


G = np.empty((nData, nParam))

for i in range(nData):
    G[i, :] = g(i)

# Plot the columns of G
fig = plt.figure(figsize=(8, 5))
ax = fig.add_subplot(111)
for i in range(G.shape[0]):
    ax.plot(G[i, :])

ax.set_title("Columns of matrix G")

Text(0.5, 1.0, 'Columns of matrix G')

Defining the Simulation#

The simulation defines the relationship between the model parameters and predicted data.

sim = simulation.LinearSimulation(mesh, G=G, model_map=model_map)

Predict Synthetic Data#

Here, we use the true model to create synthetic data which we will subsequently invert.

# Standard deviation of Gaussian noise being added
std = 0.02
np.random.seed(1)

# Create a SimPEG data object
data_obj = sim.make_synthetic_data(true_model, noise_floor=std, add_noise=True)

Define the Inverse Problem#

The inverse problem is defined by 3 things:

Data Misfit: a measure of how well our recovered model explains the field data

Regularization: constraints placed on the recovered model and a priori information

Optimization: the numerical approach used to solve the inverse problem

# Define the data misfit. Here the data misfit is the L2 norm of the weighted
# residual between the observed data and the data predicted for a given model.
# Within the data misfit, the residual between predicted and observed data are
# normalized by the data's standard deviation.
dmis = data_misfit.L2DataMisfit(simulation=sim, data=data_obj)

# Define the regularization (model objective function). Here, 'p' defines the
# the norm of the smallness term and 'q' defines the norm of the smoothness
# term.
reg = regularization.Sparse(mesh, mapping=model_map)
reg.reference_model = np.zeros(nParam)
p = 0.0
q = 0.0
reg.norms = [p, q]

# Define how the optimization problem is solved.
opt = optimization.ProjectedGNCG(
    maxIter=100, lower=-2.0, upper=2.0, maxIterLS=20, maxIterCG=30, tolCG=1e-4
)

# Here we define the inverse problem that is to be solved
inv_prob = inverse_problem.BaseInvProblem(dmis, reg, opt)

Define Inversion Directives#

Here we define any directiveas that are carried out during the inversion. This includes the cooling schedule for the trade-off parameter (beta), stopping criteria for the inversion and saving inversion results at each iteration.

# Add sensitivity weights but don't update at each beta
sensitivity_weights = directives.UpdateSensitivityWeights(every_iteration=False)

# Reach target misfit for L2 solution, then use IRLS until model stops changing.
IRLS = directives.Update_IRLS(max_irls_iterations=40, minGNiter=1, f_min_change=1e-4)

# Defining a starting value for the trade-off parameter (beta) between the data
# misfit and the regularization.
starting_beta = directives.BetaEstimate_ByEig(beta0_ratio=1e0)

# Update the preconditionner
update_Jacobi = directives.UpdatePreconditioner()

# Save output at each iteration
saveDict = directives.SaveOutputEveryIteration(save_txt=False)

# Define the directives as a list
directives_list = [sensitivity_weights, IRLS, starting_beta, update_Jacobi, saveDict]

Setting a Starting Model and Running the Inversion#

To define the inversion object, we need to define the inversion problem and the set of directives. We can then run the inversion.

# Here we combine the inverse problem and the set of directives
inv = inversion.BaseInversion(inv_prob, directives_list)

# Starting model
starting_model = 1e-4 * np.ones(nParam)

# Run inversion
recovered_model = inv.run(starting_model)

                    SimPEG.InvProblem is setting bfgsH0 to the inverse of the eval2Deriv.
                    ***Done using the default solver Pardiso and no solver_opts.***

