Python scipy.special.ndtri() Examples

def _ppf(self, q, c):
        tmp = c*sc.ndtri(q)
        return 0.25 * (tmp + np.sqrt(tmp**2 + 4))**2 

def ns_prior_transform(utheta):
#    global config.BASEMENT
    theta = np.zeros_like(utheta)*np.nan
    for i in range(len(theta)):
        if config.BASEMENT.bounds[i][0]=='uniform':
            theta[i] = utheta[i]*(config.BASEMENT.bounds[i][2]-config.BASEMENT.bounds[i][1]) + config.BASEMENT.bounds[i][1]
        elif config.BASEMENT.bounds[i][0]=='normal':
            theta[i] = config.BASEMENT.bounds[i][1] + config.BASEMENT.bounds[i][2]*ndtri(utheta[i])
        elif config.BASEMENT.bounds[i][0]=='trunc_normal':
            theta[i] = my_truncnorm_isf(utheta[i],config.BASEMENT.bounds[i][1],config.BASEMENT.bounds[i][2],config.BASEMENT.bounds[i][3],config.BASEMENT.bounds[i][4]) 
        else:
            raise ValueError('Bounds have to be "uniform", "normal" and "trunc_normal". Input from "params.csv" was "'+config.BASEMENT.bounds[i][0]+'".')
    return theta 

def _ppf(self, x, a):
        return numpy.e**(a*special.ndtri(x)) 

def _ppf(self, q, c, s):
        return numpy.exp(-s*special.ndtri(pow(1.-q, 1./c))) 

def _ppf(self, q, a):
        return 1.0/(a-special.ndtri(q*special.ndtr(a))) 

def _ppf(self, q, C, Ci, loc):
        return (numpy.dot(C, special.ndtri(q)).T+loc.T).T 

def _ppf(self, q, a, b, mu, sigma):
        fa = special.ndtr((a-mu)/sigma)
        fb = special.ndtr((b-mu)/sigma)
        return special.ndtri(q*(fb-fa) + fa)*sigma + mu 

def _ppf(self, q, c):
        return -special.ndtri(pow(1-q, 1./c)) 

def _ppf(self, q):
        val = special.ndtri(1-q/2.0)
        return 1.0/(val*val) 

def _ppf(self, q, c):
        tmp = c*special.ndtri(q)
        return 0.25*(tmp + numpy.sqrt(tmp**2 + 4))**2 

def _ppf(self, q, C, Ci):
        out = special.ndtr(numpy.dot(C, special.ndtri(q)))
        return out 

def _cdf(self, x, C, Ci):
        out = special.ndtr(numpy.dot(Ci, special.ndtri(x)))
        return out 

def setUp(self):
        np.random.seed(1983)
        shape = (2, 3)
        self.mu = np.random.normal(size=shape)
        self.sig = np.square(np.random.normal(size=shape))
        self.obs = np.random.normal(loc=self.mu, scale=self.sig, size=shape)

        n = 1000
        q = np.linspace(0. + 0.5 / n, 1. - 0.5 / n, n)
        # convert to the corresponding normal deviates
        normppf = special.ndtri
        z = normppf(q)

        forecasts = z.reshape(-1, 1, 1) * self.sig + self.mu
        self.expected = crps_ensemble(self.obs, forecasts, axis=0) 

def p_to_z(p, tail='two'):
    """Convert p-values to (unsigned) z-values.

    Parameters
    ----------
    p : array_like
        P-values
    tail : {'one', 'two'}, optional
        Whether p-values come from one-tailed or two-tailed test. Default is
        'two'.

    Returns
    -------
    z : array_like
        Z-statistics (unsigned)
    """
    eps = np.spacing(1)
    p = np.array(p)
    p[p < eps] = eps
    if tail == 'two':
        z = ndtri(1 - (p / 2))
        z = np.array(z)
    elif tail == 'one':
        z = ndtri(1 - p)
        z = np.array(z)
        z[z < 0] = 0
    else:
        raise ValueError('Argument "tail" must be one of ["one", "two"]')

    if z.shape == ():
        z = z[()]
    return z 

def _ppf(self, q):
        # Equivalent to 1.0/(norm.isf(q/2)**2) or 0.5/(erfcinv(q)**2)
        val = -sc.ndtri(q/2)
        return 1.0 / (val * val) 

def _ppf(self, q):
        return sc.ndtri((1+q)/2.0) 

def _norm_ppf(q):
    return sc.ndtri(q) 

def _ppf(self, q, a):
        return 1.0/np.asarray(a-sc.ndtri(q*_norm_cdf(a))) 

def _norm_ppf(q):
    return sc.ndtri(q) 

def _ppf(self, q):
        # Equivalent to 1.0/(norm.isf(q/2)**2) or 0.5/(erfcinv(q)**2)
        val = -sc.ndtri(q/2)
        return 1.0 / (val * val) 

def _ppf(self, q):
        return sc.ndtri((1+q)/2.0) 

def _ppf(self, q, c):
        tmp = c*sc.ndtri(q)
        return 0.25 * (tmp + np.sqrt(tmp**2 + 4))**2 

def _ppf(self, q, a):
        return 1.0/np.asarray(a-sc.ndtri(q*_norm_cdf(a))) 

def _norm_ppf(q):
    return sc.ndtri(q) 

def ndtri(x):
    """Elementwise inverse function of ndtr.

    .. note::
       Forward computation in CPU can not be done if
       `SciPy <https://www.scipy.org/>`_ is not available.

    Args:
        x (:class:`~chainer.Variable` or :ref:`ndarray`): Input variable.

    Returns:
        ~chainer.Variable: Output variable.
    """
    return Ndtri().apply((x,))[0] 

def forward_cpu(self, x):
        if not available_cpu:
            raise ImportError('SciPy is not available. Forward computation'
                              ' of ndtri in CPU can not be done.' +
                              str(_import_error))
        self.retain_outputs((0,))
        return utils.force_array(special.ndtri(x[0]), dtype=x[0].dtype), 

def label(self):
        return 'ndtri' 

def _ndtri_cpu(x, dtype):
    from scipy import special
    return numpy.vectorize(special.ndtri, otypes=[dtype])(x) 

def _ppf(self, q):
        # Equivalent to 1.0/(norm.isf(q/2)**2) or 0.5/(erfcinv(q)**2)
        val = -sc.ndtri(q/2)
        return 1.0 / (val * val) 

def _ppf(self, q):
        return sc.ndtri((1+q)/2.0)