diff --git a/code/examples_pyYggdrasil/notebooks/pyYggdrasilDemo_for_presentation.ipynb b/code/examples_pyYggdrasil/notebooks/pyYggdrasilDemo_for_presentation.ipynb
index a560df6ef366d7156800e07c36a3bc80a856da21..efd8d1e5a89ea01c75f4cdfa61a356390055b88b 100644
--- a/code/examples_pyYggdrasil/notebooks/pyYggdrasilDemo_for_presentation.ipynb
+++ b/code/examples_pyYggdrasil/notebooks/pyYggdrasilDemo_for_presentation.ipynb
@@ -14,7 +14,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
@@ -59,7 +59,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
@@ -117,7 +117,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
@@ -151,9 +151,1595 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 4,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Help on module pyYggdrasil.Random in pyYggdrasil:\n",
+ "\n",
+ "NAME\n",
+ " pyYggdrasil.Random - This submodule contains the distribution generators that are offered by yggdrasil. They operrate directly on the sample containers. Thus offer a more direct way of creating samples, than using the numpy distributions.\n",
+ "\n",
+ "CLASSES\n",
+ " pybind11_builtins.pybind11_object(builtins.object)\n",
+ " MultiCondition\n",
+ " RNDistribution\n",
+ " Beta\n",
+ " BiUniform\n",
+ " Dirichlet\n",
+ " Gamma\n",
+ " Gauss\n",
+ " MultiModal\n",
+ " Outlier\n",
+ " Uniform\n",
+ " SingleCondition\n",
+ " TreeSampler\n",
+ " eDistiType\n",
+ " pRNG\n",
+ " \n",
+ " class Beta(RNDistribution)\n",
+ " | This is the beta distribution. It is described by two parameters, Alpha and Beta. It can also be restricted to a certain part of the domain.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Beta\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Beta, Alpha: float, Beta: float, domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constructs a beta distribution, that is described by the parameters Alpha and Beta. The doamain restricts the sampling to a certain part. It can be invalid in that case the unit interval is used. Some checks are applied to see if the distribution is correct.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Beta, Alpha: float, Beta: float) -> None\n",
+ " | \n",
+ " | This constructor constructs a beta distribution, that is described by the parameters Alpha and Beta. This constructor uses the unit interval as sampling domain.\n",
+ " | \n",
+ " | getAlpha(...)\n",
+ " | getAlpha(self: pyYggdrasil.Random.Beta) -> float\n",
+ " | \n",
+ " | This function returns the Alpha parameter of the distribution.\n",
+ " | \n",
+ " | getBeta(...)\n",
+ " | getBeta(self: pyYggdrasil.Random.Beta) -> float\n",
+ " | \n",
+ " | This function returns the Beta value of the distribution.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class BiUniform(RNDistribution)\n",
+ " | This class is the implementation of a bimodal (one dimensional) distribution. It has two supports domain and and a coupling factor alpha, that describs the strength of the first mode. In the paper this distribution was known as spiky uniform.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | BiUniform\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.BiUniform, suppDom1: pyYggdrasil.Interval, suppDom2: pyYggdrasil.Interval, alpha: float, domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constructs the uniform distribution. It takes two intrvals (suppDom1 and subDom2) which describes the support of the two modes. It also takes a copupling factor alpha, which describes the strength of the first mode. It also accepts a domain, which must be a one dimensional hypercube, that serves as sampling domain. This must be valid, as for all uniform distribution, it is kind of useless. It is advised to use the other constructor.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.BiUniform, suppDom1: pyYggdrasil.Interval, suppDom2: pyYggdrasil.Interval, alpha: float) -> None\n",
+ " | \n",
+ " | This constructor constructs the uniform distribution. It takes two intrvals (suppDom1 and subDom2) which describes the support of the two modes. It also takes a copupling factor alpha, which describes the strength of the first mode. This is the constructor that does not take a sampling domain argument. The sampling domain is estimated from the two modes.\n",
+ " | \n",
+ " | getAlpha(...)\n",
+ " | getAlpha(self: pyYggdrasil.Random.BiUniform) -> float\n",
+ " | \n",
+ " | This function returns the coupling value alpha.\n",
+ " | \n",
+ " | getSuppDom1(...)\n",
+ " | getSuppDom1(self: pyYggdrasil.Random.BiUniform) -> pyYggdrasil.Interval\n",
+ " | \n",
+ " | This function returns the support domain of the first mode.\n",
+ " | \n",
+ " | getSuppDom2(...)\n",
+ " | getSuppDom2(self: pyYggdrasil.Random.BiUniform) -> pyYggdrasil.Interval\n",
+ " | \n",
+ " | This function returns the support domain of the second mode.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class Dirichlet(RNDistribution)\n",
+ " | This class implements the Dirichlet distribution. The class follows the convention of the paper and Mathematica, note that for example Wikipedia uses a different convention. A K-1 dimensional distribution is described by K many parameters, \\alpha_i > 0 for all i = 1, ... , K, parameters. Furter we requier that the sample vector x, x_i > 0 for all i = 1, ..., K-1, and further \\sum_{i = 1}^{K - 1} x_i < 1, has to hold. By default the sampling domain is uded as the unit hyper cube of the coresponding dimension.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Dirichlet\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Dirichlet, alpha: numpy.ndarray[float64[m, 1]], domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constructs a dirichlet distribution with the given parameter. Some checks on the validity of the parameter is performed. Note that the dimension of the hyper cube domain, must be one less than the one of the parameter vector. Also note that if this constructor is used, the domain must be valid, this is the _only_ requierement that is imposed. This means that it is not checked if the domain actually contains the support domain. This is generally true for the distributions with finite support.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Dirichlet, alpha: numpy.ndarray[float64[m, 1]]) -> None\n",
+ " | \n",
+ " | This constructor constructs a dirichlet distribution. If the parameter vector alpha has K komponents a K - 1 dimensional distribution is constructed. The sample domain that is used is the unit hyper cube, which is also the full supported domain.\n",
+ " | \n",
+ " | getAlpha(...)\n",
+ " | getAlpha(self: pyYggdrasil.Random.Dirichlet) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion returns a copy of the parameter vector that was used to construct the distribution.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class Gamma(RNDistribution)\n",
+ " | This is the class that generates gamma distributed samples. As every one dimensional distribution that interface seams a bit odd. There are many convention about this distribution. We use the one that is also usedby the C++ standard, which calls Alpha the shape and Beta the scale. This class also supports the sampling domain restriction.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Gamma\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Gamma, Alpha: float, Beta: float, domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constructs a gamma distribution object. The scale parameter of the distribution is set to Alpha. The rate parameter is set to Beta. If the domain is given the sampling is restricted to it. If invalid no restriction is done.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Gamma, Alpha: float, Beta: float) -> None\n",
+ " | \n",
+ " | This constructor constructs a gamma distribution object. The scale parameter of the distribution is set to Alpha. The rate parameter is set to Beta. The domain parameter is implicitly set to the invalid domain, thus no restriction is applied.\n",
+ " | \n",
+ " | getAlpha(...)\n",
+ " | getAlpha(self: pyYggdrasil.Random.Gamma) -> float\n",
+ " | \n",
+ " | This fucntion returns the shape value, that is also known as Alpha, of the distribution.\n",
+ " | \n",
+ " | getBeta(...)\n",
+ " | getBeta(self: pyYggdrasil.Random.Gamma) -> float\n",
+ " | \n",
+ " | This function returns the rate value, that is also known as Beta, of the distribution.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class Gauss(RNDistribution)\n",