model has any nan: 0
=============================== Projected GNCG ===============================
  #     beta     phi_d     phi_m       f      |proj(x-g)-x|  LS    Comment
-----------------------------------------------------------------------------
x0 has any nan: 0
2.26e+06  3.70e+03  1.02e-09  3.70e+03    2.00e+01      0
1.13e+06  2.09e+03  2.64e-04  2.38e+03    1.94e+01      0
5.65e+05  1.50e+03  6.42e-04  1.86e+03    1.90e+01      0   Skip BFGS
2.82e+05  9.23e+02  1.37e-03  1.31e+03    1.85e+01      0   Skip BFGS
1.41e+05  4.79e+02  2.48e-03  8.29e+02    1.72e+01      0   Skip BFGS
7.06e+04  2.12e+02  3.80e-03  4.79e+02    1.48e+01      0   Skip BFGS
3.53e+04  8.27e+01  5.05e-03  2.61e+02    1.32e+01      0   Skip BFGS
1.76e+04  3.10e+01  6.05e-03  1.38e+02    1.05e+01      0   Skip BFGS
Reached starting chifact with l2-norm regularization: Start IRLS steps...
irls_threshold 1.218797708962581
8.82e+03  1.27e+01  9.13e-03  9.33e+01    4.61e+00      0   Skip BFGS
1.75e+04  1.02e+01  1.05e-02  1.94e+02    1.73e+01      0
1.33e+04  2.58e+01  1.02e-02  1.61e+02    4.05e+00      0
1.03e+04  2.53e+01  1.09e-02  1.38e+02    3.03e+00      0   Skip BFGS
8.07e+03  2.43e+01  1.15e-02  1.17e+02    4.33e+00      0   Skip BFGS
6.51e+03  2.31e+01  1.17e-02  9.95e+01    5.08e+00      0   Skip BFGS
6.51e+03  2.17e+01  1.17e-02  9.79e+01    9.12e+00      0   Skip BFGS
5.30e+03  2.28e+01  1.11e-02  8.19e+01    6.64e+00      0
5.30e+03  2.05e+01  1.07e-02  7.70e+01    7.88e+00      0
5.30e+03  2.00e+01  9.84e-03  7.22e+01    9.09e+00      0
5.30e+03  1.95e+01  8.97e-03  6.70e+01    9.05e+00      0
5.30e+03  1.88e+01  8.09e-03  6.17e+01    9.64e+00      0
8.28e+03  1.79e+01  7.21e-03  7.75e+01    1.62e+01      0
8.28e+03  2.11e+01  6.13e-03  7.18e+01    1.21e+01      0
8.28e+03  2.08e+01  5.33e-03  6.50e+01    1.29e+01      0
8.28e+03  2.02e+01  4.57e-03  5.80e+01    1.75e+01      0
8.28e+03  1.89e+01  3.83e-03  5.06e+01    1.18e+01      0
1.31e+04  1.70e+01  3.12e-03  5.80e+01    1.67e+01      0
2.06e+04  1.76e+01  2.48e-03  6.87e+01    1.73e+01      0
2.06e+04  1.93e+01  2.01e-03  6.07e+01    1.21e+01      0
2.06e+04  1.82e+01  1.66e-03  5.23e+01    1.19e+01      0
3.35e+04  1.60e+01  1.35e-03  6.13e+01    1.80e+01      0
5.38e+04  1.65e+01  1.09e-03  7.50e+01    1.83e+01      0
5.38e+04  1.86e+01  8.51e-04  6.44e+01    1.27e+01      0
8.45e+04  1.75e+01  7.01e-04  7.67e+01    1.78e+01      0
8.45e+04  1.90e+01  5.68e-04  6.70e+01    1.26e+01      0
8.45e+04  1.81e+01  4.63e-04  5.72e+01    1.26e+01      0
1.36e+05  1.64e+01  3.87e-04  6.91e+01    1.81e+01      0
2.14e+05  1.74e+01  3.10e-04  8.39e+01    1.81e+01      0
2.14e+05  1.99e+01  2.54e-04  7.42e+01    1.27e+01      0
2.14e+05  1.94e+01  2.11e-04  6.46e+01    1.25e+01      0
3.34e+05  1.79e+01  1.77e-04  7.72e+01    1.77e+01      0
3.34e+05  1.92e+01  1.47e-04  6.81e+01    1.26e+01      0
3.34e+05  1.81e+01  1.23e-04  5.92e+01    1.25e+01      0
5.37e+05  1.65e+01  1.04e-04  7.23e+01    1.76e+01      0
8.42e+05  1.76e+01  8.61e-05  9.01e+01    1.78e+01      0
8.42e+05  2.01e+01  7.08e-05  7.97e+01    1.27e+01      0
8.42e+05  1.94e+01  5.95e-05  6.95e+01    1.27e+01      0
1.32e+06  1.75e+01  4.96e-05  8.31e+01    1.74e+01      0
1.32e+06  1.85e+01  3.95e-05  7.08e+01    1.27e+01      0
Reach maximum number of IRLS cycles: 40
------------------------- STOP! -------------------------
: |fc-fOld| = 0.0000e+00 <= tolF*(1+|f0|) = 3.6971e+02
: |xc-x_last| = 7.3454e-02 <= tolX*(1+|x0|) = 1.0010e-01
: |proj(x-g)-x|    = 1.2693e+01 <= tolG          = 1.0000e-01
: |proj(x-g)-x|    = 1.2693e+01 <= 1e3*eps       = 1.0000e-02
: maxIter   =     100    <= iter          =     48
------------------------- DONE! -------------------------

Plotting Results#

fig, ax = plt.subplots(1, 2, figsize=(12 * 1.2, 4 * 1.2))

# True versus recovered model
ax[0].plot(mesh.cell_centers_x, true_model, "k-")
ax[0].plot(mesh.cell_centers_x, inv_prob.l2model, "b-")
ax[0].plot(mesh.cell_centers_x, recovered_model, "r-")
ax[0].legend(("True Model", "Recovered L2 Model", "Recovered Sparse Model"))
ax[0].set_ylim([-2, 2])

# Observed versus predicted data
ax[1].plot(data_obj.dobs, "k-")
ax[1].plot(inv_prob.dpred, "ko")
ax[1].legend(("Observed Data", "Predicted Data"))

# Plot convergence
fig = plt.figure(figsize=(9, 5))
ax = fig.add_axes([0.2, 0.1, 0.7, 0.85])
ax.plot(saveDict.phi_d, "k", lw=2)

twin = ax.twinx()
twin.plot(saveDict.phi_m, "k--", lw=2)
ax.plot(np.r_[IRLS.iterStart, IRLS.iterStart], np.r_[0, np.max(saveDict.phi_d)], "k:")
ax.text(
    IRLS.iterStart,
    0.0,
    "IRLS Start",
    va="bottom",
    ha="center",
    rotation="vertical",
    size=12,
    bbox={"facecolor": "white"},
)

ax.set_ylabel(r"$\phi_d$", size=16, rotation=0)
ax.set_xlabel("Iterations", size=14)
twin.set_ylabel(r"$\phi_m$", size=16, rotation=0)

Text(865.1527777777777, 0.5, '$\\phi_m$')

Total running time of the script: (0 minutes 28.099 seconds)

Estimated memory usage: 8 MB

Download Jupyter notebook: plot_inv_2_inversion_irls.ipynb

Download Python source code: plot_inv_2_inversion_irls.py

Gallery generated by Sphinx-Gallery