zfit package

Top-level package for zfit.

class zfit.Parameter(name, value, lower_limit=None, upper_limit=None, step_size=None, floating=True, dtype=tf.float64, **kwargs)[source]

Bases: zfit.core.parameter.ZfitParameterMixin, zfit.core.parameter.TFBaseVariable, zfit.core.parameter.BaseParameter, zfit.core.interfaces.ZfitIndependentParameter

Class for fit parameters, derived from TF Variable class.

Parameters
DEFAULT_STEP_SIZE = 0.001
property lower
property upper
property has_limits

If the parameter has limits set or not.

Return type

bool

property at_limit

If the value is at the limit (or over it).

The precision is up to 1e-5 relative.

Return type

Tensor

Returns

Boolean tf.Tensor that tells whether the value is at the limits.

value()[source]
read_value()[source]
property floating
property independent
property step_size

Step size of the parameter, the estimated order of magnitude of the uncertainty.

This can be crucial to tune for the minimization. A too large step_size can produce NaNs, a too small won’t converge.

If the step size is not set, the DEFAULT_STEP_SIZE is used.

Return type

Tensor

Returns

The step size

set_value(value)[source]

Set the Parameter to value (temporarily if used in a context manager).

This operation won’t, compared to the assign, return the read value but an object that can act as a context manager.

Parameters

value (Union[int, float, complex, Tensor, ForwardRef]) – The value the parameter will take on.

randomize(minval=None, maxval=None, sampler=<built-in method uniform of numpy.random.mtrand.RandomState object>)[source]

Update the parameter with a randomised value between minval and maxval and return it.

Parameters
Return type

Tensor

Returns

The sampled value

get_params(floating=True, is_yield=None, extract_independent=True, only_floating=<class 'zfit.util.checks.NotSpecified'>)[source]
Return type

Set[ZfitParameter]

property lower_limit
property upper_limit
class SaveSliceInfo(full_name=None, full_shape=None, var_offset=None, var_shape=None, save_slice_info_def=None, import_scope=None)

Bases: object

Information on how to save this Variable as a slice.

Provides internal support for saving variables as slices of a larger variable. This API is not public and is subject to change.

Available properties:

  • full_name

  • full_shape

  • var_offset

  • var_shape

Create a SaveSliceInfo.

Parameters
  • full_name – Name of the full variable of which this Variable is a slice.

  • full_shape – Shape of the full variable, as a list of int.

  • var_offset – Offset of this Variable into the full variable, as a list of int.

  • var_shape – Shape of this Variable, as a list of int.

  • save_slice_info_defSaveSliceInfoDef protocol buffer. If not None, recreates the SaveSliceInfo object its contents. save_slice_info_def and other arguments are mutually exclusive.

  • import_scope – Optional string. Name scope to add. Only used when initializing from protocol buffer.

property spec

Computes the spec string used for saving.

to_proto(export_scope=None)

Returns a SaveSliceInfoDef() proto.

Parameters

export_scope – Optional string. Name scope to remove.

Returns

A SaveSliceInfoDef protocol buffer, or None if the Variable is not in the specified name scope.

__iter__()

Dummy method to prevent iteration.

Do not call.

NOTE(mrry): If we register __getitem__ as an overloaded operator, Python will valiantly attempt to iterate over the variable’s Tensor from 0 to infinity. Declaring this method prevents this unintended behavior.

Raises

TypeError – when invoked.

__ne__(other)

Compares two variables element-wise for equality.

add_cache_deps(cache_deps, allow_non_cachable=True)

Add dependencies that render the cache invalid if they change.

Parameters
  • cache_deps (Union[ForwardRef, Iterable[ForwardRef]]) –

  • allow_non_cachable (bool) – If True, allow cache_dependents to be non-cachables. If False, any cache_dependents that is not a ZfitCachable will raise an error.

Raises

TypeError – if one of the cache_dependents is not a ZfitCachable _and_ allow_non_cachable if False.

property aggregation
assign(value, use_locking=None, name=None, read_value=True)

Assigns a new value to this variable.

Parameters
  • value – A Tensor. The new value for this variable.

  • use_locking – If True, use locking during the assignment.

  • name – The name to use for the assignment.

  • read_value – A bool. Whether to read and return the new value of the variable or not.

Returns

If read_value is True, this method will return the new value of the variable after the assignment has completed. Otherwise, when in graph mode it will return the Operation that does the assignment, and when in eager mode it will return None.

assign_add(delta, use_locking=None, name=None, read_value=True)

Adds a value to this variable.

Parameters
  • delta – A Tensor. The value to add to this variable.

  • use_locking – If True, use locking during the operation.

  • name – The name to use for the operation.

  • read_value – A bool. Whether to read and return the new value of the variable or not.

Returns

If read_value is True, this method will return the new value of the variable after the assignment has completed. Otherwise, when in graph mode it will return the Operation that does the assignment, and when in eager mode it will return None.

assign_sub(delta, use_locking=None, name=None, read_value=True)

Subtracts a value from this variable.

Parameters
  • delta – A Tensor. The value to subtract from this variable.

  • use_locking – If True, use locking during the operation.

  • name – The name to use for the operation.

  • read_value – A bool. Whether to read and return the new value of the variable or not.

Returns

If read_value is True, this method will return the new value of the variable after the assignment has completed. Otherwise, when in graph mode it will return the Operation that does the assignment, and when in eager mode it will return None.

batch_scatter_update(sparse_delta, use_locking=False, name=None)

Assigns tf.IndexedSlices to this variable batch-wise.

Analogous to batch_gather. This assumes that this variable and the sparse_delta IndexedSlices have a series of leading dimensions that are the same for all of them, and the updates are performed on the last dimension of indices. In other words, the dimensions should be the following:

num_prefix_dims = sparse_delta.indices.ndims - 1 batch_dim = num_prefix_dims + 1 `sparse_delta.updates.shape = sparse_delta.indices.shape + var.shape[

batch_dim:]`

where

sparse_delta.updates.shape[:num_prefix_dims] == sparse_delta.indices.shape[:num_prefix_dims] == var.shape[:num_prefix_dims]

And the operation performed can be expressed as:

`var[i_1, …, i_n,
sparse_delta.indices[i_1, …, i_n, j]] = sparse_delta.updates[

i_1, …, i_n, j]`

When sparse_delta.indices is a 1D tensor, this operation is equivalent to scatter_update.

To avoid this operation one can looping over the first ndims of the variable and using scatter_update on the subtensors that result of slicing the first dimension. This is a valid option for ndims = 1, but less efficient than this implementation.

Parameters
  • sparse_deltatf.IndexedSlices to be assigned to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

property constraint

Returns the constraint function associated with this variable.

Returns

The constraint function that was passed to the variable constructor. Can be None if no constraint was passed.

copy(deep=False, name=None, **overwrite_params)
Return type

ZfitObject

count_up_to(limit)

Increments this variable until it reaches limit. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Prefer Dataset.range instead.

When that Op is run it tries to increment the variable by 1. If incrementing the variable would bring it above limit then the Op raises the exception OutOfRangeError.

If no error is raised, the Op outputs the value of the variable before the increment.

This is essentially a shortcut for count_up_to(self, limit).

Parameters

limit – value at which incrementing the variable raises an error.

Returns

A Tensor that will hold the variable value before the increment. If no other Op modifies this variable, the values produced will all be distinct.

property create

The op responsible for initializing this variable.

property device

The device this variable is on.

property dtype

The dtype of the object

Return type

DType

eval(session=None)

Evaluates and returns the value of this variable.

experimental_ref()

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use ref() instead.

static from_proto(variable_def, import_scope=None)
gather_nd(indices, name=None)

Reads the value of this variable sparsely, using gather_nd.

get_cache_deps(only_floating=True)

Return a set of all independent Parameter that this object depends on.

Parameters

only_floating (bool) – If True, only return floating Parameter

Return type

OrderedSet

get_dependencies(only_floating=True)

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use get_params instead if you want to retrieve the independent parameters or get_cache_deps in case you need the numerical cache dependents (advanced).