+ " | This class is a implementaion of the distribution interface. It implements a multi dimensional Gauss distribution. It uses Eigen matrices, which should be compatible to numpy matrices.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Gauss\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Gauss, mu: numpy.ndarray[float64[m, 1]], Sigma: numpy.ndarray[float64[m, n]], domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constructs a Gaussian distribution, with has the median mu and the covariance matrix Sigma, which is something like the VARIANCE (this is different from many other interface, also from the C++ standard. There are no direct check if Sigma is symmertric, it is feeded to the Cholescky decomposition routine of Eigen, if it is fine with it, then its good. Not that only one half of the matrix is used. The sampling is restricted to the hyper cube hc. If it is invalid, then no restrictions are applied. The class works with Eigen types, they will be automatically converted from numpy arrays.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Gauss, mu: numpy.ndarray[float64[m, 1]], Sigma: numpy.ndarray[float64[m, n]]) -> None\n",
+ " | \n",
+ " | This constructor constructs a Gauss distribution. The sampling domain is invalid, which means that there are no restrictions applied. For a discussion of the arguments see the other constructor.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class MultiCondition(pybind11_builtins.pybind11_object)\n",
+ " | This class represents many conditions at once. It can be seen as a list of SingleCondition objects. It is needed to instruct the sampler to restrict certain dimensions. There are several ways for constructing it.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | MultiCondition\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.MultiCondition, nDims: int) -> None\n",
+ " | \n",
+ " | This function constructs a MultiCondition object that does not have any restrictions. However it needs the dimension of the underling sample space to be passed to it. This is needed for internal operations.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.MultiCondition, nDims: int, cMap: Dict[int, float]) -> None\n",
+ " | \n",
+ " | This constructor constructs the MultiCondition from a map. The map is interpreted that the key of a record is interpreted as the dimension in which the condition should be applied. The associated value of the key is the value that is used for restriction. The value is interpreted on the global/original data space. Also the dimension of the underling sample space, nDims, has to be passed.\n",
+ " | \n",
+ " | 3. __init__(self: pyYggdrasil.Random.MultiCondition, nDims: int, cVec: List[pyYggdrasil.Random.SingleCondition]) -> None\n",
+ " | \n",
+ " | This constructor constructs a MultiCondition out of a list of several SingleConditions. The values are interpreted on the global/original data space. Also the dimension of the underling sample space, nDims, has to be passed. The order of the condition is irrelevant.\n",
+ " | \n",
+ " | 4. __init__(self: pyYggdrasil.Random.MultiCondition, src: pyYggdrasil.Random.MultiCondition) -> None\n",
+ " | \n",
+ " | This constructor performs a copy construction, by deep coping src and stores it in *this.\n",
+ " | \n",
+ " | __repr__(...)\n",
+ " | __repr__(self: pyYggdrasil.Random.MultiCondition) -> str\n",
+ " | \n",
+ " | This function transforms *this in a textual representation. The format is the same as when using __str__.\n",
+ " | \n",
+ " | __str__(...)\n",
+ " | __str__(self: pyYggdrasil.Random.MultiCondition) -> str\n",
+ " | \n",
+ " | This fucntion transforms the MultiCollection into a string. The format is {COND1, ... | nDims}, where COND1 is the result of a print operation on the underling SingleCondition object. nDims is the dimension of the underling sample sapce.\n",
+ " | \n",
+ " | applyConditions(...)\n",
+ " | applyConditions(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. applyConditions(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.Sample) -> None\n",
+ " | \n",
+ " | This function applies the conditions of *this to s. The values of the components of s that names restricted dimensions are set to the restricting values. This function is equivalent to first restricting the sample and then expanding the result, but more efficient.\n",
+ " | \n",
+ " | 2. applyConditions(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleList) -> None\n",
+ " | \n",
+ " | This function applies the conditions of *this to all samples inside s. The values of the components of the samples that names restricted dimensions are set to the restricting values. This function is equivalent to first restricting the container and then expanding the result, but more efficient.\n",
+ " | \n",
+ " | 3. applyConditions(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleArray) -> None\n",
+ " | \n",
+ " | This function applies the conditions of *this to all samples inside s. The values of the components of the samples that names restricted dimensions are set to the restricting values. This function is equivalent to first restricting the container and then expanding the result, but more efficient.\n",
+ " | \n",
+ " | 4. applyConditions(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleCollection) -> None\n",
+ " | \n",
+ " | This function applies the conditions of *this to all samples inside s. The values of the components of the samples that names restricted dimensions are set to the restricting values. This function is equivalent to first restricting the container and then expanding the result, but more efficient.\n",
+ " | \n",
+ " | expandSample(...)\n",
+ " | expandSample(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. expandSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.Sample) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function expands the (restricted) sample s to its full dimension and applies the condition to it. The sample must have dimension nDOF(), the returned sample will have dimensin nDims().\n",
+ " | \n",
+ " | 2. expandSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleList) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | This function expands all (restricted) samples that are stored in the sample list s to their full dimension and applies the condition to them. The samples in the sample list must have dimension nDOF(), the samples list that is returned will have dimension nDims().\n",
+ " | \n",
+ " | 3. expandSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleArray) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | This function expands all (restricted) samples that are stored in the sample array s to their full dimension and applies the condition to them. The sample array must have dimension nDOF(), the returned sample array will have dimension nDims().\n",
+ " | \n",
+ " | 4. expandSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleCollection) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | This function expands all (restricted) samples that are stored in the sample collection s to their full dimension and applies the condition to them. The sample collection must have dimension nDOF(), the returned sample collection will have dimension nDims().\n",
+ " | \n",
+ " | getConditionFor(...)\n",
+ " | getConditionFor(self: pyYggdrasil.Random.MultiCondition, arg0: int) -> pyYggdrasil.Random.SingleCondition\n",
+ " | \n",
+ " | This function returns the condition that is applied to dimension dim. If this dimension does not have a condition that is applied to it and exception is generated.\n",
+ " | \n",
+ " | getConditionMap(...)\n",
+ " | getConditionMap(self: pyYggdrasil.Random.MultiCondition) -> Dict[int, float]\n",
+ " | \n",
+ " | This function returns the underling conditions inside a map/dict. The format is such that the keys are the dimensions of the conditions, and the associated values are the values of the restriction. The conditions are expressed in tzhe original data space.\n",
+ " | \n",
+ " | getConditionVector(...)\n",
+ " | getConditionVector(self: pyYggdrasil.Random.MultiCondition) -> List[pyYggdrasil.Random.SingleCondition]\n",
+ " | \n",
+ " | This function returns the underling conditions inside a vector. The order is unspecific. The condition is expressed in the original data space.\n",
+ " | \n",
+ " | getConditionedDimensions(...)\n",
+ " | getConditionedDimensions(self: pyYggdrasil.Random.MultiCondition) -> List[int]\n",
+ " | \n",
+ " | This function returns the dimensions in which we apply conditions to. The difference between this and the getConditionVector() is, that this function only returns the dimensions and not the values.\n",
+ " | \n",
+ " | getFreeDimensions(...)\n",
+ " | getFreeDimensions(self: pyYggdrasil.Random.MultiCondition) -> List[int]\n",
+ " | \n",
+ " | This functions returns the free dimensions of the multi conditions. The free dimensions are all dimensions where no restriction is applied to. This means there are nDOF() many of them.\n",
+ " | \n",
+ " | hasConditionFor(...)\n",
+ " | hasConditionFor(self: pyYggdrasil.Random.MultiCondition, dim: int) -> bool\n",
+ " | \n",
