NA_options

Contents

Data Structure: NA_options
Description	NA inversion algorithm control data
Usage	NA_options NUM=ival where ival is the data structure number

Description

Overview

The NA_options data structure defines the control options for the nearest neighbour inverse identification algorithm, also known as the Neighbourhood Algorithm (NA).

References

•M. Sambridge, 1999a. Geophysical inversion with a neighbourhood algorithm - I. Searching the parameter space. Geophys. J. Int. , Vo.138, p.479-494.

•M. Sambridge, 1999b. Geophysical inversion with a neighbourhood algorithm - II. Appraising the ensemble. Geophys. J. Int. , Vo.138, p.727-746.

Algorithm Inversion algorithm

Usage

Algorithm

Character Name

Description

Defines the inversion algorithm type. Valid options are:

•"Fixed" - Fixed number of iterations (models)

•"Auto" - Terminate once the convergence tolerance is reached

Parallel_flag Parallel flag

Usage

Parallel_flag

Integer Value

Description

Defines the flag for carrying out the misfit model simulations in parallel (default) or sequentially. Valid options are:

•0 - Disable parallel (i.e. sequential)

•1 - Enable parallel (default)

Maximum_num_samples Number of samplings

Usage

Maximum_num_samples

Integer Value

Description

Maximum_num_samples defines the number of samplings to perform for the nearest neighbour inverse algorithm (default 20).

Num_models_in_initial_sample Number of models in initial sample

Usage

Num_models_in_initial_sample

Integer Value

Description

Num_models_in_initial_sample defines the number of models in the initial sample. This must be greater or equal to the number of models for subsequent iterations. By default Num_models_in_initial_sample = Num_models_in_sample.

Num_models_in_sample Number of models for each subsequent sampling

Usage

Num_models_in_sample

Integer Value

Description

Num_models_in_sample defines the number of models for each subsequent sampling (Default 10). By default Num_models_in_sample = Num_models_in_initial_sample.

Num_models_in_resample Number of models for neighbourhood resampling

Usage

Num_models_in_resample

Integer Value

Description

Num_models_in_resample defines the number of models for neighbourhood resampling ( nr ). The parameter nr influences the nature of the neighbourhood algorithm search which is exploratory in nature; i.e. for larger values, the sampling is spread over more cells, and conversely, for smaller values, it is restricted to fewer cells and the sampling is more localized. By default, the number of models in the resamples is set to half the number in samples. i.e. on the subsequent iteration two new models will be generated in the most optimal half of the current sample set.

Convergence_tolerance Convergence tolerance for inversion

Usage

Convergence_tolerance

Real Value

Description

Defines the convergence tolerance for the inversion process. If Convergence_tolerance is specified, the iterative process is terminated once a model achieves a misfit below the Convergence_tolerance. If Convergence_tolerance is not specified then the specified Maximum_num_samples will be performed.

Number_models_output Number of best models for output

Usage

Number_models_output

Integer Value

Description

Defines the number of best models for output to the output file. The models are output in the order of minimum misfit error.

Model_delete_option Flag options to define how many of the model results files to save or delete

Usage

Model_delete_option

Character Name

Description

Model_delete_option defines the flag options to define how many of the model results files to save or delete at the end of the optimisation procedure. Valid options are:

•"DeleteAll" - Delete all model files

•"KeepAll" - Keep all model files

•"KeepBest" - (Default) Keep best files with extensions specified by Model_results_keep. If k1\ Model_results_keep is not defined, the default is to delete all results files with the exception of data files, history files and error files.

Notes

•Model_delete_option set to "KeepBest" works in conjunction with Number_models_output. The number of best models (i.e. models with lower misfit values) kept at the end of each resample will be defined by Number_models_output.

•The type of model results files to be saved or deleted are defined by Model_clean_option and Model_results_keep.

•For Model_delete_option, Model_clean_option, Model_results_keep and Number_models_output to work in conjunction with each other, the ParaGeoInv inverse filename (.inp) and the template data filename (.dat) must be the same.

Model_clean_option Flag options to define the type of model results files to save or delete

Usage

Model_clean_option

Character Name

Description

Model_clean_option defines the flag options to define the type of model results files to save or delete at the end of the optimisation procedure. Valid options are:

•"KeepAll" - Keep all model results type (default)

•"KeepSpec" - Keep the results specified in\Model_results_keep

Model_results_keep List of output file types to keep

Usage

Model_results_keep		IDM=ival
Character Names
where	ival	Number of extension names

Description

Model_results_keep defines the list of output file types to keep. If not defined, by default, all output files are deleted with the exception of data files, history files and error files. Valid file types are:

•"History" - History files

•"Geometry" - Geometry plot files

•"PlotFile" - Result plot files

•"Grid" - Grid plot files

•"Mesh" - Mesh plot files

•"Result" - Result plot files

•"Restart" - Restart files

Output_level Output level for log file

Usage

Output_level

Integer Value

Description

Output level for log file. Valid values are:

•0 - Minimum output

•1 - Standard

•2 - Verbose (default)

Output_screen_level Output level for screen output

Usage

Output_screen_level

Character Name

Description

Output level for screen output. Valid values are:

•0 - Minimum output

•1 - Standard (default)

•2 - Verbose

Steady_state_num_sample Number of samples to define steady state

Usage

Steady_state_num_sample

Integer Value

Description

Number of samples to define steady state (Default 3).

Steady_state_flag Steady state active flag

Usage

Steady_state_flag

Integer Value

Description

Steady state active flag (Inactive by default).

Steady_state_tolerance Steady state tolerance

Usage

Steady_state_tolerance

Real Value

Description

Steady state tolerance (Default = 0.1%)