Return type

OrderedSet

get_shape()

Alias of Variable.shape.

property graph

The Graph of this variable.

graph_caching_methods = []
property handle

The handle by which this variable can be accessed.

property initial_value

Returns the Tensor used as the initial value for the variable.

initialized_value()

Returns the value of the initialized variable. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.

You should use this instead of the variable itself to initialize another variable with a value that depends on the value of this variable.

`python # Initialize 'v' with a random tensor. v = tf.Variable(tf.random.truncated_normal([10, 40])) # Use `initialized_value` to guarantee that `v` has been # initialized before its value is used to initialize `w`. # The random values are picked only once. w = tf.Variable(v.initialized_value() * 2.0) `

Returns

A Tensor holding the value of this variable after its initializer has run.

property initializer

The op responsible for initializing this variable.

instances = <_weakrefset.WeakSet object>
is_initialized(name=None)

Checks whether a resource variable has been initialized.

Outputs boolean scalar indicating whether the tensor has been initialized.

Parameters

name – A name for the operation (optional).

Returns

A Tensor of type bool.

load(value, session=None)

Load new value into this variable. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Prefer Variable.assign which has equivalent behavior in 2.X.

Writes new value to variable’s memory. Doesn’t add ops to the graph.

This convenience method requires a session where the graph containing this variable has been launched. If no session is passed, the default session is used. See tf.compat.v1.Session for more information on launching a graph and on sessions.

```python v = tf.Variable([1, 2]) init = tf.compat.v1.global_variables_initializer()

with tf.compat.v1.Session() as sess:

sess.run(init) # Usage passing the session explicitly. v.load([2, 3], sess) print(v.eval(sess)) # prints [2 3] # Usage with the default session. The ‘with’ block # above makes ‘sess’ the default session. v.load([3, 4], sess) print(v.eval()) # prints [3 4]

```

Parameters
  • value – New variable value

  • session – The session to use to evaluate this variable. If none, the default session is used.

Raises

ValueError – Session is not passed and no default session

property name
Return type

str

numpy()
property op

The op for this variable.

property params
Return type

~ParametersType

ref()

Returns a hashable reference object to this Variable.

The primary use case for this API is to put variables in a set/dictionary. We can’t put variables in a set/dictionary as variable.__hash__() is no longer available starting Tensorflow 2.0.

The following will raise an exception starting 2.0

>>> x = tf.Variable(5)
>>> y = tf.Variable(10)
>>> z = tf.Variable(10)
>>> variable_set = {x, y, z}
Traceback (most recent call last):
  ...
TypeError: Variable is unhashable. Instead, use tensor.ref() as the key.
>>> variable_dict = {x: 'five', y: 'ten'}
Traceback (most recent call last):
  ...
TypeError: Variable is unhashable. Instead, use tensor.ref() as the key.

Instead, we can use variable.ref().

>>> variable_set = {x.ref(), y.ref(), z.ref()}
>>> x.ref() in variable_set
True
>>> variable_dict = {x.ref(): 'five', y.ref(): 'ten', z.ref(): 'ten'}
>>> variable_dict[y.ref()]
'ten'

Also, the reference object provides .deref() function that returns the original Variable.

>>> x = tf.Variable(5)
>>> x.ref().deref()
<tf.Variable 'Variable:0' shape=() dtype=int32, numpy=5>
register_cacher(cacher)

Register a cacher that caches values produces by this instance; a dependent.

Parameters

cacher (Union[ForwardRef, Iterable[ForwardRef]]) –

reset_cache(reseter)
reset_cache_self()

Clear the cache of self and all dependent cachers.

scatter_add(sparse_delta, use_locking=False, name=None)

Adds tf.IndexedSlices to this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be added to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_div(sparse_delta, use_locking=False, name=None)

Divide this variable by tf.IndexedSlices.

Parameters
  • sparse_deltatf.IndexedSlices to divide this variable by.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_max(sparse_delta, use_locking=False, name=None)

Updates this variable with the max of tf.IndexedSlices and itself.

Parameters
  • sparse_deltatf.IndexedSlices to use as an argument of max with this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_min(sparse_delta, use_locking=False, name=None)

Updates this variable with the min of tf.IndexedSlices and itself.

Parameters
  • sparse_deltatf.IndexedSlices to use as an argument of min with this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_mul(sparse_delta, use_locking=False, name=None)

Multiply this variable by tf.IndexedSlices.

Parameters
  • sparse_deltatf.IndexedSlices to multiply this variable by.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_nd_add(indices, updates, name=None)

Applies sparse addition to individual values or slices in a Variable.

ref is a Tensor with rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into ref. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the K`th dimension of `ref.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) add = ref.scatter_nd_add(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(add)

```

The resulting update to ref would look like this:

[1, 13, 3, 14, 14, 6, 7, 20]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_max(indices, updates, name=None)

Updates this variable with the max of tf.IndexedSlices and itself.

ref is a Tensor with rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into ref. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the K`th dimension of `ref.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]. `

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_min(indices, updates, name=None)

Updates this variable with the min of tf.IndexedSlices and itself.

ref is a Tensor with rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into ref. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the K`th dimension of `ref.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]. `

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_sub(indices, updates, name=None)

Applies sparse subtraction to individual values or slices in a Variable.

ref is a Tensor with rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into ref. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the K`th dimension of `ref.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) op = ref.scatter_nd_sub(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(op)

```

The resulting update to ref would look like this:

[1, -9, 3, -6, -6, 6, 7, -4]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_update(indices, updates, name=None)

Applies sparse assignment to individual values or slices in a Variable.

