python类max()的实例源码

def compHistDistance(h1, h2):
  def normalize(h):    
    if np.sum(h) == 0: 
        return h
    else:
        return h / np.sum(h)

  def smoothstep(x, x_min=0., x_max=1., k=2.):
      m = 1. / (x_max - x_min)
      b = - m * x_min
      x = m * x + b
      return betainc(k, k, np.clip(x, 0., 1.))

  def fn(X, Y, k):
    return 4. * (1. - smoothstep(Y, 0, (1 - Y) * X + Y + .1)) \
      * np.sqrt(2 * X) * smoothstep(X, 0., 1. / k, 2) \
             + 2. * smoothstep(Y, 0, (1 - Y) * X + Y + .1) \
             * (1. - 2. * np.sqrt(2 * X) * smoothstep(X, 0., 1. / k, 2) - 0.5)

  h1 = normalize(h1)
  h2 = normalize(h2)

  return max(0, np.sum(fn(h2, h1, len(h1))))
  # return np.sum(np.where(h2 != 0, h2 * np.log10(h2 / (h1 + 1e-10)), 0))  # KL divergence

def effective_sample_size(x, mu, var, logger):
    """
    Calculate the effective sample size of sequence generated by MCMC.
    :param x:
    :param mu: mean of the variable
    :param var: variance of the variable
    :param logger: logg
    :return: effective sample size of the sequence
    Make sure that `mu` and `var` are correct!
    """
    # batch size, time, dimension
    b, t, d = x.shape
    ess_ = np.ones([d])
    for s in range(1, t):
        p = auto_correlation_time(x, s, mu, var)
        if np.sum(p > 0.05) == 0:
            break
        else:
            for j in range(0, d):
                if p[j] > 0.05:
                    ess_[j] += 2.0 * p[j] * (1.0 - float(s) / t)

    logger.info('ESS: max [%f] min [%f] / [%d]' % (t / np.min(ess_), t / np.max(ess_), t))
    return t / ess_

def score_fun_first_term(vals_hist,a_mid):
    sum = 0.0
    lim = int(np.max(vals_hist.keys()))
    for i in range(0, lim+1):
        if (vals_hist[i] > 0):
            inner_sum = 0.0
            for j in range(0, i):
                inner_sum += j/(1.0 + a_mid*j)
            sum += vals_hist[i]*inner_sum

    return sum 


##############################  
## in-line functions
##############################

def estimate_clipping_rect(projector, size):
    """
    Return:
    rect -- NSRect style 2d-tuple.
    flipped (bool) -- Whether y-axis is flipped.
    """
    # lt -> rt -> lb -> rb
    image_corners = [(0, 0), (size[0], 0), (0, size[1]), size]
    x_points = []
    y_points = []
    for corner in image_corners:
        x, y = map(int, projector.project_point(*corner))
        x_points.append(x)
        y_points.append(y)
    min_x = min(x_points)
    min_y = min(y_points)
    max_x = max(x_points)
    max_y = max(y_points)

    rect = ((min_x, min_y), (max_x - min_x, max_y - min_y))
    flipped = y_points[3] < 0

    return rect, flipped

def to_dim_times_two(self, bounds):
        """return boundaries in format ``[[lb0, ub0], [lb1, ub1], ...]``,
        as used by ``BoxConstraints...`` class.

        """
        if not bounds:
            b = [[None, None]]
        else:
            l = [None, None]  # figure out lenths
            for i in [0, 1]:
                try:
                    l[i] = len(bounds[i])
                except TypeError:
                    bounds[i] = [bounds[i]]
                    l[i] = 1
            b = []  # bounds in different format
            try:
                for i in range(max(l)):
                    b.append([bounds[0][i] if i < l[0] else None,
                              bounds[1][i] if i < l[1] else None])
            except (TypeError, IndexError):
                print("boundaries must be provided in the form " +
                      "[scalar_of_vector, scalar_or_vector]")
                raise
        return b

def alleviate_conditioning_in_coordinates(self, condition=1e8):
        """pass scaling from `C` to `sigma_vec`.

