How to use is_zero method in assertpy

Best Python code snippet using assertpy_python

mcmc_svm.py

Source:mcmc_svm.py

1import numpy as np2from utils import compute_target, compute_std_and_var3def compute_mcmc_svm(X, y, nu, abs_eps, max_iterations, noisy=False, noise_period=20, seed=None):4    """5    Implementation of algorithm MCMC-SVM for a value alpha of 1.6    Parameters7    ----------8    X : ndarary9        An n by d array of n observation in an d-dimensional space.10    y : ndarray11        An n by 1 array of n responses.12    alpha : float13        Parameter used for L-regularization.14    nu : float15        Constant used for L-regularization.16    abs_eps : float17        Absolute tolerance used to confirm convergence.18    max_iterations : int19        Maximum number of iterations to run.20    noisy : bool21        Flag indicating whether or not to produce logs.22    noise_period : in23        Wait noise_period iterations before logging.24    seed : int or None25        Random seed used to seed numpy.random. No seeding is performed if None is passed.26    27    Returns28    -------29    betas : ndarray30        An N x d x 1 array of the N successive beta values. N is the number of iterations 31    b_values : ndarray32        An N x d x 1 array of the N successive prior mean values of beta. N is the number of iterations 33    """34    if seed is not None:35        np.random.seed(seed)36    if noisy:37        n_fill = int(np.log10(max_iterations)) + 138    _, d = X.shape39    40    yX = y * X41    std, var = compute_std_and_var(X)42    inv_var_matrix = np.diagflat(1 / var)43    44    beta = np.random.randn(d, 1)45    betas = [beta]46    omega_inv = (nu * std) / np.abs(beta)47    lam_inv = np.abs(1 - yX @ beta)48    b_values = []49    iteration = 150    51    while True:52        if noisy and iteration % noise_period == 0:53            loss = compute_target(X, y, 1, nu, beta)54            print(f'At iteration ({str(iteration).zfill(n_fill)}/{str(max_iterations).zfill(n_fill)}), loss = {loss:.3f}')55        56        # prepping step 157        B = np.linalg.inv(inv_var_matrix @ np.diagflat(omega_inv) / (nu ** 2) + yX.T @ np.diagflat(lam_inv) @ yX)58        b = B @ yX.T @ (1 + lam_inv)59        b_values.append(b)60        61        # step 162        new_beta = np.random.multivariate_normal(b.squeeze(), B).reshape(-1, 1)63        64        if np.abs(beta - new_beta).max() <= abs_eps or iteration == max_iterations:65            if noisy:66                loss = compute_target(X, y, 1, nu, beta)67                print(f'Finished run ({str(iteration).zfill(n_fill)}/{str(max_iterations).zfill(n_fill)}), loss = {loss:.3f}')68            return np.array(betas), np.array(b_values)69        beta = new_beta70        betas.append(beta)71        72        # step 273        gap = np.abs(1 - yX @ beta)74        is_zero = np.isclose(gap, 0)75        lam_inv = np.zeros_like(gap)76        lam_inv[is_zero] = 1 / (np.random.normal(size=is_zero.sum()) ** 2)77        lam_inv[~is_zero] = np.random.wald(1 / gap[~is_zero], 1)78        79        # step 380        gap = np.abs(beta) / (nu * std)81        is_zero = np.isclose(gap, 0)82        omega_inv = np.zeros_like(gap)83        omega_inv[is_zero] = 1 / (np.random.normal(size=is_zero.sum()) ** 2)84        omega_inv[~is_zero] = np.random.wald(1 / gap[~is_zero], 1)85        iteration += 186def compute_mcmc_svm_with_nu(X, y, anu, bnu, abs_eps, max_iterations, noisy=False, noise_period=20, seed=None):87    """88    Implementation of algorithm MCMC-SVM for a value alpha of 1.89    Parameters90    ----------91    X : ndarary92        An n by d array of n observation in an d-dimensional space.93    y : ndarray94        An n by 1 array of n responses.95    alpha : float96        Parameter used for L-regularization.97    anu : float98        Constant used for the prior of nu.99    bnu : float100        Constant used for the prior of nu.101    abs_eps : float102        Absolute tolerance used to confirm convergence.103    max_iterations : int104        Maximum number of iterations to run.105    noisy : bool106        Flag indicating whether or not to produce logs.107    noise_period : in108        Wait noise_period iterations before logging.109    seed : int or None110        Random seed used to seed numpy.random. No seeding is performed if None is passed.111    112    Returns113    -------114    betas : ndarray115        An N x d x 1 array of the N successive beta values. N is the number of iterations 116    b_values : ndarray117        An N x d x 1 array of the N successive prior mean values of beta. N is the number of iterations 118    """119    if seed is not None:120        np.random.seed(seed)121    if noisy:122        n_fill = int(np.log10(max_iterations)) + 1123    124    _, d = X.shape125    126    yX = y * X127    std, var = compute_std_and_var(X)128    inv_var_matrix = np.diagflat(1 / var)129    130    beta = np.random.randn(d, 1)131    betas = [beta]132    nu = 1133    nu_values = [nu]134    omega_inv = (nu * std) / np.abs(beta)135    lam_inv = np.abs(1 - yX @ beta)136    b_values = []137    iteration = 1138    139    while True:140        if noisy and iteration % noise_period == 0:141            loss = compute_target(X, y, 1, nu, beta)142            print(f'At iteration ({str(iteration).zfill(n_fill)}/{str(max_iterations).zfill(n_fill)}), loss = {loss:.3f}')143        144        # prepping step 1145        B = np.linalg.inv(inv_var_matrix @ np.diagflat(omega_inv) / (nu ** 2) + yX.T @ np.diagflat(lam_inv) @ yX)146        b = B @ yX.T @ (1 + lam_inv)147        b_values.append(b)148        149        # step 1150        new_beta = np.random.multivariate_normal(b.squeeze(), B).reshape(-1, 1)151        152        if np.abs(beta - new_beta).max() <= abs_eps or iteration == max_iterations:153            if noisy:154                loss = compute_target(X, y, 1, nu, beta)155                print(f'Finished run ({str(iteration).zfill(n_fill)}/{str(max_iterations).zfill(n_fill)}), loss = {loss:.3f}')156            return np.array(betas), np.array(b_values), np.array(nu_values).reshape(-1, 1)157        beta = new_beta158        betas.append(beta)159        160        # step 2161        gap = np.abs(1 - yX @ beta)162        is_zero = np.isclose(gap, 0)163        lam_inv = np.zeros_like(gap)164        lam_inv[is_zero] = 1 / (np.random.normal(size=is_zero.sum()) ** 2)165        lam_inv[~is_zero] = np.random.wald(1 / gap[~is_zero], 1)166        167        # step 3168        gap = np.abs(beta) / (nu * std)169        is_zero = np.isclose(gap, 0)170        omega_inv = np.zeros_like(gap)171        omega_inv[is_zero] = 1 / (np.random.normal(size=is_zero.sum()) ** 2)172        omega_inv[~is_zero] = np.random.wald(1 / gap[~is_zero], 1)173        # step 4174        nu = 1 / np.random.gamma(shape=anu + d, scale=1 / (bnu + np.abs(beta).sum()))175        nu_values.append(nu)...

