.. _user-guide-errors:

Some common |pyretis| errors
============================

In this section, we describe some of the most common |pyretis|
warning or error messages and how these errors can be resolved.

.. _user-error-simulation:

Simulation warnings/error messages
----------------------------------

.. topic:: Error: ``SyntaxError: invalid syntax``

   This is either a plain bug in the |pyretis| source code or it can
   be caused by using Python 2.x. If the bug is related to something
   like this:

   .. code-block:: python

      class RandomGeneratorBase(metaclass=ABCMeta):
                                       ^

   This error is most likely caused by using Python 2.

   **Fix**: Use Python 3, or report this as a bug.


.. topic:: Error: ``Unknown engine XXX``

   An unknown engine was requested in the input.

   **Fix**: Specify an engine in the |pyretis| input (see the
   :ref:`engine section <user-section-engine>`).


.. topic:: Warning: ``No random seed given. Will just use "0"``

   If a random generator is needed and the seed value has not been
   set, |pyretis| will use a seed equal to 0. This message is just
   to remind the user that not setting a seed is the same as setting
   it to zero.

   **Fix**: Are you ok with this seed? If not, change it in the input
   settings.

.. topic:: Error: ``XXX settings not found: YYY``/``ValueError('Please update settings!')``

   For simulation type ``XXX`` the required settings
   ``YYY`` was not found in the input. Please read the manual
   describing the simulation you wish to perform in order to see
   what simulation settings you will need to set.

   **Fix**: Update your input settings.



.. _user-errors-plotting:

Plotting warnings/error messages
--------------------------------

.. topic:: Warning: ``UserWarning: axes.color_cycle is deprecated and replaced with axes.prop_cycle; please use the latter.``

   |pyretis| supports different versions of matplotlib. In the most recent
   matplotlib versions, the color cycle is set using ``axes.prop_cycle``,
   while ``axes.color_cycle`` was used in previous versions. Currently
   |pyretis| will try to set ``axes.color_cycle`` as the most recent version
   of matplotlib will accept this, albeit with some complaining (i.e. the
   warning message given above).

   **Fix**: N/A. Note that this behaviour might change in the future
   if support for older matplotlib versions is dropped from |pyretis|.


.. topic:: Warning: ``Using Matplotlib version < 1.4.0, please upgrade it!'``

   We recommend using a more recent matplotlib version than 1.4.
   These versions support styles for plotting which makes it easier
   to customize the figures for the reports.

   **Fix**: Upgrade matplotlib.

.. topic:: ``Overriding color cycle``

   When plotting with matplotlib, the selection of colors is typically
   handled by matplotlib and the selected plotting style. When plotting
   a lot of data series in the same plot, we might run out of distinct
   colors. |pyretis| will then create some ad-hoc colors so that the series
   can be distinguished. Note that these colors may not be good enough for
   colorblindness or gray-scale images. Here, we recommend writing your
   own plotting routine or updating the color cycle settings in the
   plot settings file.

   **Fix**: Make a custom plot or modify the |pyretis| matplotlib style.


.. _user-errors-analysis:

Analysis warnings/error messages
--------------------------------

.. topic:: ``Ignored unknown mc move: XXX``

   This warning is given when the analysis encounters a path that was not
   generated by any of the known (to |pyretis|) Monte Carlo moves. These paths
   are **NOT** included when obtaining path lengths for histograms in the
   analysis.

   **Fix**: If this happens for your own modified version of |pyretis|, double check
   the path analysis routines and the method generating the path.

.. topic:: ``The number of paths stored in path ensemble does not correspond to the number of paths seen by the path ensemble! Consider re-running the analysis using the path ensemble file!``

   The analysis can be run on either path ensemble objects or on path
   ensemble files. The path ensemble object stores a finite number of
   paths (or rather representation of paths) which means that the path
   ensemble does not necessarily store (in memory) all the paths -- some
   will be stored in a path ensemble file when the path
   ensemble object is "full". This warning means that this has happened and
   that it will be more correct to run the
   analysis using the *path ensemble file*.

   **Fix**: Run the analysis using the path ensemble file as the source (see
   the section describing the analysis tool).