        As a result, `C` is a correlation matrix, i.e., all diagonal
        entries of `C` are `1`.
        """
        if max(self.dC) / min(self.dC) > condition:
            # allows for much larger condition numbers, if axis-parallel
            if hasattr(self, 'sm') and isinstance(self.sm, sampler.GaussFullSampler):
                old_coordinate_condition = max(self.dC) / min(self.dC)
                old_condition = self.sm.condition_number
                factors = self.sm.to_correlation_matrix()
                self.sigma_vec *= factors
                self.pc /= factors
                self._updateBDfromSM(self.sm)
                utils.print_message('\ncondition in coordinate system exceeded'
                                    ' %.1e, rescaled to %.1e, '
                                    '\ncondition changed from %.1e to %.1e'
                                      % (old_coordinate_condition, max(self.dC) / min(self.dC),
                                         old_condition, self.sm.condition_number),
                                    iteration=self.countiter)

def plot_axes_scaling(self, iabscissa=1):
        from matplotlib import pyplot
        if not hasattr(self, 'D'):
            self.load()
        dat = self
        if np.max(dat.D[:, 5:]) == np.min(dat.D[:, 5:]):
            pyplot.text(0, dat.D[-1, 5],
                        'all axes scaling values equal to %s'
                        % str(dat.D[-1, 5]),
                        verticalalignment='center')
            return self  # nothing interesting to plot
        self._enter_plotting()
        pyplot.semilogy(dat.D[:, iabscissa], dat.D[:, 5:], '-b')
        # pyplot.hold(True)
        pyplot.grid(True)
        ax = array(pyplot.axis())
        # ax[1] = max(minxend, ax[1])
        pyplot.axis(ax)
        pyplot.title('Principle Axes Lengths')
        # pyplot.xticks(xticklocs)
        self._xlabel(iabscissa)
        self._finalize_plotting()
        return self

def initwithsize(self, curshape, dim):
        # DIM-dependent initialization
        if self.dim != dim:
            scale = max(1, dim ** .5 / 8.)
            self.linearTF = scale * compute_rotation(self.rseed, dim)
            # if self.zerox:
            #    self.xopt = zeros(dim) # does not work here
            # else:
            # TODO: clean this line
            self.xopt = np.hstack(dot(self.linearTF, 0.5 * np.ones((dim, 1)) / scale ** 2))

        # DIM- and POPSI-dependent initialisations of DIM*POPSI matrices
        if self.lastshape != curshape:
            self.dim = dim
            self.lastshape = curshape
            self.arrxopt = resize(self.xopt, curshape)

def update_measure(self):
        """updated noise level measure using two fitness lists ``self.fit`` and
        ``self.fitre``, return ``self.noiseS, all_individual_measures``.

        Assumes that ``self.idx`` contains the indices where the fitness
        lists differ.

        """
        lam = len(self.fit)
        idx = np.argsort(self.fit + self.fitre)
        ranks = np.argsort(idx).reshape((2, lam))
        rankDelta = ranks[0] - ranks[1] - np.sign(ranks[0] - ranks[1])

        # compute rank change limits using both ranks[0] and ranks[1]
        r = np.arange(1, 2 * lam)  # 2 * lam - 2 elements
        limits = [0.5 * (Mh.prctile(np.abs(r - (ranks[0, i] + 1 - (ranks[0, i] > ranks[1, i]))),
                                      self.theta * 50) +
                         Mh.prctile(np.abs(r - (ranks[1, i] + 1 - (ranks[1, i] > ranks[0, i]))),
                                      self.theta * 50))
                    for i in self.idx]
        # compute measurement
        #                               max: 1 rankchange in 2*lambda is always fine
        s = np.abs(rankDelta[self.idx]) - Mh.amax(limits, 1)  # lives roughly in 0..2*lambda
        self.noiseS += self.cum * (np.mean(s) - self.noiseS)
        return self.noiseS, s

def logscale_img(img_array,
                 cap=255.0,
                 coeff=1000.0):
    '''
    This scales the image according to the relation:

    logscale_img = np.log(coeff*(img/max(img))+1)/np.log(coeff)

