Skip to content

pymdea.core

Diffusion entropy analysis core methods.

DeaEngine(loader)

Run diffusion entropy analysis.

Source code in src/pymdea/core.py 
133
134
135
def __init__(self: Self, loader: DeaLoader) -> Self:
    """Run diffusion entropy analysis."""
    self.data = loader.data

analyze_with_stripes(fit_start, fit_stop, fit_method='siegel', n_stripes=20)

Run a modified diffusion entropy analysis.

Parameters:

  • fit_start (float) –

    Fraction of maximum window length at which to start linear fit.

  • fit_stop (float) –

    Fraction of maximum window length at which to stop linear fit.

  • fit_method (str {"siegel", "theilsen", "ls"}, default: 'siegel' ) –

    Linear fit method to use. By default "siegel"

  • n_stripes (int, default: 20 ) –

    Number of stripes to apply to input time-series during analysis.

Returns:

  • Self @ Engine –

    Object containing the results and inputs of the diffusion entropy analysis.

Raises:

  • ValueError –

    If n_stripes < 2. At least two stripes must be applied for DEA to provide a meaningful result.

Notes

Prefer the siegel or theilsen methods. Least squares linear fits can introduce bias when done over log-scale data, see Clauset, A., Shalizi, C.R. and Newman, M.E., 2009. Power-law distributions in empirical data. SIAM review, 51(4), pp.661-703. https://doi.org/10.1137/070710111. https://arxiv.org/pdf/0706.1062.pdf.

Source code in src/pymdea/core.py 
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
def analyze_with_stripes(
    self: Self,
    fit_start: int,
    fit_stop: int,
    fit_method: Literal["siegel", "theilsen", "ls"] = "siegel",
    n_stripes: int = 20,
) -> Self:
    """Run a modified diffusion entropy analysis.

    Parameters
    ----------
    fit_start : float
        Fraction of maximum window length at which to start linear fit.
    fit_stop : float
        Fraction of maximum window length at which to stop linear fit.
    fit_method : str {"siegel", "theilsen", "ls"}, optional
        Linear fit method to use. By default "siegel"
    n_stripes : int, optional, default: 20
        Number of stripes to apply to input time-series during
        analysis.

    Returns
    -------
    Self @ Engine
        Object containing the results and inputs of the diffusion
        entropy analysis.

    Raises
    ------
    ValueError
        If n_stripes < 2. At least two stripes must be applied for
        DEA to provide a meaningful result.

    Notes
    -----
    Prefer the siegel or theilsen methods. Least squares linear
    fits can introduce bias when done over log-scale data, see
    Clauset, A., Shalizi, C.R. and Newman, M.E., 2009. Power-law
    distributions in empirical data. SIAM review, 51(4), pp.661-703.
    https://doi.org/10.1137/070710111.
    https://arxiv.org/pdf/0706.1062.pdf.

    """
    if n_stripes < 2:  # noqa: PLR2004
        msg = "n_stripes must be greater than 1"
        raise ValueError(msg)
    self.number_of_stripes = n_stripes
    self.fit_start = fit_start
    self.fit_stop = fit_stop
    self.fit_method = fit_method
    self._apply_stripes()
    self._get_events()
    self._make_trajectory()
    self._calculate_entropy()
    self._calculate_scaling()
    self._calculate_mu()
    self.print_result()

analyze_without_stripes(fit_start, fit_stop, fit_method='siegel')

Run a regular diffusion entropy analysis.

Parameters:

  • fit_start (float) –

    Fraction of maximum window length at which to start linear fit.

  • fit_stop (float) –

    Fraction of maximum window length at which to stop linear fit.

  • fit_method (str {"siegel", "theilsen", "ls"}, default: 'siegel' ) –

    Linear fit method to use. By default "siegel"

Returns:

  • Self @ Engine –

    Object containing the results and inputs of the diffusion entropy analysis.

Notes

Prefer the siegel or theilsen methods. Least squares linear fits can introduce bias when done over log-scale data, see Clauset, A., Shalizi, C.R. and Newman, M.E., 2009. Power-law distributions in empirical data. SIAM review, 51(4), pp.661-703. https://doi.org/10.1137/070710111. https://arxiv.org/pdf/0706.1062.pdf.

Source code in src/pymdea/core.py 
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
def analyze_without_stripes(
    self: Self,
    fit_start: int,
    fit_stop: int,
    fit_method: Literal["siegel", "theilsen", "ls"] = "siegel",
) -> Self:
    """Run a regular diffusion entropy analysis.

    Parameters
    ----------
    fit_start : float
        Fraction of maximum window length at which to start linear fit.
    fit_stop : float
        Fraction of maximum window length at which to stop linear fit.
    fit_method : str {"siegel", "theilsen", "ls"}, optional
        Linear fit method to use. By default "siegel"

    Returns
    -------
    Self @ Engine
        Object containing the results and inputs of the diffusion
        entropy analysis.

    Notes
    -----
    Prefer the siegel or theilsen methods. Least squares linear
    fits can introduce bias when done over log-scale data, see
    Clauset, A., Shalizi, C.R. and Newman, M.E., 2009. Power-law
    distributions in empirical data. SIAM review, 51(4), pp.661-703.
    https://doi.org/10.1137/070710111.
    https://arxiv.org/pdf/0706.1062.pdf.

    """
    self.trajectory = self.data
    self.fit_start = fit_start
    self.fit_stop = fit_stop
    self.fit_method = fit_method
    self._calculate_entropy()
    self._calculate_scaling()
    self._calculate_mu()
    self.print_result()

print_result()

Print out result of analysis.