ref is a Tensor with rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into ref. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the K`th dimension of `ref.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) op = ref.scatter_nd_update(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(op)

```

The resulting update to ref would look like this:

[1, 11, 3, 10, 9, 6, 7, 12]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_sub(sparse_delta, use_locking=False, name=None)

Subtracts tf.IndexedSlices from this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be subtracted from this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_update(sparse_delta, use_locking=False, name=None)

Assigns tf.IndexedSlices to this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be assigned to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

set_shape(shape)
property shape

The shape of this variable.

sparse_read(indices, name=None)

Reads the value of this variable sparsely, using gather.

property synchronization
to_proto(export_scope=None)

Converts a ResourceVariable to a VariableDef protocol buffer.

Parameters

export_scope – Optional string. Name scope to remove.

Raises

RuntimeError – If run in EAGER mode.

Returns

A VariableDef protocol buffer, or None if the Variable is not in the specified name scope.

property trainable
class zfit.ComposedParameter(name, value_fn, params=<class 'zfit.util.checks.NotSpecified'>, dtype=tf.float64, dependents=<class 'zfit.util.checks.NotSpecified'>)[source]

Bases: zfit.core.parameter.BaseComposedParameter

Arbitrary composition of parameters.

A ComposedParameter allows for arbitrary combinations of parameters and correlations

Parameters
class SaveSliceInfo(full_name=None, full_shape=None, var_offset=None, var_shape=None, save_slice_info_def=None, import_scope=None)

Bases: object

Information on how to save this Variable as a slice.

Provides internal support for saving variables as slices of a larger variable. This API is not public and is subject to change.

Available properties:

  • full_name

  • full_shape

  • var_offset

  • var_shape

Create a SaveSliceInfo.

Parameters
  • full_name – Name of the full variable of which this Variable is a slice.

  • full_shape – Shape of the full variable, as a list of int.

  • var_offset – Offset of this Variable into the full variable, as a list of int.

  • var_shape – Shape of this Variable, as a list of int.

  • save_slice_info_defSaveSliceInfoDef protocol buffer. If not None, recreates the SaveSliceInfo object its contents. save_slice_info_def and other arguments are mutually exclusive.

  • import_scope – Optional string. Name scope to add. Only used when initializing from protocol buffer.

property spec

Computes the spec string used for saving.

to_proto(export_scope=None)

Returns a SaveSliceInfoDef() proto.

Parameters

export_scope – Optional string. Name scope to remove.

Returns

A SaveSliceInfoDef protocol buffer, or None if the Variable is not in the specified name scope.

__iter__()

Dummy method to prevent iteration.

Do not call.

NOTE(mrry): If we register __getitem__ as an overloaded operator, Python will valiantly attempt to iterate over the variable’s Tensor from 0 to infinity. Declaring this method prevents this unintended behavior.

Raises

TypeError – when invoked.

__ne__(other)

Compares two variables element-wise for equality.

add_cache_deps(cache_deps, allow_non_cachable=True)

Add dependencies that render the cache invalid if they change.

Parameters
  • cache_deps (Union[ForwardRef, Iterable[ForwardRef]]) –

  • allow_non_cachable (bool) – If True, allow cache_dependents to be non-cachables. If False, any cache_dependents that is not a ZfitCachable will raise an error.

Raises

TypeError – if one of the cache_dependents is not a ZfitCachable _and_ allow_non_cachable if False.

property aggregation
assign(value, use_locking=False, name=None, read_value=True)

Assigns a new value to the variable.

This is essentially a shortcut for assign(self, value).

Parameters
  • value – A Tensor. The new value for this variable.

  • use_locking – If True, use locking during the assignment.

  • name – The name of the operation to be created

  • read_value – if True, will return something which evaluates to the new value of the variable; if False will return the assign op.

Returns

The updated variable. If read_value is false, instead returns None in Eager mode and the assign op in graph mode.

assign_add(delta, use_locking=False, name=None, read_value=True)

Adds a value to this variable.

This is essentially a shortcut for assign_add(self, delta).

Parameters
  • delta – A Tensor. The value to add to this variable.

  • use_locking – If True, use locking during the operation.

  • name – The name of the operation to be created

  • read_value – if True, will return something which evaluates to the new value of the variable; if False will return the assign op.

Returns

The updated variable. If read_value is false, instead returns None in Eager mode and the assign op in graph mode.

assign_sub(delta, use_locking=False, name=None, read_value=True)

Subtracts a value from this variable.

This is essentially a shortcut for assign_sub(self, delta).

Parameters
  • delta – A Tensor. The value to subtract from this variable.

  • use_locking – If True, use locking during the operation.

  • name – The name of the operation to be created

  • read_value – if True, will return something which evaluates to the new value of the variable; if False will return the assign op.

Returns

The updated variable. If read_value is false, instead returns None in Eager mode and the assign op in graph mode.

batch_scatter_update(sparse_delta, use_locking=False, name=None)

Assigns tf.IndexedSlices to this variable batch-wise.

Analogous to batch_gather. This assumes that this variable and the sparse_delta IndexedSlices have a series of leading dimensions that are the same for all of them, and the updates are performed on the last dimension of indices. In other words, the dimensions should be the following:

num_prefix_dims = sparse_delta.indices.ndims - 1 batch_dim = num_prefix_dims + 1 `sparse_delta.updates.shape = sparse_delta.indices.shape + var.shape[

batch_dim:]`

where

sparse_delta.updates.shape[:num_prefix_dims] == sparse_delta.indices.shape[:num_prefix_dims] == var.shape[:num_prefix_dims]

And the operation performed can be expressed as:

`var[i_1, …, i_n,
sparse_delta.indices[i_1, …, i_n, j]] = sparse_delta.updates[

i_1, …, i_n, j]`

When sparse_delta.indices is a 1D tensor, this operation is equivalent to scatter_update.

To avoid this operation one can looping over the first ndims of the variable and using scatter_update on the subtensors that result of slicing the first dimension. This is a valid option for ndims = 1, but less efficient than this implementation.

Parameters
  • sparse_deltatf.IndexedSlices to be assigned to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

property constraint

Returns the constraint function associated with this variable.

Returns

The constraint function that was passed to the variable constructor. Can be None if no constraint was passed.

copy(deep=False, name=None, **overwrite_params)
Return type

ZfitObject

count_up_to(limit)

Increments this variable until it reaches limit. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Prefer Dataset.range instead.

When that Op is run it tries to increment the variable by 1. If incrementing the variable would bring it above limit then the Op raises the exception OutOfRangeError.

If no error is raised, the Op outputs the value of the variable before the increment.

This is essentially a shortcut for count_up_to(self, limit).

Parameters

limit – value at which incrementing the variable raises an error.

Returns

A Tensor that will hold the variable value before the increment. If no other Op modifies this variable, the values produced will all be distinct.

property device

The device of this variable.

property dtype

The dtype of the object

Return type

DType

eval(session=None)

In a session, computes and returns the value of this variable.

This is not a graph construction method, it does not add ops to the graph.

This convenience method requires a session where the graph containing this variable has been launched. If no session is passed, the default session is used. See tf.compat.v1.Session for more information on launching a graph and on sessions.

```python v = tf.Variable([1, 2]) init = tf.compat.v1.global_variables_initializer()

with tf.compat.v1.Session() as sess:

sess.run(init) # Usage passing the session explicitly. print(v.eval(sess)) # Usage with the default session. The ‘with’ block # above makes ‘sess’ the default session. print(v.eval())

```

Parameters

session – The session to use to evaluate this variable. If none, the default session is used.

Returns

A numpy ndarray with a copy of the value of this variable.

experimental_ref()

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use ref() instead.

property floating
static from_proto(variable_def, import_scope=None)

Returns a Variable object created from variable_def.

gather_nd(indices, name=None)

Gather slices from params into a Tensor with shape specified by indices.

See tf.gather_nd for details.

Parameters
  • indices – A Tensor. Must be one of the following types: int32, int64. Index tensor.

  • name – A name for the operation (optional).

Returns

A Tensor. Has the same type as params.

get_cache_deps(only_floating=True)

Return a set of all independent Parameter that this object depends on.

Parameters

only_floating (bool) – If True, only return floating Parameter

Return type

OrderedSet

get_dependencies(only_floating=True)

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use get_params instead if you want to retrieve the independent parameters or get_cache_deps in case you need the numerical cache dependents (advanced).

Return type

OrderedSet

get_params(floating=True, is_yield=None, extract_independent=True, only_floating=<class 'zfit.util.checks.NotSpecified'>)

Recursively collect parameters that this object depends on according to the filter criteria.

Which parameters should be included can be steered using the arguments as a filter.
  • None: do not filter on this. E.g. floating=None will return parameters that are floating as well as

    parameters that are fixed.

  • True: only return parameters that fulfil this criterion

  • False: only return parameters that do not fulfil this criterion. E.g. floating=False will return

    only parameters that are not floating.

Parameters
  • floating (Optional[bool]) – if a parameter is floating, e.g. if floating() returns True

  • is_yield (Optional[bool]) – if a parameter is a yield of the _current_ model. This won’t be applied recursively, but may include yields if they do also represent a parameter parametrizing the shape. So if the yield of the current model depends on other yields (or also non-yields), this will be included. If, however, just submodels depend on a yield (as their yield) and it is not correlated to the output of our model, they won’t be included.

  • extract_independent (Optional[bool]) – If the parameter is an independent parameter, i.e. if it is a ZfitIndependentParameter.

Return type

Set[ZfitParameter]

get_shape()

Alias of Variable.shape.

property graph

The Graph of this variable.

graph_caching_methods = []
property independent
property initial_value

Returns the Tensor used as the initial value for the variable.

Note that this is different from initialized_value() which runs the op that initializes the variable before returning its value. This method returns the tensor that is used by the op that initializes the variable.

Returns

A Tensor.

initialized_value()

Returns the value of the initialized variable. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.

You should use this instead of the variable itself to initialize another variable with a value that depends on the value of this variable.

`python # Initialize 'v' with a random tensor. v = tf.Variable(tf.random.truncated_normal([10, 40])) # Use `initialized_value` to guarantee that `v` has been # initialized before its value is used to initialize `w`. # The random values are picked only once. w = tf.Variable(v.initialized_value() * 2.0) `

Returns

A Tensor holding the value of this variable after its initializer has run.

property initializer

The initializer operation for this variable.

instances = <_weakrefset.WeakSet object>
load(value, session=None)

Load new value into this variable. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Prefer Variable.assign which has equivalent behavior in 2.X.

Writes new value to variable’s memory. Doesn’t add ops to the graph.

This convenience method requires a session where the graph containing this variable has been launched. If no session is passed, the default session is used. See tf.compat.v1.Session for more information on launching a graph and on sessions.