    Taken from the DS9 scaling algorithms page at:

    http://hea-www.harvard.edu/RD/ds9/ref/how.html

    According to that page:

    coeff = 1000.0 works well for optical images
    coeff = 100.0 works well for IR images

    '''

    logscaled_img = np.log(coeff*img_array/np.nanmax(img_array)+1)/np.log(coeff)
    return cap*logscaled_img

def genplot(x, y, fit, xdata=None, ydata=None, maxpts=10000):
    bin_range = (0, 360)
    a = (np.arange(*bin_range))
    f_a = nuth_func(a, fit[0], fit[1], fit[2])
    nuth_func_str = r'$y=%0.2f*cos(%0.2f-x)+%0.2f$' % tuple(fit)
    if xdata.size > maxpts:
        import random
        idx = random.sample(list(range(xdata.size)), 10000)
    else:
        idx = np.arange(xdata.size)
    f, ax = plt.subplots()
    ax.set_xlabel('Aspect (deg)')
    ax.set_ylabel('dh/tan(slope) (m)')
    ax.plot(xdata[idx], ydata[idx], 'k.', label='Orig pixels')
    ax.plot(x, y, 'ro', label='Bin median')
    ax.axhline(color='k')
    ax.plot(a, f_a, 'b', label=nuth_func_str)
    ax.set_xlim(*bin_range)
    pad = 0.2 * np.max([np.abs(y.min()), np.abs(y.max())])
    ax.set_ylim(y.min() - pad, y.max() + pad)
    ax.legend(prop={'size':8})
    return f 

#Function copied from from openPIV pyprocess

def center_clipping(x, percent=30):
    """
    Performs center clipping, a spectral whitening process

    need some type of spectrum flattening so that the
    speech signal more closely approximates a periodic impulse train

    Args:
        x       (array): signal data
        percent (float): percent threshold to clip

    Returns:
        cc         (array): center clipped signal
        clip_level (float): value of clipping
    """
    max_amp = np.max(np.abs(x))
    clip_level = max_amp * (percent / 100)
    positive_mask = x > clip_level
    negative_mask = x < -clip_level
    cc = np.zeros(x.shape)
    cc[positive_mask] = x[positive_mask] - clip_level
    cc[negative_mask] = x[negative_mask] + clip_level
    return cc, clip_level

def calc_tvd(sess,Generator,Data,N=50000,nbins=10):
    Xd=sess.run(Data.X,{Data.N:N})
    step,Xg=sess.run([Generator.step,Generator.X],{Generator.N:N})

    p_gen,_ = np.histogramdd(Xg,bins=nbins,range=[[0,1],[0,1],[0,1]],normed=True)
    p_dat,_ = np.histogramdd(Xd,bins=nbins,range=[[0,1],[0,1],[0,1]],normed=True)
    p_gen/=nbins**3
    p_dat/=nbins**3
    tvd=0.5*np.sum(np.abs( p_gen-p_dat ))
    mvd=np.max(np.abs( p_gen-p_dat ))

    return step,tvd, mvd

    s_tvd=make_summary(Data.name+'_tvd',tvd)
    s_mvd=make_summary(Data.name+'_mvd',mvd)

    return step,s_tvd,s_mvd
    #return make_summary('tvd/'+Generator.name,tvd)

def scatter2d(x,y,title='2dscatterplot',xlabel=None,ylabel=None):
    fig=plt.figure()
    plt.scatter(x,y)
    plt.title(title)
    if xlabel:
        plt.xlabel(xlabel)
    if ylabel:
        plt.ylabel(ylabel)

    if not 0<=np.min(x)<=np.max(x)<=1:
        raise ValueError('summary_scatter2d title:',title,' input x exceeded [0,1] range.\
                         min:',np.min(x),' max:',np.max(x))
    if not 0<=np.min(y)<=np.max(y)<=1:
        raise ValueError('summary_scatter2d title:',title,' input y exceeded [0,1] range.\
                         min:',np.min(y),' max:',np.max(y))