Source code in src/pymdea/core.py 
247
248
249
250
251
252
253
254
255
256
def print_result(self: Self) -> str:
    """Print out result of analysis."""
    self.result = "--------------------------------- \n"
    self.result = self.result + "result \n"
    self.result = self.result + f" δ: {self.delta} \n"
    self.result = self.result + f" μ (rule 1): {self.mu1} \n"
    self.result = self.result + f" μ (rule 2): {self.mu2} \n"
    self.result = self.result + "---------------------------------"
    print(self.result)  # noqa: T201
    return self

DeaLoader()

Load data for a diffusion entropy analysis.

Source code in src/pymdea/core.py 
24
25
def __init__(self: Self) -> Self:
    """Load data for a diffusion entropy analysis."""

make_diffusion_process(kind='gn', length=10000, a=0)

Generate diffusion process data.

Parameters:

  • kind (str {"cn", "gn", "fgn", "fbm"}, default: "cn" ) –

    Type of diffusion noise to generate. If "cn", generate a colored noise with spectral power a. If "gn", generate a Gaussian noise. If "fgn", generate a fractional Gaussian noise with Hurst index H = a. If "fbm", generate a fractional Brownian motion with Hurst index H=a.

  • length (int, default: 10000 ) –

    Length of time-series to generate.

  • a ((float, optiona), default: 0 ) –

    Only used if kind is "fgn", "fbm", or "cn". If kind is "fgn" or "fbm", this sets the Hurst index of the process. If kind is "cn" this sets the index of the power law spectrum for the noise, 1/(f^a).

Returns:

  • Self @ Loader –

    An instance of the Loader object.

Source code in src/pymdea/core.py 
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
81
82
83
84
85
86
87
88
89
90
91
92
def make_diffusion_process(
    self: Self,
    kind: Literal["cn", "gn", "fgn", "fbm"] = "gn",
    length: int = 10000,
    a: float = 0,
) -> Self:
    """Generate diffusion process data.

    Parameters
    ----------
    kind : str {"cn", "gn", "fgn", "fbm"}, optional, default "cn"
        Type of diffusion noise to generate. If "cn", generate a
        colored noise with spectral power `a`. If "gn", generate a
        Gaussian noise. If "fgn", generate a fractional Gaussian
        noise with Hurst index H = `a`. If "fbm", generate a
        fractional Brownian motion with Hurst index H=`a`.
    length : int, optional, default 10000
        Length of time-series to generate.
    a : float, optiona, default 0
        Only used if `kind` is "fgn", "fbm", or "cn". If `kind` is
        "fgn" or "fbm", this sets the Hurst index of the process.
        If `kind` is "cn" this sets the index of the power law
        spectrum for the noise, 1/(f^`a`).

    Returns
    -------
    Self @ Loader
        An instance of the Loader object.

    """
    if kind == "gn":
        process = stochastic.processes.noise.GaussianNoise()
        self.data = process.sample(length)
    if kind == "cn":
        process = stochastic.processes.noise.ColoredNoise(beta=a)
        self.data = process.sample(length)
    if kind == "fgn":
        process = stochastic.processes.noise.FractionalGaussianNoise(hurst=a)
        self.data = process.sample(length)
    if kind == "fbm":
        process = stochastic.processes.continuous.FractionalBrownianMotion(hurst=a)
        self.data = process.sample(length)
    return self

make_sample_data(length=100000, seed=1)

Generate an array of sample data.

Parameters:

  • length (int, default: 100000 ) –

    Number of time-steps to produce in the sample data.

  • seed (int, default: 1 ) –

    Seed for random number generation.

Returns:

  • Self @ Loader –

    An instance of the Loader object.

Source code in src/pymdea/core.py 
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
def make_sample_data(self: Self, length: int = 100000, seed: int = 1) -> np.ndarray:
    """Generate an array of sample data.

    Parameters
    ----------
    length : int, optional, default: 100000
        Number of time-steps to produce in the sample data.
    seed : int, optional, default: 1
        Seed for random number generation.

    Returns
    -------
    Self @ Loader
        An instance of the Loader object.

    """
    rng = np.random.default_rng(seed=seed)
    random_walk = rng.choice([-1, 1], size=length).cumsum()
    random_walk[0] = 0  # always start from 0
    self.seed = seed
    self.data = random_walk
    return self

read_data_file(filepath, column_name)

Read input data from file.

Parameters:

  • filepath (str) –

    System path to a file containing data. Must include the full file name, including the extension. Example: "/example/path/to/file.csv"

  • column_name (str) –

    Name of the column in the data file which contains the time series data values.

Returns:

  • Self @ Loader –

    An instance of the Loader object.

Raises:

  • ValueError –

    If filepath points to a file of type other than CSV. Support for more types of files is a work in progress.

Source code in src/pymdea/core.py 
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
def read_data_file(self: Self, filepath: str, column_name: str) -> pl.DataFrame:
    """Read input data from file.

    Parameters
    ----------
    filepath : str
        System path to a file containing data. Must include the
        full file name, including the extension. Example:
        "/example/path/to/file.csv"
    column_name : str
        Name of the column in the data file which contains the time
        series data values.

    Returns
    -------
    Self @ Loader
        An instance of the Loader object.

    Raises
    ------
    ValueError
        If filepath points to a file of type other than
        CSV. Support for more types of files is a work in
        progress.

    """
    filepath = Path(filepath)
    supported_types = [".csv"]
    if filepath.suffix not in supported_types:
        msg = f"filetype must be one of: {supported_types}."
        raise ValueError(msg)
    if filepath.suffix == ".csv":
        self.data = pl.scan_csv(filepath).select(column_name).to_numpy()
    return self