SamplerData#
- class zfit.data.SamplerData(data, *, sample_and_weights_func, sample_holder, n, weights=None, weights_holder=None, params=None, obs=None, name=None, label=None, dtype=tf.float64, use_hash=None, guarantee_limits=False)[source]#
Bases:
Data
Create a
SamplerData
object.Use constructor
from_sampler
instead.- property fixed_params#
DEPRECATED FUNCTION
Deprecated: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use
params
instead.
- classmethod from_sample(sample_func: Callable, n: ztyping.NumericalScalarType, obs: ztyping.ObsTypeInput, fixed_params=None, name: str | None = None, weights=None, dtype=None, use_hash: bool | None = None)[source]#
DEPRECATED FUNCTION
Deprecated: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use
from_sampler
instead (with an ‘r’ at the end).
- classmethod from_sampler(cls, *, sample_func=None, sample_and_weights_func=None, n, obs, params=None, fixed_params=None, name=None, label=None, dtype=None, use_hash=None, guarantee_limits=False)[source]#
Create a
SamplerData
from a sampler function. (deprecated arguments)Deprecated: SOME ARGUMENTS ARE DEPRECATED:
(fixed_params)
. They will be removed in a future version. Instructions for updating: Useparams
instead.This is a more flexible way to create a
SamplerData
. Instead of providing a fixed sample, a sampler function is provided that will be called to sample the data. If the data is used in the loss, the sampler function will updated the value in the compiled version.Note
If any method of the
SamplerData
is used to create a new data object, such aswith_obs
, the resulting data will be aData
object and not aSamplerData
object; the data will be fixed and not resampled.- Parameters:
sample_func (
Optional
[Callable
]) – A callable that takes as argumentn
and returns a sample of the data. The sample has to have the same number of observables as theobs
of theSamplerData
. IfNone
,sample_and_weights_func
has to be given.sample_and_weights_func (
Optional
[Callable
]) – A callable that takes as argumentn
and returns a tuple of the sample and the weights of the data. The sample has to have the same number of observables as theobs
of theSamplerData
. IfNone
,sample_func
has to be given.n (
Union
[int
,float
,complex
,Tensor
,ZfitParameter
]) – The number of samples to produce initially. This is used to have a first sample that can be used for compilation.obs (
Union
[str
,Iterable
[str
],Space
]) – Observables of the data. If the space has limits, the data will be cut to the limits.params (
Optional
[Mapping
[Union
[str
,ZfitParameter
],Union
[int
,float
,complex
,Tensor
,ZfitParameter
]]]) – A mapping fromParameter
or a string to a numerical value. This is used as the default values for the parameters in thesample_func
orsample_and_weights_func
and needs to fully specify the parameters.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.dtype – The dtype of the data.
use_hash (
bool
|None
) – If true, store a hash for caching. If a PDF can cache values, this option needs to be enabled for the PDF to be able to cache values.guarantee_limits (
bool
) – Guarantee that the data is within the limits. IfTrue
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.
- update_data(sample, weights=None, guarantee_limits=False)[source]#
Load a new sample into the dataset, presumably similar to the previous one.
- Parameters:
sample (
Union
[Tensor
,TensorProtocol
,int
,float
,bool
,str
,bytes
,complex
,tuple
,list
,ndarray
,generic
,ZfitUnbinnedData
]) – The new sample to load. Has to have the same number of observables as theobs
of theSamplerData
but can have a different number of events. When aZfitUnbinnedData
is given, the weights from it are used.weights (
Union
[Tensor
,TensorProtocol
,int
,float
,bool
,str
,bytes
,complex
,tuple
,list
,ndarray
,generic
,None
]) – The weights of the new sample. IfNone
, the weights are not changed. If theSamplerData
was initialized with weights, this has to be given. If theSamplerData
was initialized without weights, this cannot be given.guarantee_limits (
bool
) – Guarantee that the data is within the limits. IfTrue
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.
