Warning You're reading an old version (v5.0) of this documentation.
If you want up-to-date information, please have a look at master.
You're reading an old version (v5.0) of this documentation. If you want up-to-date information, please have a look at master.
Command Line Arguments
Normally MARE2DEM is called with just the name of the input resistivity file:
mpirun -n <np> MARE2DEM <resistivity file>
<resistivity file> is the name of the required input
Resistivity File (.resistivity). By convention, this file should have the extension
.resistivity. For example
inputModel.0.resistivity. The model found by
each inversion iteration is then output to a new resistivity file with
the iteration number incremented. For example:
inputModel.2.resistivity, … The
corresponding model responses are written to
mpirun starts up the parallel message
passing interface environment and has argument
-n <np> where
<np> should be
the number of processing cores on your system (laptop, desktop, etc).
For inversion models, MARE2DEM will also outputs the normalized data
sensitivities to the free parameters (see Normalized Sensitivity). The
sensitivity is output to the file
file>.<iter#>.sensitivity. plotMARE2DEM.m can be used to plot
the sensitivity; it can be overlain as contours or shading on inversion
models. The sensitivity can be used as a relative measure of data
sensitivity to structure, but be aware this is a linearized
approximation for what is a non-linear inverse problem, so results
should be taken with a grain of salt.
MARE2DEM has a few other command line options for more advanced usage, but most users won’t need to use these:
mpirun -n <np> MARE2DEM [-F] [-J] [-scratch <scratchfolder>] <resistivity file> [<output file>]
where the arguments in brackets  are optional:
<output file>- With this option, the output files are named
-FComputes the forward response of the input model only. The forward response is output to
<resistivity file>.resp. Note that if MARE2DEM detects no free parameters in the input model it will only compute the forward response; thus, typically the
-Fis only required when you want to compute the forward response of an inversion model (say when doing sensitivity tests of anomalous features found by inversion, where you paint over the anomaly using Mamba2D and then see how the MT or CSEM responses change).
-JOutputs the full Jacobian matrix of the input model for each iteration. This flag is provided for advanced users wishing to use the Jacobian matrix in other calculations and normal users do not need to use this; further, the output Jacobian files can be quite large. The Jacobian matrix is written to an unformatted binary data file columnwise with the naming convention
<resistivity file>.<iter#>.jacobianBin. Note that the saved J is unweighted (i.e. it has NOT been normalized by the data uncertainties).
-scratch <scratchfolder>Use the specified directory for the scratch files required for 2.5D CSEM inversion (but not needed for MT). Optimally this should be a local directory on each compute node and not a networked directory.