Python numpy.diagonal方法代碼示例

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def CRE(size):
    r"""Circular real ensemble (CRE).

    Parameters
    ----------
    size : tuple
        ``(n, n)``, where `n` is the dimension of the output matrix.

    Returns
    -------
    U : ndarray
        Orthogonal matrix drawn from the CRE (=Haar measure on O(n)).
    """
    # almost same code as for CUE
    n, m = size
    assert n == m  # ensure that `mode` in qr doesn't matter.
    A = np.random.standard_normal(size)
    Q, R = np.linalg.qr(A)
    # Q-R is not unique; to make it unique ensure that the diagonal of R is positive
    # Q' = Q*L; R' = L^{-1} *R, where L = diag(phase(diagonal(R)))
    L = np.diagonal(R)
    Q *= np.sign(L)
    return Q 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def CUE(size):
    r"""Circular unitary ensemble (CUE).

    Parameters
    ----------
    size : tuple
        ``(n, n)``, where `n` is the dimension of the output matrix.

    Returns
    -------
    U : ndarray
        Unitary matrix drawn from the CUE (=Haar measure on U(n)).
    """
    # almost same code as for CRE
    n, m = size
    assert n == m  # ensure that `mode` in qr doesn't matter.
    A = standard_normal_complex(size)
    Q, R = np.linalg.qr(A)
    # Q-R is not unique; to make it unique ensure that the diagonal of R is positive
    # Q' = Q*L; R' = L^{-1} *R, where L = diag(phase(diagonal(R)))
    L = np.diagonal(R).copy()
    L[np.abs(L) < 1.e-15] = 1.
    Q *= L / np.abs(L)
    return Q 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def O_close_1(size, a=0.01):
    r"""return an random orthogonal matrix 'close' to the Identity.

    Parameters
    ----------
    size : tuple
        ``(n, n)``, where `n` is the dimension of the output matrix.
    a : float
        Parameter determining how close the result is on `O`;
        :math:`\lim_{a \rightarrow 0} <|O-E|>_a = 0`` (where `E` is the identity).

    Returns
    -------
    O : ndarray
        Orthogonal matrix close to the identiy (for small `a`).
    """
    n, m = size
    assert n == m
    A = GOE(size) / (2. * n)**0.5  # scale such that eigenvalues are in [-1, 1]
    E = np.eye(size[0])
    Q, R = np.linalg.qr(E + a * A)
    L = np.diagonal(R)  # make QR decomposition unique & ensure Q is close to one for small `a`
    Q *= np.sign(L)
    return Q 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def log_multivariate_normal_density(X, means, covars, min_covar=1.e-7):
    """Log probability for full covariance matrices. """
    if hasattr(linalg, 'solve_triangular'):
        # only in scipy since 0.9
        solve_triangular = linalg.solve_triangular
    else:
        # slower, but works
        solve_triangular = linalg.solve
    n_samples, n_dim = X.shape
    nmix = len(means)
    log_prob = np.empty((n_samples, nmix))
    for c, (mu, cv) in enumerate(zip(means, covars)):
        try:
            cv_chol = linalg.cholesky(cv, lower=True)
        except linalg.LinAlgError:
            # The model is most probabily stuck in a component with too
            # few observations, we need to reinitialize this components
            cv_chol = linalg.cholesky(cv + min_covar * np.eye(n_dim),
                                      lower=True)
        cv_log_det = 2 * np.sum(np.log(np.diagonal(cv_chol)))
        cv_sol = solve_triangular(cv_chol, (X - mu).T, lower=True).T
        log_prob[:, c] = - .5 * (np.sum(cv_sol ** 2, axis=1) + \
                                     n_dim * np.log(2 * np.pi) + cv_log_det)

    return log_prob 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def cov(self):
        r"""The covariance matrix describing the Gaussian state.

        The diagonal elements of the covariance matrix correspond to the
        variance in the position and momentum quadratures:

        .. math::
            \mathbf{V}_{ii} = \begin{cases}
                (\Delta x_i)^2, & 0\leq i\leq N-1\\
                (\Delta p_{i-N})^2, & N\leq i\leq 2(N-1)
            \end{cases}

        where :math:`\Delta x_i` and :math:`\Delta p_i` refer to the
        position and momentum quadrature variance of mode :math:`i` respectively.