```python v = tf.Variable([1, 2]) init = tf.compat.v1.global_variables_initializer()

with tf.compat.v1.Session() as sess:

sess.run(init) # Usage passing the session explicitly. v.load([2, 3], sess) print(v.eval(sess)) # prints [2 3] # Usage with the default session. The ‘with’ block # above makes ‘sess’ the default session. v.load([3, 4], sess) print(v.eval()) # prints [3 4]

```

Parameters
  • value – New variable value

  • session – The session to use to evaluate this variable. If none, the default session is used.

Raises

ValueError – Session is not passed and no default session

property name
Return type

str

numpy()
property op

The Operation of this variable.

property params
read_value()
ref()

Returns a hashable reference object to this Variable.

The primary use case for this API is to put variables in a set/dictionary. We can’t put variables in a set/dictionary as variable.__hash__() is no longer available starting Tensorflow 2.0.

The following will raise an exception starting 2.0

>>> x = tf.Variable(5)
>>> y = tf.Variable(10)
>>> z = tf.Variable(10)
>>> variable_set = {x, y, z}
Traceback (most recent call last):
  ...
TypeError: Variable is unhashable. Instead, use tensor.ref() as the key.
>>> variable_dict = {x: 'five', y: 'ten'}
Traceback (most recent call last):
  ...
TypeError: Variable is unhashable. Instead, use tensor.ref() as the key.

Instead, we can use variable.ref().

>>> variable_set = {x.ref(), y.ref(), z.ref()}
>>> x.ref() in variable_set
True
>>> variable_dict = {x.ref(): 'five', y.ref(): 'ten', z.ref(): 'ten'}
>>> variable_dict[y.ref()]
'ten'

Also, the reference object provides .deref() function that returns the original Variable.

>>> x = tf.Variable(5)
>>> x.ref().deref()
<tf.Variable 'Variable:0' shape=() dtype=int32, numpy=5>
register_cacher(cacher)

Register a cacher that caches values produces by this instance; a dependent.

Parameters

cacher (Union[ForwardRef, Iterable[ForwardRef]]) –

reset_cache(reseter)
reset_cache_self()

Clear the cache of self and all dependent cachers.

scatter_add(sparse_delta, use_locking=False, name=None)

Adds tf.IndexedSlices to this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be added to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_div(sparse_delta, use_locking=False, name=None)

Divide this variable by tf.IndexedSlices.

Parameters
  • sparse_deltatf.IndexedSlices to divide this variable by.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_max(sparse_delta, use_locking=False, name=None)

Updates this variable with the max of tf.IndexedSlices and itself.

Parameters
  • sparse_deltatf.IndexedSlices to use as an argument of max with this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_min(sparse_delta, use_locking=False, name=None)

Updates this variable with the min of tf.IndexedSlices and itself.

Parameters
  • sparse_deltatf.IndexedSlices to use as an argument of min with this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_mul(sparse_delta, use_locking=False, name=None)

Multiply this variable by tf.IndexedSlices.

Parameters
  • sparse_deltatf.IndexedSlices to multiply this variable by.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_nd_add(indices, updates, name=None)

Applies sparse addition to individual values or slices in a Variable.

The Variable has rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into self. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the `K`th dimension of self.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, self.shape[K], ..., self.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

v = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) add = v.scatter_nd_add(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(add)

```

The resulting update to v would look like this:

[1, 13, 3, 14, 14, 6, 7, 20]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_sub(indices, updates, name=None)

Applies sparse subtraction to individual values or slices in a Variable.

Assuming the variable has rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into self. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the `K`th dimension of self.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, self.shape[K], ..., self.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

v = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) op = v.scatter_nd_sub(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(op)

```

The resulting update to v would look like this:

[1, -9, 3, -6, -6, 6, 7, -4]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_update(indices, updates, name=None)

Applies sparse assignment to individual values or slices in a Variable.

The Variable has rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into self. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the `K`th dimension of self.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, self.shape[K], ..., self.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

v = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) op = v.scatter_nd_assign(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(op)

```

The resulting update to v would look like this:

[1, 11, 3, 10, 9, 6, 7, 12]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_sub(sparse_delta, use_locking=False, name=None)

Subtracts tf.IndexedSlices from this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be subtracted from this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_update(sparse_delta, use_locking=False, name=None)

Assigns tf.IndexedSlices to this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be assigned to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

set_shape(shape)

Overrides the shape for this variable.

Parameters

shape – the TensorShape representing the overridden shape.

property shape
sparse_read(indices, name=None)

Gather slices from params axis axis according to indices.

This function supports a subset of tf.gather, see tf.gather for details on usage.

Parameters
  • indices – The index Tensor. Must be one of the following types: int32, int64. Must be in range [0, params.shape[axis]).

  • name – A name for the operation (optional).

Returns

A Tensor. Has the same type as params.

property synchronization
to_proto(export_scope=None)

Converts a Variable to a VariableDef protocol buffer.

Parameters

export_scope – Optional string. Name scope to remove.

Returns

A VariableDef protocol buffer, or None if the Variable is not in the specified name scope.

property trainable
value()
class zfit.ComplexParameter(*args, **kwargs)[source]

Bases: zfit.core.parameter.ComposedParameter

Create a complex parameter.

Note

Use the constructor class methods instead of the __init__() constructor:

class SaveSliceInfo(full_name=None, full_shape=None, var_offset=None, var_shape=None, save_slice_info_def=None, import_scope=None)

Bases: object

Information on how to save this Variable as a slice.

Provides internal support for saving variables as slices of a larger variable. This API is not public and is subject to change.

Available properties:

  • full_name

  • full_shape

  • var_offset

  • var_shape

Create a SaveSliceInfo.

Parameters
  • full_name – Name of the full variable of which this Variable is a slice.

  • full_shape – Shape of the full variable, as a list of int.

  • var_offset – Offset of this Variable into the full variable, as a list of int.

  • var_shape – Shape of this Variable, as a list of int.

  • save_slice_info_defSaveSliceInfoDef protocol buffer. If not None, recreates the SaveSliceInfo object its contents. save_slice_info_def and other arguments are mutually exclusive.

  • import_scope – Optional string. Name scope to add. Only used when initializing from protocol buffer.

property spec

Computes the spec string used for saving.

to_proto(export_scope=None)

Returns a SaveSliceInfoDef() proto.

Parameters

export_scope – Optional string. Name scope to remove.

Returns

A SaveSliceInfoDef protocol buffer, or None if the Variable is not in the specified name scope.

__iter__()

Dummy method to prevent iteration.

Do not call.

NOTE(mrry): If we register __getitem__ as an overloaded operator, Python will valiantly attempt to iterate over the variable’s Tensor from 0 to infinity. Declaring this method prevents this unintended behavior.

Raises

TypeError – when invoked.

__ne__(other)

Compares two variables element-wise for equality.

add_cache_deps(cache_deps, allow_non_cachable=True)

Add dependencies that render the cache invalid if they change.

Parameters
  • cache_deps (Union[ForwardRef, Iterable[ForwardRef]]) –

  • allow_non_cachable (bool) – If True, allow cache_dependents to be non-cachables. If False, any cache_dependents that is not a ZfitCachable will raise an error.

Raises

TypeError – if one of the cache_dependents is not a ZfitCachable _and_ allow_non_cachable if False.

property aggregation
assign(value, use_locking=False, name=None, read_value=True)

Assigns a new value to the variable.

This is essentially a shortcut for assign(self, value).

Parameters
  • value – A Tensor. The new value for this variable.

  • use_locking – If True, use locking during the assignment.