+ " | The function returns true if *this has a condition that is applied to dimension dim. Note that if dim is larger (or equal, since we have zero based indexing) an out of bound exception is generated.\n",
+ " | \n",
+ " | isValid(...)\n",
+ " | isValid(self: pyYggdrasil.Random.MultiCondition) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is valid. This should always be the case and if an invalid condition is found then there are trubles.\n",
+ " | \n",
+ " | nConditions(...)\n",
+ " | nConditions(self: pyYggdrasil.Random.MultiCondition) -> int\n",
+ " | \n",
+ " | This function returns the number of constions that *this has. Note that 0 indicates that no condition is in place.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.MultiCondition) -> int\n",
+ " | \n",
+ " | Return the dimension of the underling sample space.\n",
+ " | \n",
+ " | noRestrictions(...)\n",
+ " | noRestrictions(self: pyYggdrasil.Random.MultiCondition) -> bool\n",
+ " | \n",
+ " | This function returns true if *this does not have any restriction. It is an efficient test for nConditions == 0. Note on C++ this fucntion is called noConditions(), but it was renamed on the interface to minimize confusion.\n",
+ " | \n",
+ " | restrictSample(...)\n",
+ " | restrictSample(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. restrictSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.Sample) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function restricts the sample s, by removing all dimensions that are conditioned. The sample must have dimension nDims() and the returned sample will have dimension nDOF().\n",
+ " | \n",
+ " | 2. restrictSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleList) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | This function restricts all samples that are stored in the sample list s, by removing all dimensions that are conditioned. The sample list must have dimension nDims(), the sample list that is returned will have dimension nDOF().\n",
+ " | \n",
+ " | 3. restrictSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleArray) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | This function restricts all samples that are stored in the sample array s, by removing all dimensions that are conditioned. The sample array must have dimension nDims(), the returned sample array will have dimension nDOF().\n",
+ " | \n",
+ " | 4. restrictSample(self: pyYggdrasil.Random.MultiCondition, s: pyYggdrasil.SampleCollection) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | This function restricts all samples that are stored in the sample collection s, by removing all dimensions that are conditioned. The sample collection must have dimension nDims(), the returned sample collection will have dimension nDOF().\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class MultiModal(RNDistribution)\n",
+ " | This is a class for a multi modal distribution. A multimodal distribution is a linear combination of different other distributon. It is simmilar to the BiUniform and the outlier distribution, but way more general. It supports an arbitrary number of and kind of distribution.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | MultiModal\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(self, /, *args, **kwargs)\n",
+ " | Initialize self. See help(type(self)) for accurate signature.\n",
+ " | \n",
+ " | getWeights(...)\n",
+ " | getWeights(self: pyYggdrasil.Random.MultiModal) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the weights of the modes. The type is an Eigen Vector, thus it will be converted to a numpy array automatically.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class Outlier(RNDistribution)\n",
+ " | This is an outlier distribution, it is basically a bimodal one dimensional gaussian distribution. Both modes have the same mean, but different standard deviantions. The parameter alpha, that must be inside the range 0 to 1, describes the strength of the first mode.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Outlier\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Outlier, alpha: float, mean: float, sigma1: float, sigma2: float, domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This is the full constructor, which takes the alpha the strength of the first mode, the mean of both modes. It also takes the standard deviation of the first mode sigma1 and the standard deviation of the second mode, sigma2. Note that the other Gauss distribution takes the variance instead. As last argument A hypercube, which must have dimension one is used. The sampling generation is restricted to this domain.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Outlier, alpha: float, mean: float, sigma1: float, sigma2: float) -> None\n",
+ " | \n",
+ " | This is the full constructor, which takes the alpha the strength of the first mode, the mean of both modes. It also takes the standard deviation of the first mode sigma1 and the standard deviation of the second mode, sigma2. Note that the other Gauss distribution takes the variance instead. This version does nto take a domain argument, thus the sampling is not restricted.\n",
+ " | \n",
+ " | getAlpha(...)\n",
+ " | getAlpha(self: pyYggdrasil.Random.Outlier) -> float\n",
+ " | \n",
+ " | This fucntion returns the coupling parameter alpha, which describes the influence of the first mode.\n",
+ " | \n",
+ " | getStdDev1(...)\n",
+ " | getStdDev1(self: pyYggdrasil.Random.Outlier) -> float\n",
+ " | \n",
+ " | This function returns the standard deviation of the first mode. Note that unlike the getCoVar() function, which returns a matrix, this function returns a scalar.\n",
+ " | \n",
+ " | getStdDev2(...)\n",
+ " | getStdDev2(self: pyYggdrasil.Random.Outlier) -> float\n",
+ " | \n",
+ " | This function returns the standard deviation of the second mode. Note that unlike the getCoVar() function, which returns a matrix, this function returns a scalar.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class RNDistribution(pybind11_builtins.pybind11_object)\n",
+ " | This is the base class of all random distribution inside Yggdrasil. All of them operate on the random number generator object that is provided by pyYggdrasil. The distribution used inside Yggdrasil are pretty much what you expect. However they have an additional property, the 'sampleing domain'. In a sense this is the domain where samples will be generated. This is usefull if you want to restrict the sampling in different regions. The pdf ios not addaped to map that change, this means that when you integrate the PDF of that distribution, will not neccessaraly result in one. An invalid hyper cube means that no restriction is applied. All implementation supports a construction whiout a domain. In that case a good domain is selected, there is no guarantee if the domain is valid or not. The Vector this class returns are Eigen types, thus they will be transaprently mapped to NumPy Arrays.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(self, /, *args, **kwargs)\n",
+ " | Initialize self. See help(type(self)) for accurate signature.\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class SingleCondition(pybind11_builtins.pybind11_object)\n",
+ " | This class represents one condition. A condition is pair of an integer, which represent the dimension where this condition should be applied to. And a value, the value that should be used for that restriction. This class is not used directly but is unsed as a building block, similar to the interval.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | SingleCondition\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(self: pyYggdrasil.Random.SingleCondition, dim: int, value: float) -> None\n",
+ " | \n",
+ " | This is the building constructor, it takes a dimension, dim and the value that the conditin should have. The value is interpreted on the original/global data space.\n",
+ " | \n",
+ " | __repr__(...)\n",
+ " | __repr__(self: pyYggdrasil.Random.SingleCondition) -> str\n",
+ " | \n",
+ " | Generates a textual represenation of this, is the same function as __str__. It outputs a string with (dim: val), where dim is the dimension and val is the value that is applied.\n",
+ " | \n",
+ " | __str__(...)\n",
+ " | __str__(self: pyYggdrasil.Random.SingleCondition) -> str\n",
+ " | \n",
+ " | This function produces a textual representation of this. It outputs a string with (dim: val), where dim is the dimension and val is the value that is applied.\n",
+ " | \n",
+ " | getDim(...)\n",
+ " | getDim(self: pyYggdrasil.Random.SingleCondition) -> int\n",
+ " | \n",
+ " | This fucntion returns the dimension in which the condition should be applied to.\n",
+ " | \n",
+ " | getValue(...)\n",
+ " | getValue(self: pyYggdrasil.Random.SingleCondition) -> float\n",
+ " | \n",
+ " | This function returns the value that the condition has.\n",
+ " | \n",
+ " | isValid(...)\n",
+ " | isValid(self: pyYggdrasil.Random.SingleCondition) -> bool\n",
