GERMAINE v1.2

New GERMAINE module for modelling and FWI of TE-mode georadar data according to Lavoué et al. (2014)

This includes:

• Parameter scaling for permittivity/conductivity FWI
• Tikhonov regularization
• Preconditioned l-BFGS optimization (Nocedal & Wright 2006, Métivier & Brossier 2016) using
  the Approximate or Pseudo-Hessian
• Laplace damping
• Multiple Jupyter notebooks for FD data/wavefield visualization and computation of TD radargrams

GERMAINE v1.1

• 2-level MPI parallelization of shots and frequency groups using MPI communicator splitting
• Option to read external source wavelet from SU file and DFT to FD
• New misfit functions: logarithmic phase-amplitude and phase only according to Shin & Min (2006)
  and Bednar et al. (2007)
• Complex frequencies (Shin & Cha 2009, Kamei et al. 2012)
• Free surface boundary condition
• Jupyter notebook to transform GERMAINE FD data to TD for comparison with DENISE results
• multiple smaller bug fixes

First release of GERMAINE

Features:

• 2D Frequency Domain Finite-Difference (FDFD) Code GERMAINE solving the 2D Helmholtz equation using a 9-point FD stencil with CFS-PML absorbing boundary conditions according to

  I. Singer, E. Turkel, 2004, A perfectly matched layer for the Helmholtz equation in a semi-infinite strip. Journal of Computational Physics, 201(2), 439-465.

  Z. Chen, D. Cheng, W. Feng, H. Yang, 2013, An optimal 9-point finite difference scheme for the Helmholtz equation with PML, Int. J. Numer. Anal. Model., 10, 389-410.

• The forward wavefield is calculated via a LU-decompostion and forward/backward substitution using UMFPACK, which is part of the sparse matrix library SuiteSparse: http://faculty.cse.tamu.edu/davis/suitesparse.html

• The code is parallelized with MPI using a very simple shot parallelization

• The FWI code is based on the adjoint state-method with CG and quasi-Newton l-BFGS optimization (Nocedal & Wright 2006).

• Different approximations of the Hessian diagonal elements: approximate Hessian (Pratt et al. 1998, Operto et al. 2006), Pseudo-Hessian (Shin et al. 2001)

• FD Reverse Time Migration