  • name – The name of the operation to be created

  • read_value – if True, will return something which evaluates to the new value of the variable; if False will return the assign op.

Returns

The updated variable. If read_value is false, instead returns None in Eager mode and the assign op in graph mode.

assign_add(delta, use_locking=False, name=None, read_value=True)

Adds a value to this variable.

This is essentially a shortcut for assign_add(self, delta).

Parameters
  • delta – A Tensor. The value to add to this variable.

  • use_locking – If True, use locking during the operation.

  • name – The name of the operation to be created

  • read_value – if True, will return something which evaluates to the new value of the variable; if False will return the assign op.

Returns

The updated variable. If read_value is false, instead returns None in Eager mode and the assign op in graph mode.

assign_sub(delta, use_locking=False, name=None, read_value=True)

Subtracts a value from this variable.

This is essentially a shortcut for assign_sub(self, delta).

Parameters
  • delta – A Tensor. The value to subtract from this variable.

  • use_locking – If True, use locking during the operation.

  • name – The name of the operation to be created

  • read_value – if True, will return something which evaluates to the new value of the variable; if False will return the assign op.

Returns

The updated variable. If read_value is false, instead returns None in Eager mode and the assign op in graph mode.

batch_scatter_update(sparse_delta, use_locking=False, name=None)

Assigns tf.IndexedSlices to this variable batch-wise.

Analogous to batch_gather. This assumes that this variable and the sparse_delta IndexedSlices have a series of leading dimensions that are the same for all of them, and the updates are performed on the last dimension of indices. In other words, the dimensions should be the following:

num_prefix_dims = sparse_delta.indices.ndims - 1 batch_dim = num_prefix_dims + 1 `sparse_delta.updates.shape = sparse_delta.indices.shape + var.shape[

batch_dim:]`

where

sparse_delta.updates.shape[:num_prefix_dims] == sparse_delta.indices.shape[:num_prefix_dims] == var.shape[:num_prefix_dims]

And the operation performed can be expressed as:

`var[i_1, …, i_n,
sparse_delta.indices[i_1, …, i_n, j]] = sparse_delta.updates[

i_1, …, i_n, j]`

When sparse_delta.indices is a 1D tensor, this operation is equivalent to scatter_update.

To avoid this operation one can looping over the first ndims of the variable and using scatter_update on the subtensors that result of slicing the first dimension. This is a valid option for ndims = 1, but less efficient than this implementation.

Parameters
  • sparse_deltatf.IndexedSlices to be assigned to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

property constraint

Returns the constraint function associated with this variable.

Returns

The constraint function that was passed to the variable constructor. Can be None if no constraint was passed.

copy(deep=False, name=None, **overwrite_params)
Return type

ZfitObject

count_up_to(limit)

Increments this variable until it reaches limit. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Prefer Dataset.range instead.

When that Op is run it tries to increment the variable by 1. If incrementing the variable would bring it above limit then the Op raises the exception OutOfRangeError.

If no error is raised, the Op outputs the value of the variable before the increment.

This is essentially a shortcut for count_up_to(self, limit).

Parameters

limit – value at which incrementing the variable raises an error.

Returns

A Tensor that will hold the variable value before the increment. If no other Op modifies this variable, the values produced will all be distinct.

property device

The device of this variable.

property dtype

The dtype of the object

Return type

DType

eval(session=None)

In a session, computes and returns the value of this variable.

This is not a graph construction method, it does not add ops to the graph.

This convenience method requires a session where the graph containing this variable has been launched. If no session is passed, the default session is used. See tf.compat.v1.Session for more information on launching a graph and on sessions.

```python v = tf.Variable([1, 2]) init = tf.compat.v1.global_variables_initializer()

with tf.compat.v1.Session() as sess:

sess.run(init) # Usage passing the session explicitly. print(v.eval(sess)) # Usage with the default session. The ‘with’ block # above makes ‘sess’ the default session. print(v.eval())

```

Parameters

session – The session to use to evaluate this variable. If none, the default session is used.

Returns

A numpy ndarray with a copy of the value of this variable.

experimental_ref()

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use ref() instead.

property floating
static from_proto(variable_def, import_scope=None)

Returns a Variable object created from variable_def.

gather_nd(indices, name=None)

Gather slices from params into a Tensor with shape specified by indices.

See tf.gather_nd for details.

Parameters
  • indices – A Tensor. Must be one of the following types: int32, int64. Index tensor.

  • name – A name for the operation (optional).

Returns

A Tensor. Has the same type as params.

get_cache_deps(only_floating=True)

Return a set of all independent Parameter that this object depends on.

Parameters

only_floating (bool) – If True, only return floating Parameter

Return type

OrderedSet

get_dependencies(only_floating=True)

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use get_params instead if you want to retrieve the independent parameters or get_cache_deps in case you need the numerical cache dependents (advanced).

Return type

OrderedSet

get_params(floating=True, is_yield=None, extract_independent=True, only_floating=<class 'zfit.util.checks.NotSpecified'>)

Recursively collect parameters that this object depends on according to the filter criteria.

Which parameters should be included can be steered using the arguments as a filter.
  • None: do not filter on this. E.g. floating=None will return parameters that are floating as well as

    parameters that are fixed.

  • True: only return parameters that fulfil this criterion

  • False: only return parameters that do not fulfil this criterion. E.g. floating=False will return

    only parameters that are not floating.

Parameters
  • floating (Optional[bool]) – if a parameter is floating, e.g. if floating() returns True

  • is_yield (Optional[bool]) – if a parameter is a yield of the _current_ model. This won’t be applied recursively, but may include yields if they do also represent a parameter parametrizing the shape. So if the yield of the current model depends on other yields (or also non-yields), this will be included. If, however, just submodels depend on a yield (as their yield) and it is not correlated to the output of our model, they won’t be included.

  • extract_independent (Optional[bool]) – If the parameter is an independent parameter, i.e. if it is a ZfitIndependentParameter.

Return type

Set[ZfitParameter]

get_shape()

Alias of Variable.shape.

property graph

The Graph of this variable.

graph_caching_methods = []
property independent
property initial_value

Returns the Tensor used as the initial value for the variable.

Note that this is different from initialized_value() which runs the op that initializes the variable before returning its value. This method returns the tensor that is used by the op that initializes the variable.

Returns

A Tensor.

initialized_value()

Returns the value of the initialized variable. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.

You should use this instead of the variable itself to initialize another variable with a value that depends on the value of this variable.

`python # Initialize 'v' with a random tensor. v = tf.Variable(tf.random.truncated_normal([10, 40])) # Use `initialized_value` to guarantee that `v` has been # initialized before its value is used to initialize `w`. # The random values are picked only once. w = tf.Variable(v.initialized_value() * 2.0) `

Returns

A Tensor holding the value of this variable after its initializer has run.

property initializer

The initializer operation for this variable.

instances = <_weakrefset.WeakSet object>
load(value, session=None)

Load new value into this variable. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Prefer Variable.assign which has equivalent behavior in 2.X.

Writes new value to variable’s memory. Doesn’t add ops to the graph.

This convenience method requires a session where the graph containing this variable has been launched. If no session is passed, the default session is used. See tf.compat.v1.Session for more information on launching a graph and on sessions.

```python v = tf.Variable([1, 2]) init = tf.compat.v1.global_variables_initializer()

with tf.compat.v1.Session() as sess:

sess.run(init) # Usage passing the session explicitly. v.load([2, 3], sess) print(v.eval(sess)) # prints [2 3] # Usage with the default session. The ‘with’ block # above makes ‘sess’ the default session. v.load([3, 4], sess) print(v.eval()) # prints [3 4]

```

Parameters
  • value – New variable value

  • session – The session to use to evaluate this variable. If none, the default session is used.

Raises

ValueError – Session is not passed and no default session

property name
Return type

str

numpy()
property op

The Operation of this variable.

property params
read_value()
ref()

Returns a hashable reference object to this Variable.

