fitting_option.rst

Fitting Options

Options that control the benchmarking process are set here.

Software (software)

Software is used to select the fitting software to benchmark, this should be a newline-separated list. Available options are:

• bumps (default software)
• ceres (external software -- see :ref:`external-instructions`)
• CUTEst (external software -- see :ref:`external-instructions`)
• dfo (external software -- see :ref:`extra_dependencies`)
• gofit (external software -- see :ref:`extra_dependencies`)
• gradient_free (external software -- see :ref:`extra_dependencies`)
• gsl (external software -- see :ref:`external-instructions`)
• horace (external software -- see :ref:`external-instructions`)
• levmar (external software -- see :ref:`extra_dependencies`)
• lmfit (external software -- see :ref:`extra_dependencies`)
• mantid (external software -- see :ref:`external-instructions`)
• matlab (external software -- see :ref:`external-instructions`)
• matlab_curve (external software -- see :ref:`external-instructions`)
• matlab_opt (external software -- see :ref:`external-instructions`)
• matlab_stats (external software -- see :ref:`external-instructions`)
• minuit (external software -- see :ref:`extra_dependencies`)
• nlopt (external software -- see :ref:`extra_dependencies`)
• ralfit (external software -- see :ref:`external-instructions`)
• scipy (default software)
• scipy_ls (default software)
• scipy_go
• theseus (external software -- see :ref:`external-instructions`)

Default software options are scipy and scipy_ls

[FITTING]
software: bumps
          dfo
          minuit
          scipy
          scipy_ls
          scipy_go

Warning

Software must be listed to be here to be run. Any minimizers set in :ref:`minimizer_option` will not be run if the software is not also present in this list.

Number of minimizer runs (num_runs)

Sets the number of runs to average each fit over.

Default is 5

[FITTING]
num_runs: 5

Algorithm type (algorithm_type)

This is used to select what type of algorithm is used within a specific software. For a full list of available minimizers for each algorithm type, see :ref:`minimizer_types`. The options are:

• all - all minimizers
• ls - least-squares fitting algorithms
• deriv_free - derivative free algorithms (these are algorithms that cannot use information about derivatives -- e.g., the Simplex method in Mantid), see :ref:`deriv_free`.
• general - minimizers which solve a generic min f(x)
• simplex - derivative free simplex based algorithms e.g. Nelder-Mead, see :ref:`Simplex <simplex>`
• trust_region - algorithms which employ a trust region approach, see :ref:`trust_region`
• levenberg-marquardt - minimizers that use the Levenberg Marquardt algorithm, see :ref:`Levenberg-Marquardt <levenberg_marquardt>`.
• gauss_newton - minimizers that use the Gauss Newton algorithm, see :ref:`Gauss-Newton <gauss_newton>`
• bfgs - minimizers that use the BFGS algorithm, see :ref:`BFGS <bfgs>`
• conjugate_gradient - Conjugate Gradient algorithms, see :ref:`Conjugate Gradient <conjugate_gradient>`
• steepest_descent - Steepest Descent algorithms, see :ref:`Steepest Descent <steepest_descent>`
• global_optimization - Global Optimization algorithms

Default is all

[FITTING]
algorithm_type: all

Warning

Choosing an option other than all may deselect certain minimizers set in the options file

Jacobian method (jac_method)

This sets the Jacobian used. Choosing multiple options via a new line seperated list will result in all combinations being benchmarked. Current Jacobian methods are:

• analytic - uses the analytic Jacobian extracted from the fitting problem.
• scipy - uses :ref:`SciPy's finite difference Jacobian approximations <scipy-jac>`.
• default - uses the default derivative approximation implemented in the minimizer.
• numdifftools - uses the python package :ref:`numdifftools <numdifftools-jac>`.

Default is default

[FITTING]
jac_method: scipy

Warning

Currently analytic Jacobians are only available for problems that use the cutest and NIST parsers.

Hessian method (hes_method)

This sets the Hessian used. Choosing multiple options via a new line seperated list will result in all combinations being benchmarked. Current Hessian methods are:

• default - Hessian information is not passed to minimizers
• analytic - uses the analytic Hessian extracted from the fitting problem.
• scipy - uses :ref:`SciPy's finite difference approximations <scipy-hes>`.
• numdifftools - uses the python package :ref:`numdifftools <numdifftools-hes>`.

Default is default

[FITTING]
hes_method: default

Warning

Currently analytic Hessians are only available for problems that use the cutest and NIST parsers.

Cost function (cost_func_type)

This sets the cost functions to be used for the given data. Choosing multiple options via a new line seperated list will result in all combinations being benchmarked. Currently supported cost functions are:

• nlls - This sets the cost function to be non-weighted non-linear least squares, :class:`~fitbenchmarking.cost_func.nlls_cost_func.NLLSCostFunc`.
• weighted_nlls - This sets the cost function to be weighted non-linear least squares, :class:`~fitbenchmarking.cost_func.weighted_nlls_cost_func.WeightedNLLSCostFunc`.
• hellinger_nlls - This sets the cost function to be the Hellinger cost function, :class:`~fitbenchmarking.cost_func.hellinger_nlls_cost_func.HellingerNLLSCostFunc`.
• poisson - This sets the cost function to be the Poisson Deviation cost function, :class:`~fitbenchmarking.cost_func.poisson_cost_func.PoissonCostFunc`.

Default is weighted_nlls

[FITTING]
cost_func_type: weighted_nlls

Maximum Runtime (max_runtime)

This sets the maximum runtime a minimizer has to solve one benchmark problem num_runs number of times, where num_runs is another option a user can set. If the minimizer is still running after the maximum time has elapsed, then this result will be skipped and FitBenchmarking will move on to the next minimizer / benchmark dataset combination. The main purpose of this option is to get to result tables quicker by limit the runtime.

max_runtime is set by specifying a number in unit of seconds. Please note that depending on platform the time specified with max_runtime may not match entirely with the absolute run-times specified in tables. Hence you may have to experiment a bit with this option to get the cutoff you want.

Default is 600 seconds

[FITTING]
max_runtime: 600