Overview
Solvers and processing Programs for inverse problems
sigclear-optim provides Programs and liboptim for fitting observations, predictive filtering, delay and dip estimation, denoising and regularized inversion.
Quick example
Fit a parameterized model
sgfit --default=pressure,altitude \
model=linear optim=mse \
<observations.sg >prediction.sgUse man sgfit for model definitions, objectives and quality-control targets.
Included Programs
Apply fitting and inversion methods
| Program | Purpose |
|---|---|
sgfit | Fit parameterized models group by group. |
sgfiltertv | Denoise data with total-variation regularization. |
sgfilterlp | Apply linear predictive filtering and harmonic decomposition. |
sgfilterfx | Apply smoothing in the frequency-space domain. |
sgdelay | Estimate nonstationary delay through phase approximation and inversion. |
sgdippwd | Estimate dip through plane-wave destruction. |
Open a Program reference with man PROGRAM, for example man sgfit.
Libraries and APIs
Develop specialized solvers with liboptim
The Package installs liboptim and its public optim.h interface. Custom Programs can combine these solvers with SOIG data access and DSP operations.
Model choices
Match the objective to the research problem
Mean squared error
Model approximately Gaussian residuals.
Mean absolute error
Use a robust objective suited to Laplacian residuals.
Regularization
Apply smoothness or structure-preserving prior information.
Custom operators
Connect specialized forward and adjoint operations.