Configuration#

Each run of unfold requires a configuration file in the TOML format. The configuration file is divided into sections:

paths: provides the input file paths.
output: describes the properties of the output.
inversion: defines inversion parameters, how the inversion operation is defined.
model: defines the parameters for the sklearn ElasticNet that is used.
execution: describes the execution of the inversion, including computer specific properties like the number of threads.

Note

All sections and parameters are expected in the configuration file. There are no optional parameters with defaults.

We provide an example configuration file here.

paths section#

There are five configurables for this section:

overlappogram: path to the overlappogram image to be inverted.
weights: path to the accompanying weights used in the inversion. Weights should be in units of \(\frac{1}{\sigma}\) where \(\sigma\) is the uncertainty or standard deviation. Weights are optional and the keyword can be omitted to run in weightless mode.
mask: path to the accompanying mask used in the inversion. This mask is optional and the keyword can be omitted to run without a mask.
response: path to the instrument response.
gnt: path to the file containing atomic physics values from Chianti, the G(n, t) function.

output section#

There are four configurables for this section:

prefix: the string that output files begin with
make_spectral: if true, makes spectrally pure images as output. otherwise, these files are not made.
directory: path to the directory where output files are written.
overwrite: if true, output files will be overwritten. otherwise, the program will fail writing if a file already exists.

inversion section#

There are six configurables for this section:

solution_fov_width: an integer for the field-of-view width in pixels used in the solution. We suggest 2.
detector_row_range: a list of two integers defining the range of detector rows to invert. For example, [10, 35] would run between 10 and 35.
field_angle_range: a list of two integers defining the range of field angles to use in inversion. Units are arc seconds.
response_dependency_name: for now, only “logt” is supported.
response_dependency_list: a list of floats defining the logarithm of the temperature used in the response dependency.
smooth_over: the method of smoothing, currently only supports “dependence”

model section#

This section defines the parameters used by Scikit-Learn’s ElasticNet. You can find more exhaustive descriptions of the parameters at that link in the sklearn documentation.

There are six configurables for this section:

alphas: A list of floating point numbers defining the various values of alpha to iterate over. Alpha is a constant in sklearn that multiplies a penalty term.
rhos: A list of floating point numbers defining the various values of rho to iterate over. This corresponds to sklearn’s l1_ratio parameter.
warm_start: Boolean indicating whether a warm start should be used when training the model. Note that if the mode (see the execution section) is “row” this has no effect.
tol: The tolerance for the optimization.
max_iter: The maximum number of iterations.
selection: Either set to “cyclic” or “random”. If set to “random”, a random coefficient is updated every iteration rather than looping over features sequentially by default.

execution section#

There are three configurables for this section:

num_threads: The number of threads to use when optimizing.
mode_switch_thread_count: Only used if mode is set to “hybrid”. In that case, when the number of remaining threads is less than this value, the optimization switches from “chunked” to “row”.
mode: The optimization mode can be set to three different values: “row”, “chunked”, or “hybrid”. See Optimization modes for a description of what these do.