- resample(params=None, *, n=None, param_values=None)[source]#
Update the sample by newly sampling inplace, affecting objects that have it internally, like likelihoods. (deprecated arguments)
Deprecated: SOME ARGUMENTS ARE DEPRECATED:
(param_values)
. They will be removed in a future version. Instructions for updating: Useparams
instead.The values of the parameters used to sample the PDF are taken from the creation of the Sampler and won’t change by setting their values, i.e. using
param.set_values
. Parameter values (some or all) can be overridden by providing theparams
argument.- Parameters:
params (
Optional
[Mapping
[Union
[str
,ZfitParameter
],Union
[int
,float
,complex
,Tensor
,ZfitParameter
]]]) – a mapping fromParameter
or string to avalue
so that the sampler will use this value for the sampling. If not given, theparams
will be used.n (
Union
[Tensor
,TensorProtocol
,int
,float
,bool
,str
,bytes
,complex
,tuple
,list
,ndarray
,generic
]) – the number of samples to produce. If theSampler
was created with anything else then a numerical or tf.Tensor, this can’t be used.
- add_cache_deps(cache_deps, allow_non_cachable=True)#
Add dependencies that render the cache invalid if they change.
- property data_range#
DEPRECATED FUNCTION
Deprecated: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use
space
instead.
- enable_hashing()#
Enable hashing for this data object if it was disabled.
A hash allows some objects to be cached and reused. If a hash is enabled, the data object will be hashed and the hash _can_ be used for caching. This can speedup various objects, however, it maybe doesn’t have an effect at all. For example, if an object was already called before with the data object, the hash will probably not be used, as the object is already compiled.
- classmethod from_asdf(asdf_obj, *, reuse_params=None)#
Load an object from an asdf file.
Args#
asdf_obj: Object reuse_params:If parameters, the parameters
will be reused if they are given. If a parameter is given, it will be used as the parameter with the same name. If a parameter is not given, a new parameter will be created.
- classmethod from_dict(dict_, *, reuse_params=None)#
Creates an object from a dictionary structure as generated by
to_dict
.- Parameters:
dict – Dictionary structure.
reuse_params – If parameters, the parameters will be reused if they are given. If a parameter is given, it will be used as the parameter with the same name. If a parameter is not given, a new parameter will be created.
- Returns:
The deserialized object.
- classmethod from_json(cls, json, *, reuse_params=None)#
Load an object from a json string.
- Parameters:
json (
str
) – Serialized object in a JSON string.reuse_params – If parameters, the parameters will be reused if they are given. If a parameter is given, it will be used as the parameter with the same name. If a parameter is not given, a new parameter will be created.
- Return type:
- Returns:
The deserialized object.
- classmethod from_mapping(mapping, obs=None, *, weights=None, label=None, name=None, dtype=None, use_hash=None, guarantee_limits=False)#
Create a
Data
from a mapping of observables to arrays.- Parameters:
mapping (
Mapping
[str
,Union
[Tensor
,TensorProtocol
,int
,float
,bool
,str
,bytes
,complex
,tuple
,list
,ndarray
,generic
]]) – A mapping from the observables to the data, with the observables as keys and the data as values.obs (
Union
[str
,Iterable
[str
],Space
]) –Space of the data. The space is used to define the observables and the limits of the data. If the
Space
has limits, these will be used to cut the data. If the data is already cut, useguarantee_limits
for a possible performance improvement.They will be matched to the data in the same order. Can be omitted, in which case the keys of the mapping are used as observables.
weights (
Union
[Tensor
,TensorProtocol
,int
,float
,bool
,str
,bytes
,complex
,tuple
,list
,ndarray
,generic
,None
]) –Weights of the data. Has to be 1-D and match the shape of the data (nevents). Note that a weighted dataset may not be supported by all methods or need additional approximations to correct for the weights, taking more time.
Can also be a string that is a column in the dataframe. By default, look for a column
""
, i.e., an empty string.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.dtype (
DType
) – dtype of the datause_hash (
bool
|None
) – If true, store a hash for caching. If a PDF can cache values, this option needs to be enabled for the PDF to be able to cache values.guarantee_limits (
bool
|None
) – Guarantee that the data is within the limits. IfTrue
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.
- Returns:
- data.init.returns|
zfit.Data
orzfit.BinnedData
: A
Data
object containing the unbinned data or aBinnedData
if the obs is binned.
- data.init.returns|
- Return type:
|@doc
- Raises:
ValueError – If the observables are not in the mapping.