+ " | \n",
+ " | This function returns true id *this is valid. In Yggdrasil invalid conditions could occure, but they will not surface in pyYggdrasil. This fucntion should return always true, if not am internal error has manifested.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class TreeSampler(pybind11_builtins.pybind11_object)\n",
+ " | This class allows the sampling from a valid fitted tree. It processed the tree and copies the important/needed part, thus the tree is not needed to exist longer. There are some limitations in the implementation. For example the structure that is generated an stored inside *this is not suited to compute pdf values. The pdf of samples that are generated by this can be computed as a byproduct of the sampling, but no pdf of a sample can be computed, for this the tree is needed. Hower it supports the same sampling functions as the random distributions does. Note that the probability that the generating functions returns are conditioned pdfs. This means that the probability that is returned is not the same as the one computed by the tree, in the case of conditioned samples.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | TreeSampler\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.TreeSampler, tree: pyYggdrasil.DETree) -> None\n",
+ " | \n",
+ " | This constructor will construct a sampler that is not restricted. This means unconditional sampling is applied.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.TreeSampler, tree: pyYggdrasil.DETree, condi: pyYggdrasil.Random.MultiCondition) -> None\n",
+ " | \n",
+ " | This constructor will build a tree sampler that is restricted as described by the MultiCondition condi. This is the canonical way of generating a conditioned sampler.\n",
+ " | \n",
+ " | 3. __init__(self: pyYggdrasil.Random.TreeSampler, tree: pyYggdrasil.DETree, cVector: List[pyYggdrasil.Random.SingleCondition]) -> None\n",
+ " | \n",
+ " | This will create a tree sampler out of a tree and a condition vector. Internaly a MultiCondition is constructed using the supplied cVector and then the sampler is constructed. This constructor is provided for convenience.\n",
+ " | \n",
+ " | 4. __init__(self: pyYggdrasil.Random.TreeSampler, tree: pyYggdrasil.DETree, cMap: Dict[int, float]) -> None\n",
+ " | \n",
+ " | This will create a conditioned tree sampler. The conditions are extracted from the provided map/dict in the same way as a MultiCondition would be constructed.\n",
+ " | \n",
+ " | 5. __init__(self: pyYggdrasil.Random.TreeSampler, src: pyYggdrasil.Random.TreeSampler) -> None\n",
+ " | \n",
+ " | This will construct a tree sampler by deep coping src.\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates a single sample from the tree sampler. geni is used to generate the needed randomness.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | This function generates N samples and store them in a SampleArray container. The randomness is drawn from the geni object.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function generates N samples and also its associatd pdf values, they are stored in a pair. The first element of the returned pair is a SampleArray, where the samples are stored. The second element is an Eigen vector, that can be used like a NumPy array, that contains the pdf values. The randomness is drawn from geni.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | This function generates N samples and stores them inside a SampleCollection. The randomness is generated by geni.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function generates N samples, using geni as random source and the associated pdf values of the generated samples. The function returns a pair, the first element contains the SampleCollection, where the samplesa are stored in. The second element is a NumPy compatible array that contains the probabilities of the samples.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | This function generates N samples and stores them in a SampleList object. The randomness generated by geni.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function generates N samples and their associated pdf values. The fucntion returns a pair, its first element is a SampleList that stores the generated samples. The second element is a NumPy vector with the pdf values. The randomness that is needed is generated by geni.\n",
+ " | \n",
+ " | generateSamplesPDF(...)\n",
+ " | generateSamplesPDF(self: pyYggdrasil.Random.TreeSampler, g: pyYggdrasil.Random.pRNG) -> Tuple[pyYggdrasil.Sample, float]\n",
+ " | \n",
+ " | This function generates a pair. first is the generated sample and second is the pdf of that sample. The randomness is drawn from geni.\n",
+ " | \n",
+ " | getConditions(...)\n",
+ " | getConditions(self: pyYggdrasil.Random.TreeSampler) -> pyYggdrasil.Random.MultiCondition\n",
+ " | \n",
+ " | This fucntion returns the conditions that where used to select the leafes that enters the tree. Note that this is not what was passed upon the construction to the sampler. The conditions that were passed to the constructor, is mapped to the root data space and stored. This function will first apply a transformation to get back to the original data space. If *this is over constrained an error will be generated.\n",
+ " | \n",
+ " | getGlobalDataSpace(...)\n",
+ " | getGlobalDataSpace(self: pyYggdrasil.Random.TreeSampler) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the domain where the tree is defined on, also known as the global/original data space. This variable was copied from the tree during the constuction.\n",
+ " | \n",
+ " | getInvPxC(...)\n",
+ " | getInvPxC(self: pyYggdrasil.Random.TreeSampler) -> float\n",
+ " | \n",
+ " | This function will return the scalling constant of the probability. This factor is used to transform unconditioned probability to conditioned probability.\n",
+ " | \n",
+ " | getProxyDomain(...)\n",
+ " | getProxyDomain(self: pyYggdrasil.Random.TreeSampler, i: int) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This function will return the hyper cube of the ith leaf. There are two very important things here. First the domain is relative to the root data space. Second the order of the proxy is arbitraraly, but fix during the lifetime of *this. The index ranges from 0 to nProxies() - 1.\n",
+ " | \n",
+ " | nConditions(...)\n",
+ " | nConditions(self: pyYggdrasil.Random.TreeSampler) -> int\n",
+ " | \n",
+ " | This function returns the number of conditions that are used in the sampler.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.TreeSampler) -> int\n",
+ " | \n",
+ " | This function returns the dimension of the underlying sample space.\n",
+ " | \n",
+ " | nProxy(...)\n",
+ " | nProxy(self: pyYggdrasil.Random.TreeSampler) -> int\n",
+ " | \n",
+ " | This function returns the number of leafs that where able to meet the conditions. The name Proxy comes from the C++ side and does not bear any meaning in pyYggdrasil.\n",
+ " | \n",
+ " | noRestrictions(...)\n",
+ " | noRestrictions(self: pyYggdrasil.Random.TreeSampler) -> bool\n",
+ " | \n",
+ " | This function returns true, the sampler does not have any conditions that are applied. This is like testing if nConditions() == 0, but is a bit more efficient.\n",
+ " | \n",
+ " | notOverConstrained(...)\n",
+ " | notOverConstrained(self: pyYggdrasil.Random.TreeSampler) -> bool\n",
+ " | \n",
+ " | This fucntion returns true if at least some leaf in the tree was able to meet the condition that was given upon constructing. This function is used to indicate that the sampler can be used. If this function returns false, then some of the fucntion will genertae errors. This fucntion is like testing nProxy() != 0, but more efficient.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class Uniform(RNDistribution)\n",
+ " | This class allows to sample multidimensional uniform distributions. Note that all dimensions are independend form each other. This class is mainly described by the support domain, this is the domain in which samples are generated. This class also supports the concept of a sampling domain. For historical reaons the two of them needs not to be the same, however it is strongly recomended to not exploit that future.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Uniform\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Uniform, suppDom: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constuct a multidimensional uniform distribution on the domain described by suppDom. This is the domain where samples can be drawn. All dimensions are independend from each other.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Uniform, suppDom: pyYggdrasil.HyperCube, domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This is a constructor that takes as first argument the support domain, suppDom, meaning is the same as in the other supported constructor. As second argument it takes another domain, the sampling domain, which is called for considtency simply domain, which needs to be at least as big as the support domain, it can be invalid. In this setting it does not cary any meaning, it is only provided for consistency reasons. It is strongly advised to not use it.\n",