        Note that if the covariance matrix is purely diagonal, then this
        corresponds to squeezing :math:`z=re^{i\phi}` where :math:`\phi=0`,
        and :math:`\Delta x_i = e^{-2r}`, :math:`\Delta p_i = e^{2r}`.

        Returns:
          array: the :math:`2N\times 2N` covariance matrix.
        """
        return self._cov 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def test_diagonal(self):
        a = np.arange(12).reshape((3, 4))
        assert_equal(a.diagonal(), [0, 5, 10])
        assert_equal(a.diagonal(0), [0, 5, 10])
        assert_equal(a.diagonal(1), [1, 6, 11])
        assert_equal(a.diagonal(-1), [4, 9])

        b = np.arange(8).reshape((2, 2, 2))
        assert_equal(b.diagonal(), [[0, 6], [1, 7]])
        assert_equal(b.diagonal(0), [[0, 6], [1, 7]])
        assert_equal(b.diagonal(1), [[2], [3]])
        assert_equal(b.diagonal(-1), [[4], [5]])
        assert_raises(ValueError, b.diagonal, axis1=0, axis2=0)
        assert_equal(b.diagonal(0, 1, 2), [[0, 3], [4, 7]])
        assert_equal(b.diagonal(0, 0, 1), [[0, 6], [1, 7]])
        assert_equal(b.diagonal(offset=1, axis1=0, axis2=2), [[1], [3]])
        # Order of axis argument doesn't matter:
        assert_equal(b.diagonal(0, 2, 1), [[0, 3], [4, 7]]) 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def _log_multivariate_normal_density_full(X, means, covars, min_covar=1.e-7):
    """Log probability for full covariance matrices.
    """
    n_samples, n_dim = X.shape
    nmix = len(means)
    log_prob = np.empty((n_samples, nmix))
    for c, (mu, cv) in enumerate(zip(means, covars)):
        try:
            cv_chol = linalg.cholesky(cv, lower=True)
        except linalg.LinAlgError:
            # The model is most probably stuck in a component with too
            # few observations, we need to reinitialize this components
            cv_chol = linalg.cholesky(cv + min_covar * np.eye(n_dim),
                                      lower=True)
        cv_log_det = 2 * np.sum(np.log(np.diagonal(cv_chol)))
        cv_sol = linalg.solve_triangular(cv_chol, (X - mu).T, lower=True).T
        log_prob[:, c] = - .5 * (np.sum(cv_sol ** 2, axis=1) +
                                 n_dim * np.log(2 * np.pi) + cv_log_det)

    return log_prob 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def sp_to_adjency(C, threshinf=0.2, threshsup=1.8):
    """ Thresholds the structure matrix in order to compute an adjency matrix.
    All values between threshinf and threshsup are considered representing connected nodes and set to 1. Else are set to 0
    Parameters
    ----------
    C : ndarray, shape (n_nodes,n_nodes)
        The structure matrix to threshold
    threshinf : float
        The minimum value of distance from which the new value is set to 1
    threshsup : float
        The maximum value of distance from which the new value is set to 1
    Returns
    -------
    C : ndarray, shape (n_nodes,n_nodes)
        The threshold matrix. Each element is in {0,1}
    """
    H = np.zeros_like(C)
    np.fill_diagonal(H, np.diagonal(C))
    C = C - H
    C = np.minimum(np.maximum(C, threshinf), threshsup)
    C[C == threshsup] = 0
    C[C != 0] = 1

    return C 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def diagflat(v, k=0):
  """Returns a 2-d array with flattened `v` as diagonal.

  Args:
    v: array_like of any rank. Gets flattened when setting as diagonal. Could be
      an ndarray, a Tensor or any object that can be converted to a Tensor using
      `tf.convert_to_tensor`.
    k: Position of the diagonal. Defaults to 0, the main diagonal. Positive
      values refer to diagonals shifted right, negative values refer to
      diagonals shifted left.