- classmethod from_numpy(obs, array, *, weights=None, name=None, label=None, dtype=None, use_hash=None, guarantee_limits=False)#
Create
Data
from anp.array
.- Parameters:
obs (
Union
[str
,Iterable
[str
],Space
]) – Space of the data. The space is used to define the observables and the limits of the data. If theSpace
has limits, these will be used to cut the data. If the data is already cut, useguarantee_limits
for a possible performance improvement.array (
ndarray
) – Numpy array containing the data. Has to be of shape (nevents, nobs) or, if only one observable, (nevents) is also possible.weights (
Union
[Tensor
,None
,ndarray
]) – Weights of the data. Has to be 1-D and match the shape of the data (nevents). Note that a weighted dataset may not be supported by all methods or need additional approximations to correct for the weights, taking more time.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.dtype (
DType
) – dtype of the data.use_hash – If true, store a hash for caching. If a PDF can cache values, this option needs to be enabled for the PDF to be able to cache values.
guarantee_limits (
bool
) – Guarantee that the data is within the limits. IfTrue
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.
- Returns:
- data.init.returns|
zfit.Data
orzfit.BinnedData
: A
Data
object containing the unbinned data or aBinnedData
if the obs is binned.
- data.init.returns|
- Return type:
|@doc
- Raises:
TypeError – If the array is not a numpy array.
- classmethod from_pandas(df, obs=None, *, weights=None, name=None, label=None, dtype=None, use_hash=None, guarantee_limits=False)#
Create a
Data
from a pandas DataFrame. Ifobs
isNone
, columns are used as obs.- Parameters:
df (
DataFrame
) – pandas DataFrame that contains the data. Ifobs
isNone
, columns are used as obs. Can be a superset of obs.obs (
Union
[str
,Iterable
[str
],Space
]) –Space of the data. The space is used to define the observables and the limits of the data. If the
Space
has limits, these will be used to cut the data. If the data is already cut, useguarantee_limits
for a possible performance improvement.If
None
, columns are used as obs.weights (
Union
[Tensor
,None
,ndarray
,str
]) – Weights of the data. Has to be 1-D and match the shape of the data (nevents). Note that a weighted dataset may not be supported by all methods or need additional approximations to correct for the weights, taking more time.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.guarantee_limits (
bool
) –Guarantee that the data is within the limits. If
True
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.For example, if the data is apd.DataFrame
and the limitsof
obs
have already been enforced through aquery
on the DataFrame, the limits are guaranteed to be correct and the data will not be checked again. Possible speedup, should not have any effect on the result.dtype (
DType
) – The DType of the return value. Defaults to the zfit default (usually float64).use_hash (
bool
|None
) – If true, store a hash for caching. If a PDF can cache values, this option needs to be enabled for the PDF to be able to cache values.
- Returns:
- data.init.returns|
zfit.Data
orzfit.BinnedData
: A
Data
object containing the unbinned data or aBinnedData
if the obs is binned.
- data.init.returns|
- Return type:
|@doc
- Raises:
ValueError – If the observables are not in the dataframe.
- classmethod from_root(path, treepath, obs=None, *, weights=None, obs_alias=None, name=None, label=None, dtype=None, root_dir_options=None, use_hash=None, branches=None, branches_alias=None)#
Create a
Data
from a ROOT file. Arguments are passed touproot
.The arguments are passed to uproot directly.
- Parameters:
path (
str
) – Path to the root file.treepath (
str
) – Name of the tree in the root file.obs (
ZfitSpace
) – Observables of the data. This will also be the columns of the data if not obs_alias is given.weights (
Union
[Tensor
,None
,ndarray
,str
]) – Weights of the data. Has to be 1-D and match the shape of the data (nevents). Can be a column of the ROOT file by using a string corresponding to a column.obs_alias (
Mapping
[str
,str
] |None
) – A mapping from theobs
(as keys) to the actualbranches
(as values) in the root file. This allows to have differentobservable
names, independent of the branch name in the file.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.dtype (
DType
) – dtype of the data.root_dir_options – Options passed to uproot.
use_hash (
bool
|None
) – IfTrue
, a hash of the data is created and is used to identify it in caching.