codegen.py

Source:codegen.py

1def is_zero(x):2  return str(x) == "0" or str(x) == "0.0"3def is_one(x):4  return str(x) == "1" or str(x) == "1.0"5class codegen:6  def __init__(self, s):7    if(type(s) is str):8      self.str = s9    else:10      self.str = str(s)11  def __str__(self):12    return self.str13  def __eq__(a, b):14    return a.str == b.str15  def __neg__(a):16    if(is_zero(a)):17      return a18    else:19      return codegen("(-"+a.str+")")20  def __add__(a, b):21    if(is_zero(a)):22      return b23    elif(is_zero(b)):24      return a25    elif(a == b):26      return codegen(2)*a27    else:28      return codegen("("+a.str+"+"+b.str+")")29  def __sub__(a, b):30    if(is_zero(a)):31      return codegen("-"+b.str)32    elif(is_zero(b)):33      return a34    elif(a == b):35      return codegen(0)36    else:37      return codegen("("+a.str+"-"+b.str+")")38  def __mul__(a, b):39    if(is_zero(a) or is_zero(b)):40      return codegen(0)41    elif(is_one(b)):42      return a43    elif(is_one(a)):44      return b45    elif(a == b):46      return a**codegen(2)47    else:48      return codegen("("+a.str+"*"+b.str+")")49  def __div__(a, b):50    if(is_zero(a)):51      return codegen(0)52    elif(is_zero(b)):53      raise(ZeroDivisionError("division by zero"))54    elif(is_one(b)):55      return a56    elif(a == b):57      return codegen(1)58    else:59      return codegen("("+a.str+"/"+b.str+")")60  def __pow__(a, b):61    if(is_zero(a) or is_one(a) or is_one(b)):62      return a63    elif(is_zero(b)):64      return codegen(1)65    else:66      return codegen("("+a.str+"**"+b.str+")")67  def exp(a):68    return codegen("exp("+a.str+")")69  def log(a):70    return codegen("log("+a.str+")")71  def sin(a):72    return codegen("sin("+a.str+")")73  def cos(a):74    return codegen("cos("+a.str+")")75  def tan(a):76    return codegen("tan("+a.str+")")77  def cot(a):...

gru.py

Source:gru.py

1from torch import nn2class GRU(nn.Module):3    def __init__(self, state_space, hidden_size, enable_recurrent = True):4        super(GRU, self).__init__()5        self.enable_recurrent = enable_recurrent6        self.hidden_size = hidden_size7        if enable_recurrent:8            self.gru = nn.GRU(state_space, hidden_size)9            for name, param in self.gru.named_parameters():10                if 'bias' in name: nn.init.constant_(param, 0)11                elif 'weight' in name: nn.init.orthogonal_(param)12    def forward(self, states, hxs, masks):13        if inputs.size(0) == hxs.size(0):14            x, hxs = self.gru(inputs.unsqueeze(0), (hxs * masks).unsqueeze(0))15            x, hxs = x.squeeze(0), hxs.squeeze(0)16        else:17            n = hxs.size(0)18            t = int(x.size(0) / n)19            x = inputs.view(t, n, inputs.size(1))20            masks = masks.view(t, n)21            is_zero = ((mask[1:] == 0.0)\22                    .any(dim = -1).nonzero().squeeze().cpu())23            if is_zero.dim() == 0: is_zero = [is_zero.item() + 1]24            else: is_zero = (is_zero + 1).numpy().tolist()25            is_zero = [0] + is_zero + [t]26            hxs = hxs.unsqueeze(0)27            outputs = []28            for i in range(len(is_zero) - 1):29                start = is_zero[i]30                end = is_zero[i + 1]31                score, hxs = self.gru(x[start: end],32                        hxs * mask[start].view(1, -1, 1))33                outputs.append(score)34            x = torch.cat(outputs, dim = 0)35            x = x.view(t * n, -1)36            hxs = hxs.squeeze(0)...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.