{{ result_name }}


{% if table_description %}
{{ table_description }}
{% endif %}

{% if table_format %}

{{ table_format }}

{% endif %}
{{ problem_dropdown }} {{ minimizer_dropdown }} {{ probsize_checkbox }}

Key:

{{ cbar }}

Table:

{{ table }}

Clicking a result in the tables will give more details, such as graphs of the fit against the data and the parameters that the minimizer found.

Clicking the problem names will take you to details of the best minimizer.

Clicking the software name will take you to FitBenchmarking Read the Docs documentation for the selected software.

Hovering over each minimizer name will display its matching algorithm types out of those selected in options.

{% if report_failed_min %}

Failed minimizers

The `algorithm_type` has been selected to be {{algorithm_type}} and not all minimizers specified in the options setup file are compatible. If this was not intended, please review current options setup and re-run FitBenchmarking.

{% for software, minimizer in unselected_minimzers.items() %} {% if minimizer != [] %} Within {{software}}, the following minimizer(s) were not run due to the choice of {{algorithm_type}} for the algorithm type:

    {% for m in minimizer -%} {{ m }}
    {%- endfor %}
{% endif %} {% endfor %}

{% endif %} {% if failed_problems != [] %}

Failed problems

For the following problems all the selected minimizers and software packages raise an exception:

    {% for problem in failed_problems %} {{ problem }}
    {% endfor %}

This could be due to incompatible problem-cost function combinations or the `algorithm_type` set in the options. Please review the logs and current options setup and re-run FitBenchmarking to see comparisons for these problems.

{% endif %}

Errors

The superscripts in the tables denote current error handling within FitBenchmarking. Currently the errors correspond to:

{% for flag, message in error_message.items() %} {% endfor %}
Error flag Error message
{{ flag }} {{ message }}
{% if has_pp %}

For a more interative plot (Dash), switch to "Online version" by clicking the button below. This will only work if the Dash app is running.


{% for offline_plot in pp_filenames %}
{% endfor %}

Fits are taken from all benchmarks, so if FitBenchmarking is run with ``n`` problems and ``m`` cost functions, the resulting profile plots will have ``n*m`` steps on the y-axis.
This is a Dolan-Moré performance profile. The solvers appearing in the top left corner may be considered the best performing on this test set. See Dolan and Moré (2001) for more information.

The plot is such that lines can be removed by clicking on the respective minimizer name in the dropdown or hidden by unselecting the minimizer in the legend. If a solver is removed through the dropdown, the graph gets recalculated based on the best minimizer among the ones shown. On the other hand, if a solver is unselected in the legend, the line gets temporarily hidden and the plot is not recalculated. We recommend only analysing graphs where lines are removed via the dropdown.

{% endif %}