+ " | \n",
+ " | getSupportDomain(...)\n",
+ " | getSupportDomain(self: pyYggdrasil.Random.Uniform) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This function returns the support domain. This is the first argument of the constructor.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class eDistiType(pybind11_builtins.pybind11_object)\n",
+ " | This is an enum to encode the kind of distribution should be generated, when a canonical distribution is requested.\n",
+ " | \n",
+ " | Members:\n",
+ " | \n",
+ " | NoDistri : This encodes no distribution, this is a useless value.\n",
+ " | \n",
+ " | Gauss : This is value for requesting Gaussian distributions.\n",
+ " | \n",
+ " | Dirichlet : Type for encodeing Dirichlet distributed samples (convention of the paper and Mathematica).\n",
+ " | \n",
+ " | Outlier : Type for encouding an outlier distribution. An outlier distribution is a binomial Gauss distrbution, where both Gauss distributions have the same mean.\n",
+ " | \n",
+ " | Uniform : Type for encoding uniform distributions.\n",
+ " | \n",
+ " | Gamma : Type for encoding a gamma distribution.\n",
+ " | \n",
+ " | Beta : Type for encoding Beta distributions.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | eDistiType\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __eq__ = (...)\n",
+ " | (self: object, arg0: object) -> bool\n",
+ " | \n",
+ " | __getstate__ = (...)\n",
+ " | (self: object) -> int_\n",
+ " | \n",
+ " | __hash__ = (...)\n",
+ " | (self: object) -> int_\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(self: pyYggdrasil.Random.eDistiType, arg0: int) -> None\n",
+ " | \n",
+ " | __int__(...)\n",
+ " | __int__(self: pyYggdrasil.Random.eDistiType) -> int\n",
+ " | \n",
+ " | __ne__ = (...)\n",
+ " | (self: object, arg0: object) -> bool\n",
+ " | \n",
+ " | __repr__ = (...)\n",
+ " | (self: handle) -> str\n",
+ " | \n",
+ " | __setstate__ = (...)\n",
+ " | (self: pyYggdrasil.Random.eDistiType, arg0: int) -> None\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Data descriptors defined here:\n",
+ " | \n",
+ " | __members__\n",
+ " | \n",
+ " | name\n",
+ " | (self: handle) -> str\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Data and other attributes defined here:\n",
+ " | \n",
+ " | Beta = eDistiType.Beta\n",
+ " | \n",
+ " | Dirichlet = eDistiType.Dirichlet\n",
+ " | \n",
+ " | Gamma = eDistiType.Gamma\n",
+ " | \n",
+ " | Gauss = eDistiType.Gauss\n",
+ " | \n",
+ " | NoDistri = eDistiType.NoDistri\n",
+ " | \n",
+ " | Outlier = eDistiType.Outlier\n",
+ " | \n",
+ " | Uniform = eDistiType.Uniform\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ " \n",
+ " class pRNG(pybind11_builtins.pybind11_object)\n",
+ " | This is the implementation of a (pseudo) random number generator. It is a wrapper arround the Mersen Twister Engine that is implemented by the C++ standard. It is needed to generate random numbers from the Distributions that are offered by py/Yggdrasil.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | pRNG\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.pRNG) -> None\n",
+ " | \n",
+ " | This is the default constructor, it will generate a default seeded random number generator.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.pRNG, seed: int) -> None\n",
+ " | \n",
+ " | This constructor constructs a pRNG and uses the provided seed.\n",
+ " | \n",
+ " | 3. __init__(self: pyYggdrasil.Random.pRNG, seed: int, N: int) -> None\n",
+ " | \n",
+ " | This construct a pRNG and seeds it with the provided seed. But it will advance the internal state of the genrator by N. This is equivalent with first constructing a generator and then calling discard.\n",
+ " | \n",
+ " | discard(...)\n",
+ " | discard(self: pyYggdrasil.Random.pRNG, N: int) -> None\n",
+ " | \n",
+ " | This function advances the internal state of *this by N. It is equivalent by calling the sample fucntion N times, but is more efficient.\n",
+ " | \n",
+ " | sample(...)\n",
+ " | sample(self: pyYggdrasil.Random.pRNG) -> int\n",
+ " | \n",
+ " | This function generate a new sample form the random numnber generation.\n",
+ " | \n",
+ " | seed(...)\n",
+ " | seed(self: pyYggdrasil.Random.pRNG, newSeed: int) -> None\n",
+ " | \n",
+ " | This function reinitalizes *this with the provided seed. This function is as reconstructing *this, but with new seed as argument.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ "\n",
+ "FUNCTIONS\n",
+ " canDistribution(...) method of builtins.PyCapsule instance\n",
+ " canDistribution(dType: pyYggdrasil.Random.eDistiType, dKind: int) -> pyYggdrasil.Random.RNDistribution\n",
+ " \n",
+ " This function returns the canonical distribution. The canonical distributions are the ones that where used in the paper. dType encodes which kind of distribution is requested, this value is of type eDistriType. dKind is used to select the distribution inside the clsss. The intention of dKind was once to encode the dimension, but it has long lost that meaning, also negative values are alloed. See the manual for a list with supported distributions. If the distribution does not exists, an error is generated.\n",
+ " \n",
+ " canDistributionName(...) method of builtins.PyCapsule instance\n",
+ " canDistributionName(dType: pyYggdrasil.Random.eDistiType, dKind: int) -> str\n",
+ " \n",
+ " This fucntion gives a name to the distribution. This name is small and does not contain any funny charactor like space. It can thus be used as filename. See the canDistribution() fucntion for a discussion of the argumenst.\n",
+ " \n",
+ " creatMultiModal(...) method of builtins.PyCapsule instance\n",
+ " creatMultiModal(*args, **kwargs)\n",
+ " Overloaded function.\n",
+ " \n",
+ " 1. creatMultiModal(w1: float, m1: pyYggdrasil.Random.RNDistribution, m2: pyYggdrasil.Random.RNDistribution, domain: pyYggdrasil.HyperCube = The domain that is used to restrict the sample range, if invalid, the domain is estimated.) -> pyYggdrasil.Random.RNDistribution\n",
+ " \n",
+ " This function creates a bimodal distribution out of the given modes.The weight of the first mode is w1, the weight of the second mode is implicitly given by 1-w1. The domain is optional and defaults to the invalid domain. This means that it is tried to estimate a domain from the underling distributions.\n",
+ " \n",
+ " 2. creatMultiModal(w1: float, m1: pyYggdrasil.Random.RNDistribution, w2: float, m2: pyYggdrasil.Random.RNDistribution, m3: pyYggdrasil.Random.RNDistribution, domain: pyYggdrasil.HyperCube = The domain that is used to restrict the sample range, if invalid, the domain is estimated.) -> pyYggdrasil.Random.RNDistribution\n",
+ " \n",
+ " This function creates a trimodal distribution out of the given modes.The weight of the first mode is w1, the weight of the second mode is w2, the third weight is estimated by 1-w1-w2. The domain is optional and defaults to the invalid domain. This means that it is tried to estimate a domain from the underling distributions.\n",
+ " \n",
+ " 3. creatMultiModal(w1: float, m1: pyYggdrasil.Random.RNDistribution, w2: float, m2: pyYggdrasil.Random.RNDistribution, w3: float, m3: pyYggdrasil.Random.RNDistribution, m4: pyYggdrasil.Random.RNDistribution, domain: pyYggdrasil.HyperCube = The domain that is used to restrict the sample range, if invalid, the domain is estimated.) -> pyYggdrasil.Random.RNDistribution\n",
+ " \n",
+ " This function creates a distribution that is composed out of four given modes. The last weight is given implicitly such that the weights sum up to one. The domain is optional and defaults to the invalid domain. This means that it is tried to estimate a domain from the underling distributions.\n",
+ " \n",
+ " 4. creatMultiModal(w1: float, m1: pyYggdrasil.Random.RNDistribution, w2: float, m2: pyYggdrasil.Random.RNDistribution, w3: float, m3: pyYggdrasil.Random.RNDistribution, w4: float, m4: pyYggdrasil.Random.RNDistribution, m5: pyYggdrasil.Random.RNDistribution, domain: pyYggdrasil.HyperCube = The domain that is used to restrict the sample range, if invalid, the domain is estimated.) -> pyYggdrasil.Random.RNDistribution\n",
+ " \n",
+ " This function creates a distribution that is composed out of five given modes. The last weight is given implicitly such that the weights sum up to one. The domain is optional and defaults to the invalid domain. This means that it is tried to estimate a domain from the underling distributions.\n",
+ " \n",
+ " estMeanCovar(...) method of builtins.PyCapsule instance\n",
+ " estMeanCovar(*args, **kwargs)\n",
