Reading this work Scalable posterior approximations for large-scale Bayesian inverse ...https://arxiv.org/abs/1510.06053 by Cui, Marzouk and Willcox, can Julia Mads and in particular the calibrate module be extended to squared Hellinger distance in the data-misfit function, the likelihood-informed parameter subspace and likelihood-informed state reduction? Likelihood informed approaches appear in computational statistics but I see in this work that it is also used in geophysical applications as demonstrated in the groundwater aquifer inversion example which is close to the one you use in A Computationally Efficient.... and Mads examples.
Monty,
Reading this work Scalable posterior approximations for large-scale Bayesian inverse ...https://arxiv.org/abs/1510.06053 by Cui, Marzouk and Willcox, can Julia Mads and in particular the calibrate module be extended to squared Hellinger distance in the data-misfit function, the likelihood-informed parameter subspace and likelihood-informed state reduction? Likelihood informed approaches appear in computational statistics but I see in this work that it is also used in geophysical applications as demonstrated in the groundwater aquifer inversion example which is close to the one you use in A Computationally Efficient.... and Mads examples.