- Returns:
A
Data
object containing the unbinned data.- Return type:
zfit.Data
- classmethod from_tensor(obs, tensor, *, weights=None, name=None, label=None, dtype=None, use_hash=None, guarantee_limits=False)#
Create a
Data
from atf.Tensor
- Parameters:
obs (
Union
[str
,Iterable
[str
],Space
]) – Space of the data. The space is used to define the observables and the limits of the data. If theSpace
has limits, these will be used to cut the data. If the data is already cut, useguarantee_limits
for a possible performance improvement.tensor (
Tensor
) – Tensor containing the data. Has to be of shape (nevents, nobs) or, if only one observable, (nevents) is also possible.weights (
Union
[Tensor
,None
,ndarray
]) – Weights of the data. Has to be 1-D and match the shape of the data (nevents). Note that a weighted dataset may not be supported by all methods or need additional approximations to correct for the weights, taking more time.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.dtype (
DType
) – dtype of the data.use_hash – If true, store a hash for caching. If a PDF can cache values, this option needs to be enabled for the PDF to be able to cache values.
guarantee_limits (
bool
) – Guarantee that the data is within the limits. IfTrue
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.
- Returns:
- data.init.returns|
zfit.Data
orzfit.BinnedData
: A
Data
object containing the unbinned data or aBinnedData
if the obs is binned.
- data.init.returns|
- Return type:
|@doc
- Raises:
TypeError – If the tensor is not a tensorflow tensor.
- property nevents#
DEPRECATED FUNCTION
Deprecated: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use
nentries
(for an int) orsamplesize
for a float corresponding to the sum of weights) instead.
- register_cacher(cacher)#
Register a
cacher
that caches values produces by this instance; a dependent.- Parameters:
cacher (ztyping.CacherOrCachersType)
- reset_cache_self()#
Clear the cache of self and all dependent cachers.
- set_data_range(data_range)#
DEPRECATED FUNCTION
Deprecated: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Do not change the range, preferably use pandas or similar, or use
with_obs
instead.
- set_weights(weights: ztyping.WeightsInputType)#
Set (temporarily) the weights of the dataset. (deprecated)
Deprecated: THIS FUNCTION IS DEPRECATED. It will be removed in a future version. Instructions for updating: Use
with_weights
instead.
- to_asdf()#
Convert the object to an asdf file.
- to_binned(space, *, name=None, label=None, use_hash=None)#
Bins the data using
space
and returns aBinnedData
object.- Parameters:
space (
Space
) – The space to bin the data in.name (
str
|None
) – Name of the data. This can possibly be used for future identification, with possible implications on the serialization and deserialization of the data. The name should therefore be “machine-readable” and not contain special characters. (currently not used for a special purpose) For a human-readable name or description, use the label.label (
str
|None
) – Human-readable name or label of the data for a better description, to be used with plots etc. Can contain arbitrary characters. Has no programmatical functional purpose as identification.use_hash (
bool
|None
) – If true, store a hash for caching. If a PDF can cache values, this option needs to be enabled for the PDF to be able to cache values.
- Returns:
A new
BinnedData
object containing the binned data.- Return type:
zfit.BinnedData
- to_dict()#
Convert the object to a nested dictionary structure.
- Returns:
The dictionary structure.
- Return type:
- to_numpy()#
Return the data as a numpy array.
Pandas DataFrame equivalent method :returns: The data as a numpy array. :rtype: np.ndarray
- to_pandas(obs=None, weightsname=None)#
Create a
pd.DataFrame
fromobs
as columns and return it.- Parameters:
- Returns:
A
pd.DataFrame
containing the data and the weights (if present).- Return type:
pd.DataFrame
- unstack_x(obs=None, always_list=None)#
Return the unstacked data: a list of tensors or a single Tensor.
- value(obs=None, axis=None)#
Return the data as a numpy-like object in
obs
order.- Parameters:
obs (
Union
[str
,Iterable
[str
],Space
]) – Observables to return. IfNone
, all observables are returned. Can be a subset of the original observables. If a string is given, a 1-D array is returned with shape (nevents,). If a list of strings or azfit.Space
is given, a 2-D array is returned with shape (nevents, nobs).axis (
int
|None
) – If given, the axis to return instead of the full data. Ifobs
is a string, this has to beNone
.
- Return type:
Tensor
Returns:
- property weights#
Get the weights of the data.
- with_obs(obs, *, guarantee_limits=False)#
Create a new
Data
with a subset of the data using the obs.- Parameters:
obs (
Union
[str
,Iterable
[str
],Space
]) – Observables to return. Has to be a subset of the original observables.guarantee_limits (
bool
) – Guarantee that the data is within the limits. IfTrue
, the data will not be checked and _is assumed_ to be within the limits, possibly because it was already cut before. This can lead to a performance improvement as the data does not have to be checked.
- Returns:
A new
Data
object containing the subset of the data.- Return type:
zfit.Data