+ " Overloaded function.\n",
+ " \n",
+ " 1. estMeanCovar(x: pyYggdrasil.SampleArray) -> Tuple[numpy.ndarray[float64[m, 1]], numpy.ndarray[float64[m, n]]]\n",
+ " \n",
+ " This function calculates the mean and the covariance of the samples that are stored inside the sample array. The fucntion returns a pair, the first memeber contains the mean and the second contains the covariance. Notice the type is always an Eigen::Matrix type, this means it will be converted to a numpy array. Also notice that this holds true even for one dimensional samples.\n",
+ " \n",
+ " 2. estMeanCovar(x: pyYggdrasil.SampleList) -> Tuple[numpy.ndarray[float64[m, 1]], numpy.ndarray[float64[m, n]]]\n",
+ " \n",
+ " This function calculates the mean and the covariance of the samples that are stored inside the sample list. The fucntion returns a pair, the first memeber contains the mean and the second contains the covariance. Notice the type is always an Eigen::Matrix type, this means it will be converted to a numpy array. Also notice that this holds true even for one dimensional samples.\n",
+ " \n",
+ " 3. estMeanCovar(x: pyYggdrasil.SampleCollection) -> Tuple[numpy.ndarray[float64[m, 1]], numpy.ndarray[float64[m, n]]]\n",
+ " \n",
+ " This function calculates the mean and the covariance of the samples that are stored inside the sample collection. The fucntion returns a pair, the first memeber contains the mean and the second contains the covariance. Notice the type is always an Eigen::Matrix type, this means it will be converted to a numpy array. Also notice that this holds true even for one dimensional samples.\n",
+ "\n",
+ "FILE\n",
+ " (built-in)\n",
+ "\n",
+ "\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# The generators are inside a submodule.\n",
@@ -163,9 +1749,130 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 5,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Help on class Gauss in module pyYggdrasil.Random:\n",
+ "\n",
+ "class Gauss(RNDistribution)\n",
+ " | This class is a implementaion of the distribution interface. It implements a multi dimensional Gauss distribution. It uses Eigen matrices, which should be compatible to numpy matrices.\n",
+ " | \n",
+ " | Method resolution order:\n",
+ " | Gauss\n",
+ " | RNDistribution\n",
+ " | pybind11_builtins.pybind11_object\n",
+ " | builtins.object\n",
+ " | \n",
+ " | Methods defined here:\n",
+ " | \n",
+ " | __init__(...)\n",
+ " | __init__(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. __init__(self: pyYggdrasil.Random.Gauss, mu: numpy.ndarray[float64[m, 1]], Sigma: numpy.ndarray[float64[m, n]], domain: pyYggdrasil.HyperCube) -> None\n",
+ " | \n",
+ " | This constructor constructs a Gaussian distribution, with has the median mu and the covariance matrix Sigma, which is something like the VARIANCE (this is different from many other interface, also from the C++ standard. There are no direct check if Sigma is symmertric, it is feeded to the Cholescky decomposition routine of Eigen, if it is fine with it, then its good. Not that only one half of the matrix is used. The sampling is restricted to the hyper cube hc. If it is invalid, then no restrictions are applied. The class works with Eigen types, they will be automatically converted from numpy arrays.\n",
+ " | \n",
+ " | 2. __init__(self: pyYggdrasil.Random.Gauss, mu: numpy.ndarray[float64[m, 1]], Sigma: numpy.ndarray[float64[m, n]]) -> None\n",
+ " | \n",
+ " | This constructor constructs a Gauss distribution. The sampling domain is invalid, which means that there are no restrictions applied. For a discussion of the arguments see the other constructor.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Methods inherited from RNDistribution:\n",
+ " | \n",
+ " | generateSamples(...)\n",
+ " | generateSamples(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG) -> pyYggdrasil.Sample\n",
+ " | \n",
+ " | This function generates one sample from the underling distribution. The probqability is guaranteed to be positive and non zero. This guarantees comes from \"nomerical zeros\" that are effects of the fiunite size arethmetic.\n",
+ " | \n",
+ " | generateSamplesArray(...)\n",
+ " | generateSamplesArray(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleArray\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. The samples are stored inside a sample array. It is guaranteed tha the pdf fucntion of all samples inside the return value will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesArrayPDF(...)\n",
+ " | generateSamplesArrayPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleArray, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Generate N samples that are drawn from the distribution. This fucntion returns a pair, the first element is a sample array which contains the drawn sample. The second elements are the pdf values of those samples.\n",
+ " | \n",
+ " | generateSamplesCollection(...)\n",
+ " | generateSamplesCollection(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleCollection\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. The samples are stored inside a sample colelction.\n",
+ " | \n",
+ " | generateSamplesCollectionPDF(...)\n",
+ " | generateSamplesCollectionPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleCollection, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | This function drawns N samples from the distribution. It returns a pair, where the first memeber is a sample collection, which the drawn samples. The second memeber is a vector that contains the probabilioty of said memeber.\n",
+ " | \n",
+ " | generateSamplesList(...)\n",
+ " | generateSamplesList(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> pyYggdrasil.SampleList\n",
+ " | \n",
+ " | Drawn N sample from the underling distribution. The samples are stored inside a sample list. Not that the pdf of all samples will be greater than zero.\n",
+ " | \n",
+ " | generateSamplesListPDF(...)\n",
+ " | generateSamplesListPDF(self: pyYggdrasil.Random.RNDistribution, g: pyYggdrasil.Random.pRNG, N: int) -> Tuple[pyYggdrasil.SampleList, numpy.ndarray[float64[m, 1]]]\n",
+ " | \n",
+ " | Drawn N samples from the underling distribution. This function returns a pair, the first memeber contains a sample list, with the drawn samples. The second member is an array that stores the associated pdfs.\n",
+ " | \n",
+ " | getCoVar(...)\n",
+ " | getCoVar(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, n]]\n",
+ " | \n",
+ " | This function returns the theoretical covariance of the distribution. This is an Eigen matrix, that should be converted into a numpy matrix.\n",
+ " | \n",
+ " | getMean(...)\n",
+ " | getMean(self: pyYggdrasil.Random.RNDistribution) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function returns the mean of the underling distribution. The value is an Eigen::VectorXd, pybind11 should convert it into an numpy array automatically. Note that this involves coping.\n",
+ " | \n",
+ " | getSampleDomain(...)\n",
+ " | getSampleDomain(self: pyYggdrasil.Random.RNDistribution) -> pyYggdrasil.HyperCube\n",
+ " | \n",
+ " | This fucntion returns the sampling domain of *this. The sampling domain is the domain where in principle samples could occure. If this domain is infinite, aka. unbounded, then an invalid cube is returned.\n",
+ " | \n",
+ " | isUnbounded(...)\n",
+ " | isUnbounded(self: pyYggdrasil.Random.RNDistribution) -> bool\n",
+ " | \n",
+ " | This function returns true if *this is undbounded. This means that are no restriction, beside some hardware ones, where the samples could lie.\n",
+ " | \n",
+ " | nDims(...)\n",
+ " | nDims(self: pyYggdrasil.Random.RNDistribution) -> int\n",
+ " | \n",
+ " | Return the dimensions of the samples that are generated.\n",
+ " | \n",
+ " | pdf(...)\n",
+ " | pdf(*args, **kwargs)\n",
+ " | Overloaded function.\n",
+ " | \n",
+ " | 1. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.Sample) -> float\n",
+ " | \n",
+ " | Calculate the pdf of the given sample s. Note that the result can be zero in finite arethmentic, but in infinite arethmetic the sample could have a positive value.\n",
+ " | \n",
+ " | 2. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleArray) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This function calculates the pdf of all samples that are inside the sample array.\n",
+ " | \n",
+ " | 3. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleList) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | Calcualte the pdf of the samples that are stored inside the sample array.\n",
+ " | \n",
+ " | 4. pdf(self: pyYggdrasil.Random.RNDistribution, s: pyYggdrasil.SampleCollection) -> numpy.ndarray[float64[m, 1]]\n",
+ " | \n",
+ " | This fucntion calculates the probability of the samples that are inside the given sample collection.\n",
+ " | \n",
+ " | ----------------------------------------------------------------------\n",
