robustData and blockyModel

[makeabhishek]

What is the difference between robustData – Robust (L1 norm mimicking) data reweighting and blockyModel – Robust (L1 norm mimicking) model roughness reweighting .

What is different mathematical formulation for each ?

[halbmy]

It is, as the description says, an L1 norm minimization (actually an L2 norm with adapted weights):

• one (robustData) for the data misfit, i.e. to reduce the effect of outliers
• the other (blockModel) for the model roughness, i.e. to enhance gradients in the model
  Both can be independently set.

[makeabhishek]

Thanks for the description. Can we use both together?

Is there reference to look the mathematical description for the two conditions robustData and blockModel ?

[halbmy]

Yes, setting independently means you can set both, only one, or none.

There is nothing about L1 norm written on pygimli.org but in some textbooks or papers (Claerbout & Muir 1973).