src.cmfsapy.dimension package¶
Submodules¶
src.cmfsapy.dimension.cmfsa module¶
-
src.cmfsapy.dimension.cmfsa.
calibrate
(n=2500, myk=20, emb_dims=array([2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80]), N_realiz=100, powers=[- 1, 1, 2, 3], box=None, load_data=False, save_data=False)¶ Computes regression coefs on calibration dataset with known intrinsic dimensionality
- Parameters
n (int) – sample size per realization
myk (int) – maximal neighborhood size
of int emb_dims (numpy.ndarray) – embedding dimensions to estimate dimension
N_realiz (int) – number of realizations per embedding dimension
of in powers (list) – powers to be used in the polynomial basis
or float box (None) – box size for periodic boundary (default=None)
or str load_data (bool) – if a string, then took as path to data
or str save_data (bool) – if a string then took as path to save
- Returns
regression coefficients
- Return type
list of float
-
src.cmfsapy.dimension.cmfsa.
cmfsa
(X, k, powers=None, alphas=None, boxsize=None)¶ Computes corrigated dimension estimations on dataset @todo: Hard-wire default values from the article
- Parameters
of float X (numpy.ndarray) – data
k (int) –
of float powers (list) – powers
of float alphas (list) – regression coeffitients
boxsize (float) –
- Returns
src.cmfsapy.dimension.correction module¶
-
src.cmfsapy.dimension.correction.
compute_mFSA_correction_coef
(d, E, powers=[1, 2])¶ Compute the regression coefficients with Orthogonal Distance Regression
- Parameters
of float d (numpy.ndarray) – dimension values
of float E (numpy.ndarray) – relative error
of float powers (numpy.ndarray) – the powers of the polynomial to include in the regression
- Returns
regression coefficients
- Return type
numpy.ndarray of float
-
src.cmfsapy.dimension.correction.
correct_estimates
(d, alpha, powers)¶ Correct mFSA estimates given rergression coefficients and the coresponding powers of the polynomial
- Parameters
d (float) – measured mFSA value(s)
of float alpha (numpy.ndarray) – regression coefs
of float powers (numpy.ndarray) – powers of the polynomial
- Returns
corrigated-mFSA value(s)
- Return type
float
-
src.cmfsapy.dimension.correction.
correct_mFSA
(d, E, powers)¶ Correct mFSA values given the relative error of the measurements fit
- Parameters
of float d (numpy.ndarray) – mFSA values
of float E (numpy.ndarray) – relative error of mFSA values
of float powers (numpy.ndarray) –
- Returns
corrected estimates
- Return type
numpy.ndarray of float
-
src.cmfsapy.dimension.correction.
polynom_func
(p, x, powers=[1, 2, 3])¶ Computes the value of polynomial expression with given mixing coefficients and powers
- Parameters
of float p (list) – mixing coefficients (length has to match the length of powers)
or numpy.ndarray of float x (float) – the value of the variable
of float powers (list) – the powers of (length has to match the length of p)
- Returns
the value of the polynomial at the place x
- Return type
float or numpy.ndarray of float
src.cmfsapy.dimension.fsa module¶
-
src.cmfsapy.dimension.fsa.
fsa
(X, k, boxsize=None)¶ Measure local Szepesvari-Farahmand dimension, distances are computed by the cKDTree algoritm
- Parameters
X (arraylike) – data series [n x dim] shape
k – maximum k value
boxsize – apply d-toroidal distance computation with edge-size =boxsize, see ckdtree class for more
- Returns
local estimates, distances, indicees
-
src.cmfsapy.dimension.fsa.
get_dists_inds_ck
(X, k, boxsize)¶ computes the kNN distances and indices
- Parameters
X (numpy.ndarray) – 2D array with data shape: (ndata, n_vars)
k (int) – neighborhood size
boxsize (float) – circular boundary condition to [0, boxsice] interval for all input dimensions if not None.
- Returns
KNN distances and indices
-
src.cmfsapy.dimension.fsa.
ml_dims
(X, k2, k1=1)¶ Maximum likelihood estimator af intrinsic dimension (Levina-Bickel)
-
src.cmfsapy.dimension.fsa.
ml_estimator
(normed_dists)¶
-
src.cmfsapy.dimension.fsa.
szepes_ml
(local_d)¶ maximum likelihood estimator from local FSA estimates (for k=1)
- Parameters
of float local_d (numpy.ndarray) – local FSA estimates
- Returns
global ML-FSA estimate
-
src.cmfsapy.dimension.fsa.
szepesvari_dimensionality
(dists)¶ Compute szepesvari dimensions from kNN distances
- Parameters
dists –
- Returns