+ " | Static methods inherited from pybind11_builtins.pybind11_object:\n",
+ " | \n",
+ " | __new__(*args, **kwargs) from pybind11_builtins.pybind11_type\n",
+ " | Create and return a new object. See help(type) for accurate signature.\n",
+ "\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# This is rather crowded, we only want Gauss, for the beginning.\n",
@@ -193,7 +1900,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 6,
"metadata": {},
"outputs": [],
"source": [
@@ -225,7 +1932,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 7,
"metadata": {},
"outputs": [],
"source": [
@@ -257,7 +1964,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 8,
"metadata": {},
"outputs": [],
"source": [
@@ -272,9 +1979,27 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 9,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "The estimated mean is: [-0.00012161 -0.00025211 -0.00071741]\n",
+ "The theoretical mean is: [0. 0. 0.]\n",
+ "\n",
+ "The empirical covariance is:\n",
+ "[[0.59849472 0.24901251 0.49819026]\n",
+ " [0.24901251 0.39954903 0.59908904]\n",
+ " [0.49819026 0.59908904 0.99831205]]\n",
+ "The theoretical covariance is:\n",
+ "[[0.6 0.25 0.5 ]\n",
+ " [0.25 0.4 0.6 ]\n",
+ " [0.5 0.6 1. ]]\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# We estimate mean and covariance of the generated samples\n",
@@ -311,7 +2036,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 10,
"metadata": {},
"outputs": [],
"source": [
@@ -342,7 +2067,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 11,
"metadata": {},
"outputs": [],
"source": [
@@ -356,9 +2081,20 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 12,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Numbers of leafs: 6846\n",
+ "Deepest depth: 17\n",
+ "Mean depth: 14.602687700847211\n",
+ "Std depth: 2.278560112982685\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# A samll inspection of the tree\n",
@@ -383,9 +2119,17 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 13,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "It took 3.471541404724121 seconds.\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# Experiment parameters\n",
@@ -439,9 +2183,22 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 14,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 1500x1000 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
"#\n",
"# Now plotting the result\n",
@@ -482,7 +2239,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 15,
"metadata": {},
"outputs": [],
"source": [
@@ -507,7 +2264,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 16,
"metadata": {},
"outputs": [],
"source": [
@@ -532,9 +2289,17 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 17,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "The sampler contains 31 many domains out of 6846 that are inside the tree.\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# Generate the conditioned sampler\n",
@@ -545,6 +2310,12 @@
" cMap = cond_dict # The condition\n",
")\n",
"\n",
+ "# How many domains have joined the sampler\n",
+ "print(\"The sampler contains {} many domains out of {} that are inside the tree.\".format(\n",
+ " cond_sampler.nProxy(), \n",
+ " stat.getLeafCount())\n",
+ ")\n",
+ "\n",
"#\n",
"# Get the condition object\n",
"conditions = cond_sampler.getConditions()"
@@ -560,7 +2331,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 18,
"metadata": {},
"outputs": [],
"source": [
@@ -574,9 +2345,27 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 19,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "The estimated mean is: [0.00904994 0. 0. ]\n",
+ "\n",
+ "The empirical covariance is:\n",
+ "[[0.2891625 0. 0. ]\n",
+ " [0. 0. 0. ]\n",
+ " [0. 0. 0. ]]\n",
+ "\n",
+ "The theoretical (true) covariance is:\n",
+ "[[0.2875 0. 0. ]\n",
+ " [0. 0. 0. ]\n",
+ " [0. 0. 0. ]]\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# We estimate mean and covariance of the generated samples\n",
@@ -598,23 +2387,50 @@
"metadata": {},
"source": [
"### Plotting the Result\n",
- "We will compare the analytical PDF with the one that is given by the tree sampler.\n",
- "Plotting this will result in quite some boiler code, it is an interesting example, but more not.\n",
+ "We will compare the analytical pdf with the one given by the estimator and a histogram of randomly drawn samples.\n",
"\n",
- "The ``TreeSampler`` can not be used to compute the PDF of a sample directly.\n",
- "But can compute the PDF of a sample that is generated as a by product.\n",
- "The reason is that not everything of the tree is copied and computing the PDF value is possible, but extremely inefficient, so it is not supported."
+ "The tree sampler can not be used to compute the pdf of samples.\n",
+ "For that we will use the tree itself and use it to compute the (unconditioned) pdfs.\n",
+ "We will then use the sampler to transform the unconditioned pdfs to conditioned ones.\n",
+ "We will also do the same for the analytical pdfs.\n"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "#### Preparation\n",
+ "Here we will make some preparation for the plotting."
]
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 20,
"metadata": {},
"outputs": [],
"source": [
- "# this are the bounds for the plotting\n",
+ "# This are the bounds for the plotting\n",
"plotLowBound = -3.0 # not Symetric because of the legend\n",
- "plotUppBound = 2.0"
+ "plotUppBound = 2.0\n",
+ "\n",
+ "# This is the numbers of samples we use for plotting (~ grid points)\n",
+ "plotSamples = int((plotUppBound - plotLowBound) * 600)\n",
+ "\n",
+ "# This is tzhe grid we use for plotting it is in \n",
+ "# one dimenion and not in a container format.\n",
+ "condSampler_aSamples_fd = np.linspace(\n",
+ " start = plotLowBound, \n",
+ " stop = plotUppBound,\n",
+ " num = plotSamples\n",
+ ")\n",
+ "\n",
+ "# Transform it in a container.\n",
+ "# This is a restricted sample\n",
+ "condSampler_resSample = pyY.SampleArray(condSampler_aSamples_fd, pyY.eMatrixViewSample.Row)\n",
+ "\n",
+ "# Now we will expand it, such that we get full \n",
+ "# samples that we can use.\n",
+ "condSampler_qSamples = conditions.expandSample(condSampler_resSample)"
]
},
{
@@ -627,97 +2443,122 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 21,
"metadata": {},
"outputs": [],
"source": [
- "# This is the numbers of samples we use for plotting (~ grid points)\n",
- "plotSamples = int((plotUppBound - plotLowBound) * 600)\n",
- "\n",
- "# Now generate a grid of the first dimension\n",
- "condSampler_aSamples_fd = np.linspace(\n",
- " start = plotLowBound, \n",
- " stop = plotUppBound,\n",
- " num = plotSamples\n",
- ")\n",
- "\n",
- "# Store the samples inside a sample array \n",
- "# This container will later be expanded\n",
- "restricted_plot_samples_array = pyY.SampleArray(\n",
- " mat = condSampler_aSamples_fd, # The matrix we load from\n",
- " view = pyY.eMatrixViewSample.Row # Each row contains one sample \n",
- ")\n",
+ "#\n",
+ "# Computing of the analytical (unconditioned) pdf values.\n",
+ "# The samples that we use are compling to the condition.\n",
+ "condSampler_qSamples_exPDF = mmodeDist.pdf(condSampler_qSamples)\n",
"\n",
- "# Expand the restricted samples, this will automatically set the conditions\n",
- "condSampler_aSamples_arrayC = conditions.expandSample(restricted_plot_samples_array)\n",
+ "# We we transform the pdf such that it is conditioned.\n",
+ "# For that we can use the tree sampler, that supports the value.\n",
+ "condSampler_qSamples_exPDF *= cond_sampler.getInvPxC()"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "#### Empirical PDF\n",
+ "Now we get the empirical pdf values.\n",
+ "For that we will use the tree and the transform the unconditioned pdfs, which are generated by the tree to conditioned one, by applying the constant.\n",
"\n",
+ "This is needed because the sampler can not compute the pdf of a sample on its own."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 42,
+ "metadata": {},
+ "outputs": [],
+ "source": [
"#\n",
- "# Get the analytical pdf values; The values are unconditioned \n",
- "condSampler_aSamples_exPDF = mmodeDist.pdf(condSampler_aSamples_arrayC)\n",
+ "# Computing of the empirical (unconditioned) pdf values.\n",
+ "# The samples that we use are compling to the condition.\n",
+ "condSampler_qSamples_empPDF = tree.evaluateSamplesAt(condSampler_qSamples)\n",
"\n",
- "# Now we apply the condition by hand\n",
- "condSampler_aSamples_exPDF *= cond_sampler.getInvPxC()"
+ "# We we transform the pdf such that it is conditioned.\n",
+ "# For that we can use the tree sampler, that supports the value.\n",
+ "condSampler_qSamples_empPDF *= cond_sampler.getInvPxC()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "#### Generate Tree Estimate\n",
- "Here we will compute the PDF values that comes from the ``TreeSampler``."