The primary use case for this API is to put variables in a set/dictionary. We can’t put variables in a set/dictionary as variable.__hash__() is no longer available starting Tensorflow 2.0.

The following will raise an exception starting 2.0

>>> x = tf.Variable(5)
>>> y = tf.Variable(10)
>>> z = tf.Variable(10)
>>> variable_set = {x, y, z}
Traceback (most recent call last):
  ...
TypeError: Variable is unhashable. Instead, use tensor.ref() as the key.
>>> variable_dict = {x: 'five', y: 'ten'}
Traceback (most recent call last):
  ...
TypeError: Variable is unhashable. Instead, use tensor.ref() as the key.

Instead, we can use variable.ref().

>>> variable_set = {x.ref(), y.ref(), z.ref()}
>>> x.ref() in variable_set
True
>>> variable_dict = {x.ref(): 'five', y.ref(): 'ten', z.ref(): 'ten'}
>>> variable_dict[y.ref()]
'ten'

Also, the reference object provides .deref() function that returns the original Variable.

>>> x = tf.Variable(5)
>>> x.ref().deref()
<tf.Variable 'Variable:0' shape=() dtype=int32, numpy=5>
register_cacher(cacher)

Register a cacher that caches values produces by this instance; a dependent.

Parameters

cacher (Union[ForwardRef, Iterable[ForwardRef]]) –

reset_cache(reseter)
reset_cache_self()

Clear the cache of self and all dependent cachers.

scatter_add(sparse_delta, use_locking=False, name=None)

Adds tf.IndexedSlices to this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be added to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_div(sparse_delta, use_locking=False, name=None)

Divide this variable by tf.IndexedSlices.

Parameters
  • sparse_deltatf.IndexedSlices to divide this variable by.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_max(sparse_delta, use_locking=False, name=None)

Updates this variable with the max of tf.IndexedSlices and itself.

Parameters
  • sparse_deltatf.IndexedSlices to use as an argument of max with this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_min(sparse_delta, use_locking=False, name=None)

Updates this variable with the min of tf.IndexedSlices and itself.

Parameters
  • sparse_deltatf.IndexedSlices to use as an argument of min with this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_mul(sparse_delta, use_locking=False, name=None)

Multiply this variable by tf.IndexedSlices.

Parameters
  • sparse_deltatf.IndexedSlices to multiply this variable by.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_nd_add(indices, updates, name=None)

Applies sparse addition to individual values or slices in a Variable.

The Variable has rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into self. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the `K`th dimension of self.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, self.shape[K], ..., self.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

v = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) add = v.scatter_nd_add(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(add)

```

The resulting update to v would look like this:

[1, 13, 3, 14, 14, 6, 7, 20]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_sub(indices, updates, name=None)

Applies sparse subtraction to individual values or slices in a Variable.

Assuming the variable has rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into self. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the `K`th dimension of self.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, self.shape[K], ..., self.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

v = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) op = v.scatter_nd_sub(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(op)

```

The resulting update to v would look like this:

[1, -9, 3, -6, -6, 6, 7, -4]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_nd_update(indices, updates, name=None)

Applies sparse assignment to individual values or slices in a Variable.

The Variable has rank P and indices is a Tensor of rank Q.

indices must be integer tensor, containing indices into self. It must be shape [d_0, …, d_{Q-2}, K] where 0 < K <= P.

The innermost dimension of indices (with length K) corresponds to indices into elements (if K = P) or slices (if K < P) along the `K`th dimension of self.

updates is Tensor of rank Q-1+P-K with shape:

` [d_0, ..., d_{Q-2}, self.shape[K], ..., self.shape[P-1]]. `

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that update would look like this:

```python

v = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8]) indices = tf.constant([[4], [3], [1] ,[7]]) updates = tf.constant([9, 10, 11, 12]) op = v.scatter_nd_assign(indices, updates) with tf.compat.v1.Session() as sess:

print sess.run(op)

```

The resulting update to v would look like this:

[1, 11, 3, 10, 9, 6, 7, 12]

See tf.scatter_nd for more details about how to make updates to slices.

Parameters
  • indices – The indices to be used in the operation.

  • updates – The values to be used in the operation.

  • name – the name of the operation.

Returns

The updated variable.

scatter_sub(sparse_delta, use_locking=False, name=None)

Subtracts tf.IndexedSlices from this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be subtracted from this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

scatter_update(sparse_delta, use_locking=False, name=None)

Assigns tf.IndexedSlices to this variable.

Parameters
  • sparse_deltatf.IndexedSlices to be assigned to this variable.

  • use_locking – If True, use locking during the operation.

  • name – the name of the operation.

Returns

The updated variable.

Raises

TypeError – if sparse_delta is not an IndexedSlices.

set_shape(shape)

Overrides the shape for this variable.

Parameters

shape – the TensorShape representing the overridden shape.

property shape
sparse_read(indices, name=None)

Gather slices from params axis axis according to indices.

This function supports a subset of tf.gather, see tf.gather for details on usage.

Parameters
  • indices – The index Tensor. Must be one of the following types: int32, int64. Must be in range [0, params.shape[axis]).

  • name – A name for the operation (optional).

Returns

A Tensor. Has the same type as params.

property synchronization
to_proto(export_scope=None)

Converts a Variable to a VariableDef protocol buffer.

Parameters

export_scope – Optional string. Name scope to remove.

Returns

A VariableDef protocol buffer, or None if the Variable is not in the specified name scope.

property trainable
value()
classmethod from_cartesian(name, real, imag, dtype=tf.complex128, floating=True)[source]

Create a complex parameter from cartesian coordinates.

Parameters
  • name – Name of the parameter.

  • real – Real part of the complex number.

  • imag – Imaginary part of the complex number.

classmethod from_polar(name, mod, arg, dtype=tf.complex128, floating=True, **kwargs)[source]

Create a complex parameter from polar coordinates.

Parameters
  • name – Name of the parameter.

  • real – Modulus (r) the complex number.

  • imag – Argument (phi) of the complex number.

property conj

Returns a complex conjugated copy of the complex parameter.

property real

Real part of the complex parameter.

property imag

Imaginary part of the complex parameter.

property mod

Modulus (r) of the complex parameter.

property arg

Argument (phi) of the complex parameter.

zfit.convert_to_parameter(value, name=None, prefer_constant=True, dependents=None)[source]

Convert a numerical to a constant/floating parameter or return if already a parameter.

Parameters
  • value

  • name

  • prefer_constant – If True, create a ConstantParameter instead of a Parameter, if possible.

Return type

ZfitParameter

class zfit.Space(obs=None, limits=None, axes=None, rect_limits=None, name='Space')[source]

Bases: zfit.core.space.BaseSpace

Define a space with the name (obs) of the axes (and it’s number) and possibly it’s limits.

A space can be thought of as coordinates, possibly with the definition of a range (limits). For most use-cases, it is sufficient to specify a Space via observables; simple string identifiers. They can be multidimensional.

Observables are like the columns of a spreadsheet/dataframe, and are therefore needed for any object that does numerical operations or holds data in order to match the right axes. On object creation, the observables are assigned using a Space. This is often used as the default space of an object and can be used as the default norm_range, sampling limits etc.

Axes are the same concept as observables, but numbers, indexes, and are used inside an object. There, axes 0 corresponds to the 0th data column we get (which corresponds to a certain observable).

Every space can have limits; they are either rectangular or an arbitrary function (together with rectangular limits). Spaces can be combined (multiplied) to create higher dimensional spaces. Spaces can be added, which combines them into one Space consisting of two disconnected limits.

So integrating over the space consisting of the two added disconnected ranges, e.g. 0 to 1 and 2 to 3 will return the sum of the two separate integrals.

lower_band = zfit.Space('obs1', (0, 1))
upper_band = zfit.Space('obs1', (2, 3))
combined_obs = lower_band + upper_band
integral_comb = model.integrate(limits=combined_obs)
# which is equivalent to the lower
integral_sep = model.integrate(limits=lower_band) + model.integrate(limits=upper_band)
assert integral_comb == integral_sep

In principle, the same behavior could also be achieved by specifying an arbitrary function. Using the addition allows for certain optimizations inside.