    plt.xlim([0,1])
    plt.ylim([0,1])
    return fig

def get_next_batch(experience, model, num_actions, gamma, batch_size):
    batch_indices = np.random.randint(low=0, high=len(experience),
                                      size=batch_size)
    batch = [experience[i] for i in batch_indices]
    X = np.zeros((batch_size, 80, 80, 4))
    Y = np.zeros((batch_size, num_actions))
    for i in range(len(batch)):
        s_t, a_t, r_t, s_tp1, game_over = batch[i]
        X[i] = s_t
        Y[i] = model.predict(s_t)[0]
        Q_sa = np.max(model.predict(s_tp1)[0])
        if game_over:
            Y[i, a_t] = r_t
        else:
            Y[i, a_t] = r_t + gamma * Q_sa
    return X, Y


############################# main ###############################

# initialize parameters

def forward(self, input):
        """During the forward pass, it inhibits all inhibitions below some 
        threshold :math:`?`, typically :math:`0`. In other words, it computes point-wise 

        .. math:: y=max(0,x) 

        Parameters
        ----------
        x : float32
            The activation (the summed, weighted input of a neuron).

        Returns
        -------
        float32
            The output of the rectify function applied to the activation.
        """
        self.last_forward = input
        return np.maximum(0.0, input)

def forward(self, input):
        """:math:`\\varphi(\\mathbf{x})_j =
        \\frac{e^{\mathbf{x}_j}}{\sum_{k=1}^K e^{\mathbf{x}_k}}`
        where :math:`K` is the total number of neurons in the layer. This
        activation function gets applied row-wise.

        Parameters
        ----------
        x : float32
            The activation (the summed, weighted input of a neuron).

        Returns
        -------
        float32 where the sum of the row is 1 and each single value is in [0, 1]
            The output of the softmax function applied to the activation.
        """
        assert np.ndim(input) == 2
        self.last_forward = input
        x = input - np.max(input, axis=1, keepdims=True)
        exp_x = np.exp(x)
        s = exp_x / np.sum(exp_x, axis=1, keepdims=True)
        return s

def main(max_iter):
    # prepare
    npdl.utils.random.set_seed(1234)

    # data
    digits = load_digits()

    X_train = digits.data
    X_train /= np.max(X_train)

    Y_train = digits.target
    n_classes = np.unique(Y_train).size

    # model
    model = npdl.model.Model()
    model.add(npdl.layers.Dense(n_out=500, n_in=64, activation=npdl.activations.ReLU()))
    model.add(npdl.layers.Dense(n_out=n_classes, activation=npdl.activations.Softmax()))
    model.compile(loss=npdl.objectives.SCCE(), optimizer=npdl.optimizers.SGD(lr=0.005))

    # train
    model.fit(X_train, npdl.utils.data.one_hot(Y_train), max_iter=max_iter, validation_split=0.1)

def _updateMaxTextSize(self, x):
        ## Informs that the maximum tick size orthogonal to the axis has
        ## changed; we use this to decide whether the item needs to be resized
        ## to accomodate.
        if self.orientation in ['left', 'right']:
            mx = max(self.textWidth, x)
            if mx > self.textWidth or mx < self.textWidth-10:
                self.textWidth = mx
                if self.style['autoExpandTextSpace'] is True:
                    self._updateWidth()
                    #return True  ## size has changed
        else:
            mx = max(self.textHeight, x)
            if mx > self.textHeight or mx < self.textHeight-10:
                self.textHeight = mx
                if self.style['autoExpandTextSpace'] is True:
                    self._updateHeight()
                    #return True  ## size has changed

def _updateHeight(self):
        if not self.isVisible():
            h = 0
        else:
            if self.fixedHeight is None:
                if not self.style['showValues']:
                    h = 0
                elif self.style['autoExpandTextSpace'] is True:
                    h = self.textHeight
                else:
                    h = self.style['tickTextHeight']
                h += self.style['tickTextOffset'][1] if self.style['showValues'] else 0
                h += max(0, self.style['tickLength'])
                if self.label.isVisible():
                    h += self.label.boundingRect().height() * 0.8
            else:
                h = self.fixedHeight

        self.setMaximumHeight(h)
        self.setMinimumHeight(h)
        self.picture = None