+ "#### Generate Histogram Data\n",
+ "Here we will generate the data for the histogram.\n",
+ "We will generate a number of samples and then restrict them.\n",
+ "(We could also use the `getDimensionalArray()` function, but for the sake of demonstration, we will restrict it, it is formal more correct. But I recommend doing the other thing.)\n",
+ "\n",
+ "We will also check that the samples are inside the plot range."
]
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 52,
"metadata": {},
"outputs": [],
"source": [
"# To get a good coverage, we need a lot of samples\n",
- "plotSamples *= 10\n",
+ "plotSamples *= 20\n",
+ "\n",
+ "# Generate random samples\n",
+ "condSampler_histSamplesFull = cond_sampler.generateSamplesCollection(g = geni, N = plotSamples)\n",
"\n",
- "# Generate random samples together with their PDF\n",
- "condSampler_tSamples, condSampler_tSamples_pdf = cond_sampler.generateSamplesArrayPDF(g = geni, N = plotSamples)\n",
+ "# Restrict the samples, this will return a collection of dimension one.\n",
+ "condSampler_histSamplesRes_coll = conditions.restrictSample(condSampler_histSamplesFull)\n",
"\n",
- "# Extract the first dimensions, since it is the only that is not prescribed\n",
- "condSampler_tSamples_fd = condSampler_tSamples.getDimensionArray(0)\n",
+ "# Convert the container into a real matrix\n",
+ "condSampler_histSamplesRes = condSampler_histSamplesRes_coll.getMatrix(pyY.eMatrixViewSample.Row)\n",
+ "\n",
+ "# You could also do, it is directer, but the previous two lines are more formal.\n",
+ "#condSampler_histSamplesRes = condSampler_tSamples.condSampler_histSamplesFull(0)\n",
"\n",
"# Get all the values that are inside the range\n",
- "insidePlotRange = (condSampler_tSamples_fd >= plotLowBound) & (condSampler_tSamples_fd <= plotUppBound)"
+ "insidePlotRange = (condSampler_histSamplesRes >= plotLowBound) & (condSampler_histSamplesRes <= plotUppBound)\n",
+ "\n",
+ "# Extract them\n",
+ "condSampler_histSamplesRes = condSampler_histSamplesRes[insidePlotRange]"
]
},
{
- "cell_type": "code",
- "execution_count": null,
+ "cell_type": "markdown",
"metadata": {},
- "outputs": [],
"source": [
- "#\n",
- "# Extract all the samples that are inside the range\n",
- "condSampler_tSamples_fd = condSampler_tSamples_fd[insidePlotRange]\n",
- "condSampler_tSamples_pdf = condSampler_tSamples_pdf[insidePlotRange]\n",
- "\n",
- "#\n",
- "# Now we must sort them such that they can be plotted nicely\n",
- "orderIdx = np.argsort(condSampler_tSamples_fd)\n",
- "\n",
- "# Apply the ordering\n",
- "condSampler_tSamples_fd = condSampler_tSamples_fd[orderIdx]\n",
- "condSampler_tSamples_pdf = condSampler_tSamples_pdf[orderIdx]"
+ "#### Final Plotting"
]
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 51,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 1500x1000 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
"plt.title(\"Comparison of the tree and analytic pdf\")\n",
"plt.xlabel(\"x\")\n",
"plt.ylabel(\"pdf\")\n",
"\n",
- "plt.plot(condSampler_aSamples_fd, condSampler_aSamples_exPDF, label=\"Analytic PDF\")\n",
- "plt.plot(condSampler_tSamples_fd, condSampler_tSamples_pdf, label=\"TreeSampler PDF\")\n",
+ "plt.hist(condSampler_histSamplesRes, density = True, \n",
+ " range = (plotLowBound, plotUppBound), bins = 200,\n",
+ " label = \"Histogram\")\n",
+ "plt.plot(condSampler_aSamples_fd, condSampler_qSamples_exPDF, label=\"Analytic PDF\")\n",
+ "plt.plot(condSampler_aSamples_fd, condSampler_qSamples_empPDF, label=\"Empiric PDF\")\n",
"\n",
"plt.legend(loc=2)\n",
"plt.show()"
@@ -742,9 +2583,17 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 25,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "It took 3.0863523483276367 seconds.\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# Defining the parameters\n",
@@ -783,7 +2632,7 @@
" builder = builder)\n",
" \n",
" # Generate the sampler\n",
- " treeSampler_i = pyR.TreeSampler(fitTree_i)\n",
+ " treeSampler_i = pyR.TreeSampler(fitTree_i, conditions)\n",
" \n",
" # Perform the resampling\n",
" bootSamples_i = treeSampler_i.generateSamplesCollection(g = geni, N = testSize)\n",
@@ -813,9 +2662,22 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 26,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 1500x1000 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
"#\n",
"# Now plotting the result\n",
@@ -851,7 +2713,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 27,
"metadata": {},
"outputs": [],
"source": [
@@ -874,7 +2736,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 28,
"metadata": {},
"outputs": [],
"source": [
@@ -888,9 +2750,33 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 29,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "The estimated mean is: [ 0.00045477 -0.00011629 -0.00081433]\n",
+ "The mean of the source is: [-0.00012161 -0.00025211 -0.00071741]\n",
+ "The theoretical mean is: [0. 0. 0.]\n",
+ "\n",
+ "The empirical covariance (bootstrapped):\n",
+ "[[0.6625591 0.25005117 0.50482876]\n",
+ " [0.25005117 0.43041613 0.60634894]\n",
+ " [0.50482876 0.60634894 1.0434699 ]]\n",
+ "The covariance of the source is:\n",
+ "[[0.59849472 0.24901251 0.49819026]\n",
+ " [0.24901251 0.39954903 0.59908904]\n",
+ " [0.49819026 0.59908904 0.99831205]]\n",
+ "\n",
+ "The theoretical (true) covariance is:\n",
+ "[[0.6 0.25 0.5 ]\n",
+ " [0.25 0.4 0.6 ]\n",
+ " [0.5 0.6 1. ]]\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# We estimate mean and covariance of the generated samples\n",
@@ -928,9 +2814,17 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 30,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "It took 3.59222412109375 seconds.\n"
+ ]
+ }
+ ],
"source": [
"#\n",
"# Defining the parameters\n",
@@ -1010,9 +2904,22 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 31,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 1500x1000 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
"#\n",
"# Now plotting the result\n",