Parameters
AUTO_FILL = <object object>
ANY = <Any>
ANY_LOWER = <Any Lower Limit>
ANY_UPPER = <Any Upper Limit>
get_limits(obs=None, axes=None)[source]
Return type

Union[Dict[str, Union[Dict[Tuple[int], ZfitLimit], Dict[Tuple[str], ZfitLimit]]], Dict[Tuple[int], ZfitLimit], Dict[Tuple[str], ZfitLimit]]

property limits
property rect_limits

Return the rectangular limits as np.ndarray``tf.Tensor if they are set and not false.

The rectangular limits can be used for sampling. They do not in general represent the limits of the object as a functional limit can be set and to check if something is inside the limits, the method inside() should be used.

In order to test if the limits are False or None, it is recommended to use the appropriate methods limits_are_false and limits_are_set.

Return type

Tuple[Union[ndarray, Tensor, float], Union[ndarray, Tensor, float]]

Returns

The lower and upper limits.

Raises

LimitsNotSpecifiedError – If there are not limits set or they are False.

property rect_limits_np

Return the rectangular limits as np.ndarray. Raise error if not possible.

Rectangular limits are returned as numpy arrays which can be useful when doing checks that do not need to be involved in the computation later on as they allow direct interaction with Python as compared to tf.Tensor inside a graph function.

In order to test if the limits are False or None, it is recommended to use the appropriate methods limits_are_false and limits_are_set.

Return type

Tuple[ndarray, ndarray]

Returns

A tuple of two np.ndarray with shape (1, n_obs) typically. The last

dimension is always n_obs, the first can be vectorized. This allows unstacking with z.unstack_x() as can be done with data.

Raises
property rect_lower

The lower, rectangular limits, equivalent to rect_limits[0] with shape (…, n_obs)

Return type

Union[ndarray, Tensor, float]

Returns

The lower, rectangular limits as np.ndarray or tf.Tensor

Raises

LimitsNotSpecifiedError – If the limits are not set or are false

property rect_upper

The upper, rectangular limits, equivalent to rect_limits[1] with shape (…, n_obs)

Return type

Union[ndarray, Tensor, None, bool]

Returns

The upper, rectangular limits as np.ndarray or tf.Tensor

Raises

LimitsNotSpecifiedError – If the limits are not set or are false

rect_area()[source]

Calculate the total rectangular area of all the limits and axes. Useful, for example, for MC integration.

Return type

Union[float, ndarray, Tensor]

property rect_limits_are_tensors

Return True if the rectangular limits are tensors.

If a limit with tensors is evaluated inside a graph context, comparison operations will fail.

Return type

bool

Returns

If the rectangular limits are tensors.

property has_rect_limits

If there are limits and whether they are rectangular.

Return type

bool

property limits_are_false

If the limits have been set to False, so the object on purpose does not contain limits.

Return type

bool

Returns

True if limits is False

property has_limits

Whether there are limits set and they are not false.

Returns:

Return type

bool

property limits_are_set
property n_events

Return the number of events, the dimension of the first shape.

Return type

Optional[int]

Returns

Number of events, the dimension of the first shape. If this is > 1 or None,

it’s vectorized.

property limit2d

DEPRECATED FUNCTION

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Depreceated, use .rect_limits or .inside to check if a value is inside or userect_limits to receive the rectangular limits.

property limits1d

return the tuple(low_1, …, low_n, up_1, …, up_n).

Return type

Tuple[float]

Returns

So low_1, low_2, up_1, up_2 = space.limits1d for several, 1 obs limits.

low_1 to up_1 is the first interval, low_2 to up_2 is the second interval etc.

Raises

RuntimeError – if the conditions (n_obs or n_limits) are not satisfied.

Type

Simplified .limits for exactly 1 obs, n limits

property lower
property upper
property n_limits

The number of different limits.

Return type

int

Returns

int >= 1

property iter_limits

REMOVED.Return the limits, either as Space objects or as pure limits-tuple. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Iterate over the space directly and use the limits from the spaces.

This makes iterating over limits easier: for limit in space.iter_limits() allows to, for example, pass limit to a function that can deal with simple limits only or if as_tuple is True the limit can be directly used to calculate something.

Example

for lower, upper in space.iter_limits(as_tuple=True):
    integrals = integrate(lower, upper)  # calculate integral
integral = sum(integrals)
Returns

Return type

List[Space] or List[limit,…]

with_limits(limits=None, rect_limits=None, name=None)[source]

Return a copy of the space with the new limits (and the new name).

Parameters
Return type

ZfitSpace

Returns

Copy of the current object with the new limits.

reorder_x(x, *, x_obs=None, x_axes=None, func_obs=None, func_axes=None)[source]

Reorder x in the last dimension either according to its own obs or assuming a function ordered with func_obs.

There are two obs or axes around: the one associated with this Coordinate object and the one associated with x. If x_obs or x_axes is given, then this is assumed to be the obs resp. the axes of x and x will be reordered according to self.obs resp. self.axes.

If func_obs resp. func_axes is given, then x is assumed to have self.obs resp. self.axes and will be reordered to align with a function ordered with func_obs resp. func_axes.

Switching func_obs for x_obs resp. func_axes for x_axes inverts the reordering of x.

Parameters
  • x (Union[Tensor, ndarray]) – Tensor to be reordered, last dimension should be n_obs resp. n_axes

  • x_obs (Union[str, Iterable[str], Space, None]) – Observables associated with x. If both, x_obs and x_axes are given, this has precedency over the latter.

  • x_axes (Union[int, Iterable[int], None]) – Axes associated with x.

  • func_obs (Union[str, Iterable[str], Space, None]) – Observables associated with a function that x will be given to. Reorders x accordingly and assumes self.obs to be the obs of x. If both, func_obs and func_axes are given, this has precedency over the latter.

  • func_axes (Union[int, Iterable[int], None]) – Axe associated with a function that x will be given to. Reorders x accordingly and assumes self.axes to be the axes of x.

Return type

Union[ndarray, Tensor]

Returns

The reordered array-like object

with_obs(obs, allow_superset=True, allow_subset=True)[source]

Create a new Space that has obs; sorted by or set or dropped.

The behavior is as follows:

  • obs are already set:

    • input obs are None: the observables will be dropped. If no axes are set, an error will be raised, as no coordinates will be assigned to this instance anymore.

    • input obs are not None: the instance will be sorted by the incoming obs. If axes or other objects have an associated order (e.g. data, limits,…), they will be reordered as well. If a strict subset is given (and allow_subset is True), only a subset will be returned. This can be used to take a subspace of limits, data etc. If a strict superset is given (and allow_superset is True), the obs will be sorted accordingly as if the obs not contained in the instances obs were not in the input obs.

  • obs are not set:

    • if the input obs are None, the same object is returned.

    • if the input obs are not None, they will be set as-is and now correspond to the already existing axes in the object.

Parameters
  • obs (Union[str, Iterable[str], Space, None]) – Observables to sort/associate this instance with

  • allow_superset (bool) – if False and a strict superset of the own observables is given, an error

  • raised. (is) –

  • allow_subset (bool) – if False and a strict subset of the own observables is given, an error

  • raised.

Return type

ZfitSpace

Returns

A copy of the object with the new ordering/observables

Raises
with_axes(axes, allow_superset=True, allow_subset=True)[source]

Create a new instance that has axes; sorted by or set or dropped.

The behavior is as follows:

  • axes are already set:

    • input axes are None: the axes will be dropped. If no observables are set, an error will be raised, as no coordinates will be assigned to this instance anymore.

    • input axes are not None: the instance will be sorted by the incoming axes. If obs or other objects have an associated order (e.g. data, limits,…), they will be reordered as well. If a strict subset is given (and allow_subset is True), only a subset will be returned. This can be used to retrieve a subspace of limits, data etc. If a strict superset is given (and allow_superset is True), the axes will be sorted accordingly as if the axes not contained in the instances axes were not present in the input axes.