  Returns:
    2-d ndarray.
  """
  v = asarray(v)
  return diag(tf.reshape(v.data, [-1]), k) 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def get_like_from_mats(ky_mat, l_mat, alpha, name):
    """ compute the likelihood from the covariance matrix

    :param ky_mat: the covariance matrix

    :return: float, likelihood
    """
    # catch linear algebra errors
    labels = _global_training_labels[name]

    # calculate likelihood
    like = (-0.5 * np.matmul(labels, alpha) -
            np.sum(np.log(np.diagonal(l_mat))) -
            math.log(2 * np.pi) * ky_mat.shape[1] / 2)

    return like


#######################################
##### KY MATRIX FUNCTIONS and gradients
####################################### 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def zz(matrix, nb):
    r"""Zig-zag traversal of the input matrix
    :param matrix: input matrix
    :param nb: number of coefficients to keep
    :return: an array of nb coefficients
    """
    flipped = np.fliplr(matrix)
    rows, cols = flipped.shape  # nb of columns

    coefficient_list = []

    for loop, i in enumerate(range(cols - 1, -rows, -1)):
        anti_diagonal = np.diagonal(flipped, i)

        # reversing even diagonals prioritizes the X resolution
        # reversing odd diagonals prioritizes the Y resolution
        # for square matrices, the information content is the same only when nb covers half of the matrix
        #  e.g. [ nb = n*(n+1)/2 ]
        if loop % 2 == 0:
            anti_diagonal = anti_diagonal[::-1]  # reverse anti_diagonal

        coefficient_list.extend([x for x in anti_diagonal])

    # flattened = [val for sublist in coefficient_list for val in sublist]
    return coefficient_list[:nb] 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def _fit(self, X, model_dump_path=None, verbose=True):
    """
    :param X: shape (n_videos, n_frames, n_descriptors_per_image, n_dim_descriptor)
    :param model_dump_path: (optional) path where the fitted model shall be dumped
    :param verbose - boolean that controls the verbosity
    :return: fitted Fisher vector object
    """
    assert X.ndim == 4
    self.feature_dim = X.shape[-1]

    X = X.reshape(-1, X.shape[-1])

    # fit GMM and store params of fitted model
    self.gmm = gmm = GaussianMixture(n_components=self.n_kernels, covariance_type=self.covariance_type, max_iter=1000).fit(X)
    self.covars = gmm.covariances_
    self.means = gmm.means_
    self.weights = gmm.weights_

    # if cov_type is diagonal - make sure that covars holds a diagonal matrix
    if self.covariance_type == 'diag':
      cov_matrices = np.empty(shape=(self.n_kernels, self.covars.shape[1], self.covars.shape[1]))
      for i in range(self.n_kernels):
        cov_matrices[i, :, :] = np.diag(self.covars[i, :])
      self.covars = cov_matrices

    assert self.covars.ndim == 3
    self.fitted = True
    if verbose:
      print('fitted GMM with %i kernels'%self.n_kernels)

    if model_dump_path:
      with open(model_dump_path, 'wb') as f:
        pickle.dump(self,f, protocol=4)
      if verbose:
        print('Dumped fitted model to', model_dump_path)

    return self 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def init_amps(self, eris):
        time0 = time.clock(), time.time()
        mo_e = eris.fock.diagonal()
        nocc = self.nocc
        eia = mo_e[:nocc,None] - mo_e[None,nocc:]
        eijab = lib.direct_sum('ia,jb->ijab',eia,eia)
        t1 = eris.fock[:nocc,nocc:] / eia
        eris_oovv = np.array(eris.oovv)
        t2 = eris_oovv/eijab
        self.emp2 = 0.25*einsum('ijab,ijab',t2,eris_oovv.conj()).real
        logger.info(self, 'Init t2, MP2 energy = %.15g', self.emp2)
        logger.timer(self, 'init mp2', *time0)
        return self.emp2, t1, t2 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def make_cholesky_unique(chol):
    """Make a lower triangular cholesky factor unique.

    Cholesky factors are only unique with the additional requirement that all diagonal
    elements are positive. This is done automatically by np.linalg.cholesky.
    Since we calucate cholesky factors by QR decompositions we have to do it manually.

    It is obvious from that this is admissible because:

    chol sign_swither sign_switcher.T chol.T = chol chol.T

    """
    sign_switcher = np.sign(np.diag(np.diagonal(chol)))
    return chol @ sign_switcher 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import diagonal [as 別名]
def test_diagonal():
    b1 = np.matrix([[1,2],[3,4]])
    diag_b1 = np.matrix([[1, 4]])
    array_b1 = np.array([1, 4])

    assert_equal(b1.diagonal(), diag_b1)
    assert_equal(np.diagonal(b1), array_b1)
    assert_equal(np.diag(b1), array_b1)