sierra.core.utils

Miscellaneous bits used in mutiple places but that don’t fit anywhere else.

• dir_create_checked(): Create a directory idempotently.

• path_exists(): Check if a path exists, trying multiple times.

• get_primary_axis(): Determine axis in a bivariate batch criteria is the primary axis.

• exp_range_calc(): Get the range of experiments to run/do stuff with. SUPER USEFUL.

• exp_include_filter(): Calculate which experiments to include in a calculation for something.

• apply_to_expdef(): Apply a generated XML modifictions to an experiment definition.

• pickle_modifications(): After applying XML modifications, pickle changes for later retrieval.

• exp_template_path(): Calculate the path to the template input file in the batch experiment root.

• get_n_robots(): Get the # robots used for a specific Experiment.

• df_fill(): Fill missing cells in a dataframe according to the specified fill policy.

sierra.core.utils.dir_create_checked(path: Union[Path, str], exist_ok: bool) → None

Create a directory idempotently.

If the directory exists and it shouldn’t, raise an error.

sierra.core.utils.path_exists(path: Union[Path, str]) → bool

Check if a path exists, trying multiple times.

This is necessary for working on HPC systems where if a given directory/filesystem is under heavy pressure the first check or two might time out as the FS goes and executes the query over the network.

sierra.core.utils.get_primary_axis(criteria, primary_axis_bc: List, cmdopts: Dict[str, Any]) → int

Determine axis in a bivariate batch criteria is the primary axis.

This is obtained on a per-query basis depending on the query context, or can be overriden on the cmdline.

sierra.core.utils.exp_range_calc(cmdopts: Dict[str, Any], root_dir: Path, criteria) → List[Path]

Get the range of experiments to run/do stuff with. SUPER USEFUL.

sierra.core.utils.exp_include_filter(inc_spec: Optional[str], target: List, n_exps: int)

Calculate which experiments to include in a calculation for something.

Take a input list of experiment numbers to include, and returns the sublist specified by the inc_spec (of the form [x:y]). inc_spec is an absolute specification; if a given performance measure excludes exp0 then that case is handled internally so that array/list shapes work out when generating graphs if this function is used consistently everywhere.

sierra.core.utils.apply_to_expdef(var, exp_def: XMLExpDef) → Tuple[Optional[TagRmList], Optional[TagAddList], Optional[AttrChangeSet]]

Apply a generated XML modifictions to an experiment definition.

In this order:

1. Remove existing XML tags

2. Add new XML tags

3. Change existing XML attributes

sierra.core.utils.pickle_modifications(adds: Optional[TagAddList], chgs: Optional[AttrChangeSet], path: Path) → None

After applying XML modifications, pickle changes for later retrieval.

sierra.core.utils.exp_template_path(cmdopts: Dict[str, Any], batch_input_root: Path, dirname: str) → Path

Calculate the path to the template input file in the batch experiment root.

The file at this path will be Used as the de-facto template for generating per-run input files.

sierra.core.utils.get_n_robots(main_config: Dict[str, Any], cmdopts: Dict[str, Any], exp_input_root: Path, exp_def: XMLExpDef) → int

Get the # robots used for a specific Experiment.

sierra.core.utils.df_fill(df: DataFrame, policy: str) → DataFrame

Fill missing cells in a dataframe according to the specified fill policy.

• ArenaExtent: Representation of a 2D or 3D section/chunk/volume of the arena.

• Sigmoid: Sigmoid activation function.

• ReLu: Rectified Linear Unit (ReLU) activation function.

class sierra.core.utils.ArenaExtent(dims: Vector3D, origin: Vector3D = (0, 0, 0))

Representation of a 2D or 3D section/chunk/volume of the arena.

Inheritance

__dict__ = mappingproxy({'__module__': 'sierra.core.utils', '__doc__': 'Representation of a 2D or 3D section/chunk/volume of the arena.', 'from_corners': <staticmethod object>, '__init__': <function ArenaExtent.__init__>, 'contains': <function ArenaExtent.contains>, 'area': <function ArenaExtent.area>, 'xsize': <function ArenaExtent.xsize>, 'ysize': <function ArenaExtent.ysize>, 'zsize': <function ArenaExtent.zsize>, 'origin': <function ArenaExtent.origin>, '__str__': <function ArenaExtent.__str__>, '__dict__': <attribute '__dict__' of 'ArenaExtent' objects>, '__weakref__': <attribute '__weakref__' of 'ArenaExtent' objects>, '__annotations__': {}})

__doc__ = 'Representation of a 2D or 3D section/chunk/volume of the arena.'

__init__(dims: Vector3D, origin: Vector3D = (0, 0, 0)) → None

__module__ = 'sierra.core.utils'

__str__() → str

Return str(self).

__weakref__

list of weak references to the object (if defined)

area() → float

contains(pt: Vector3D) → bool

static from_corners(ll: Vector3D, ur: Vector3D) → ArenaExtent

Initialize an extent via LL and UR corners.

As opposed to an origin and a set of dimensions.

origin() → Vector3D

xsize() → int

ysize() → int

zsize() → int

class sierra.core.utils.Sigmoid(x: float)

Sigmoid activation function.

(1)\[f(x) =\]

rac{1}{1+e^{-x}}

Inheritance

__call__() → float

Call self as a function.

__dict__ = mappingproxy({'__module__': 'sierra.core.utils', '__doc__': '\n    Sigmoid activation function.\n\n    .. math::\n       f(x) = \x0crac{1}{1+e^{-x}}\n\n    ', '__init__': <function Sigmoid.__init__>, '__call__': <function Sigmoid.__call__>, '__dict__': <attribute '__dict__' of 'Sigmoid' objects>, '__weakref__': <attribute '__weakref__' of 'Sigmoid' objects>, '__annotations__': {}})

__doc__ = '\n    Sigmoid activation function.\n\n    .. math::\n       f(x) = \x0crac{1}{1+e^{-x}}\n\n    '

__init__(x: float) → None

__module__ = 'sierra.core.utils'

__weakref__

list of weak references to the object (if defined)

class sierra.core.utils.ReLu(x: float)

Rectified Linear Unit (ReLU) activation function.

(2)\[\begin{aligned} f(x) = max(0,x) &= x \textit{if} x > 0 &= 0 \textit{else} \end{aligned}\]

Inheritance

__call__()

Call self as a function.

__dict__ = mappingproxy({'__module__': 'sierra.core.utils', '__doc__': '\n    Rectified Linear Unit (ReLU) activation function.\n\n    .. math::\n\n       \\begin{aligned}\n            f(x) = max(0,x) &= x \\textit{if} x > 0\n                            &= 0 \\textit{else}\n       \\end{aligned}\n    ', '__init__': <function ReLu.__init__>, '__call__': <function ReLu.__call__>, '__dict__': <attribute '__dict__' of 'ReLu' objects>, '__weakref__': <attribute '__weakref__' of 'ReLu' objects>, '__annotations__': {}})

__doc__ = '\n    Rectified Linear Unit (ReLU) activation function.\n\n    .. math::\n\n       \\begin{aligned}\n            f(x) = max(0,x) &= x \\textit{if} x > 0\n                            &= 0 \\textit{else}\n       \\end{aligned}\n    '

__init__(x: float)

__module__ = 'sierra.core.utils'

__weakref__

list of weak references to the object (if defined)

• utf8open

sierra.core.utils.utf8open

Explictly specify that the type of file being opened is UTF-8, which is should be for almost everything in SIERRA.

functools.partial(<built-in function open>, encoding='UTF-8')