  • axes are not set:

    • if the input axes are None, the same object is returned.

    • if the input axes are not None, they will be set as-is and now correspond to the already existing obs in the object.

Parameters
  • axes (Union[int, Iterable[int], None]) – Axes to sort/associate this instance with

  • allow_superset (bool) – if False and a strict superset of the own axeservables is given, an error

  • raised. (is) –

  • allow_subset (bool) – if False and a strict subset of the own axeservables is given, an error

  • raised.

Return type

ZfitSpace

Returns

A copy of the object with the new ordering/axes

Raises
with_coords(coords, allow_superset=True, allow_subset=True)[source]

Create a new Space with reordered observables and/or axes.

The behavior is that _at least one coordinate (obs or axes) has to be set in both instances (the space itself or in coords). If both match, observables is taken as the defining coordinate. The space is sorted according to the defining coordinate and the other coordinate is sorted as well. If either the space did not have the “weaker coordinate” (e.g. both have observables, but only coords has axes), then the resulting Space will have both. If both have both coordinates, obs and axes, and sorting for obs results in non-matchin axes results in axes being dropped.

Parameters
  • coords (ZfitOrderableDimensional) – An instance of Coordinates

  • allow_superset (bool) – If False and a strict superset is given, an error is raised

  • allow_subset (bool) – If False and a strict subset is given, an error is raised

Returns

Return type

Space

Raises
with_autofill_axes(overwrite=False)[source]

Overwrite the axes of the current object with axes corresponding to range(len(n_obs)).

This effectively fills with (0, 1, 2,…) and can be used mostly when an object enters a PDF or similar. overwrite allows to remove the axis first in case there are already some set.

object.obs -> ('x', 'z', 'y')
object.axes -> None

object.with_autofill_axes()

object.obs -> ('x', 'z', 'y')
object.axes -> (0, 1, 2)
Parameters

overwrite (bool) – If axes are already set, replace the axes with the autofilled ones. If axes is already set and overwrite is False, raise an error.

Return type

Space

Returns

The object with the new axes

Raises

AxesIncompatibleError – if the axes are already set and overwrite is False.

get_subspace(obs=None, axes=None, name=None)[source]

Create a Space consisting of only a subset of the obs/axes (only one allowed).

Parameters
Return type

Space

Returns

A space containing only a subspace (and sublimits etc.)

with_obs_axes(**kwargs)[source]
get_obs_axes(obs=None, axes=None)[source]
property obs_axes
area(**kwargs)[source]
copy(**overwrite_kwargs)[source]

Create a new Space using the current attributes and overwriting with overwrite_overwrite_kwargs.

Parameters
  • name – The new name. If not given, the new instance will be named the same as the current one.

  • **overwrite_kwargs

Return type

Space

Returns

Space

property limit1d
classmethod from_axes(cls, axes, limits=None, rect_limits=None, name=None)[source]

Create a space from axes instead of from obs. (deprecated)

Warning: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use directly the class to create a Space. E.g. zfit.Space(axes=(0, 1), …)

Parameters
Return type

Space

Returns

Space

__eq__(other)

Compares two Limits for equality without graph mode allowed.

Returns:

Raises

IllegalInGraphModeError – it the comparison happens with tensors in a graph context.

Return type

bool

__le__(other)

Set-like comparison for compatibility. If an object is less_equal to another, the limits are combatible.

This can be used to determine whether a fitting range specification can handle another limit.

Return type

bool

Returns

Result of the comparison

Raises

IllegalInGraphModeError – it the comparison happens with tensors in a graph context.

add(*other)

Add the limits of the spaces. Only works for the same obs.

In case the observables are different, the order of the first space is taken.

Parameters

other (Union[Space, Iterable[Space]]) –

Returns

Return type

Space

property axes

The axes (“obs with int”) the space is defined in.

Returns:

Return type

Optional[Tuple[int]]

combine(*other)

Combine spaces with different obs (but consistent limits).

Parameters

other (Union[Space, Iterable[Space]]) –

Returns

Return type

Space

equal(other, allow_graph)

Compare the limits on equality. For ANY objects, this also returns true.

If called inside a graph context and the limits are tensors, this will return a symbolic tf.Tensor.

Return type

Union[bool, Tensor]

Returns

Result of the comparison

Raises

IllegalInGraphModeError – it the comparison happens with tensors in a graph context.

filter(x, guarantee_limits=False, axis=None)

Filter x by removing the elements along axis that are not inside the limits.

This is similar to tf.boolean_mask.

Parameters
  • x (Union[ndarray, Tensor, Data]) – Values to be checked whether they are inside of the limits. If not, the corresonding element (in the specified axis) is removed. The shape is expected to have the last dimension equal to n_obs.

  • guarantee_limits (bool) – Guarantee that the values are already inside the rectangular limits.

  • axis (Optional[int]) – The axis to remove the elements from. Defaults to 0.

Return type

Union[ndarray, Tensor]

Returns

Return an object with the same shape as x except that along axis elements have been

removed.

get_reorder_indices(obs=None, axes=None)

Indices that would order the instances obs as obs respectively the instances axes as axes.

Parameters
  • obs (Union[str, Iterable[str], Space, None]) – Observables that the instances obs should be ordered to. Does not reorder, but just return the indices that could be used to reorder.

  • axes (Union[int, Iterable[int], None]) – Axes that the instances obs should be ordered to. Does not reorder, but just return the indices that could be used to reorder.

Return type

Tuple[int]

Returns

New indices that would reorder the instances obs to be obs respectively axes.

Raises

CoordinatesUnderdefinedError – If neither obs nor axes is given

get_sublimits()
inside(x, guarantee_limits=False)

Test if x is inside the limits.

This function should be used to test if values are inside the limits. If the given x is already inside the rectangular limits, e.g. because it was sampled from within them

Parameters
  • x (Union[ndarray, Tensor, Data]) – Values to be checked whether they are inside of the limits. The shape is expected to have the last dimension equal to n_obs.

  • guarantee_limits (bool) – Guarantee that the values are already inside the rectangular limits.

Return type

Union[ndarray, Tensor, Data]

Returns

Return a boolean tensor-like object with the same shape as the input x except of the

last dimension removed.

less_equal(other, allow_graph)

Set-like comparison for compatibility. If an object is less_equal to another, the limits are combatible.

This can be used to determine whether a fitting range specification can handle another limit.

If called inside a graph context and the limits are tensors, this will return a symbolic tf.Tensor.

Parameters
  • other – Any other object to compare with

  • allow_graph – If False and the function returns a symbolic tensor, raise IllegalInGraphModeError instead.

Returns

Result of the comparison

Raises

IllegalInGraphModeError – it the comparison happens with tensors in a graph context.

property n_obs

Return the number of observables/axes.

Return type

int

Returns

int >= 1

property name

The name of the object.

Return type

str

property obs

The observables (“axes with str”)the space is defined in.

Returns:

Return type

Optional[Tuple[str, …]]

zfit.convert_to_space(obs=None, axes=None, limits=None, *, overwrite_limits=False, one_dim_limits_only=True, simple_limits_only=True)[source]

Convert limits to a Space object if not already None or False.

Parameters
Returns

Return type

Union[Space, False, None]

Raises

OverdefinedError – if obs or axes is a Space and axes respectively obs is not None.

zfit.supports(*, norm_range=False, multiple_limits=False)[source]

Decorator: Add (mandatory for some methods) on a method to control what it can handle.

If any of the flags is set to False, it will check the arguments and, in case they match a flag (say if a norm_range is passed while the norm_range flag is set to False), it will raise a corresponding exception (in this example a NormRangeNotImplementedError) that will be catched by an earlier function that knows how to handle things.

Parameters
  • norm_range (bool) – If False, no norm_range argument will be passed through resp. will be None

  • multiple_limits (bool) – If False, only simple limits are to be expected and no iteration is therefore required.

Return type

Callable

Subpackages