我们从Python开源项目中，提取了以下50个代码示例，用于说明如何使用numpy.ptp()。
def compute_group(cls, data, scales, **params): n = len(data) if n < 3: return pd.DataFrame() weight = data.get('weight') if params['trim']: range_y = data['y'].min(), data['y'].max() else: range_y = scales.y.dimension() dens = compute_density(data['y'], weight, range_y, **params) dens['y'] = dens['x'] dens['x'] = np.mean([data['x'].min(), data['x'].max()]) # Compute width if x has multiple values if len(np.unique(data['x'])) > 1: dens['width'] = np.ptp(data['x']) * 0.9 return dens
def draw_group(data, panel_params, coord, ax, **params): data = coord.transform(data, panel_params) fill = to_rgba(data['fill'], data['alpha']) color = to_rgba(data['color'], data['alpha']) ranges = coord.range(panel_params) # For perfect circles the width/height of the circle(ellipse) # should factor in the dimensions of axes bbox = ax.get_window_extent().transformed( ax.figure.dpi_scale_trans.inverted()) ax_width, ax_height = bbox.width, bbox.height factor = ((ax_width/ax_height) * np.ptp(ranges.y)/np.ptp(ranges.x)) size = data.loc[0, 'binwidth'] * params['dotsize'] offsets = data['stackpos'] * params['stackratio'] if params['binaxis'] == 'x': width, height = size, size*factor xpos, ypos = data['x'], data['y'] + height*offsets elif params['binaxis'] == 'y': width, height = size/factor, size xpos, ypos = data['x'] + width*offsets, data['y'] circles = [] for xy in zip(xpos, ypos): patch = mpatches.Ellipse(xy, width=width, height=height) circles.append(patch) coll = mcoll.PatchCollection(circles, edgecolors=color, facecolors=fill) ax.add_collection(coll)
def fit(self, X, y=None): """Fit it. Parameters ---------- X : array, shape (n_epochs, n_times) The data for one channel. y : None Redundant. Necessary to be compatible with sklearn API. """ deltas = np.ptp(X, axis=1) self.deltas_ = deltas keep = deltas <= self.thresh # XXX: actually go over all the folds before setting the min # in skopt. Otherwise, may confuse skopt. if self.thresh < np.min(np.ptp(X, axis=1)): assert np.sum(keep) == 0 keep = deltas <= np.min(np.ptp(X, axis=1)) self.mean_ = _slicemean(X, keep, axis=0) return self
def _vote_bad_epochs(self, epochs): """Each channel votes for an epoch as good or bad. Parameters ---------- epochs : instance of mne.Epochs The epochs object for which bad epochs must be found. """ n_epochs = len(epochs) picks = _handle_picks(info=epochs.info, picks=self.picks) drop_log = np.zeros((n_epochs, len(epochs.ch_names))) bad_sensor_counts = np.zeros((len(epochs), )) ch_names = [epochs.ch_names[p] for p in picks] deltas = np.ptp(epochs.get_data()[:, picks], axis=-1).T threshes = [self.threshes_[ch_name] for ch_name in ch_names] for ch_idx, (delta, thresh) in enumerate(zip(deltas, threshes)): bad_epochs_idx = np.where(delta > thresh)[0] # TODO: combine for different ch types bad_sensor_counts[bad_epochs_idx] += 1 drop_log[bad_epochs_idx, picks[ch_idx]] = 1 return drop_log, bad_sensor_counts
def extend_limits(values, fraction=0.10, tolerance=1e-2): """ Extend the values of a list by a fractional amount """ values = np.array(values) finite_indices = np.isfinite(values) if np.sum(finite_indices) == 0: raise ValueError("no finite values provided") lower_limit, upper_limit = np.min(values[finite_indices]), np.max(values[finite_indices]) ptp_value = np.ptp([lower_limit, upper_limit]) new_limits = lower_limit - fraction * ptp_value, ptp_value * fraction + upper_limit if np.abs(new_limits[0] - new_limits[1]) < tolerance: if np.abs(new_limits[0]) < tolerance: # Arbitrary limits, since we"ve just been passed zeros offset = 1 else: offset = np.abs(new_limits[0]) * fraction new_limits = new_limits[0] - offset, offset + new_limits[0] return np.array(new_limits)
def calculate_fractional_overlap(interest_range, comparison_range): """ Calculate how much of the range of interest overlaps with the comparison range. """ if not (interest_range[-1] >= comparison_range[0] \ and comparison_range[-1] >= interest_range[0]): return 0.0 # No overlap elif (interest_range[0] >= comparison_range[0] \ and interest_range[-1] <= comparison_range[-1]): return 1.0 # Total overlap else: # Some overlap. Which side? if interest_range[0] < comparison_range[0]: # Left hand side width = interest_range[-1] - comparison_range[0] else: # Right hand side width = comparison_range[-1] - interest_range[0] return width/np.ptp(interest_range) # Fractional overlap
def update_roi_xy_size(self): """ Update the cursor size showing the optimizer scan area for the XY image. """ hpos = self.roi_xy.pos()[0] vpos = self.roi_xy.pos()[1] hsize = self.roi_xy.size()[0] vsize = self.roi_xy.size()[1] hcenter = hpos + 0.5 * hsize vcenter = vpos + 0.5 * vsize if self.adjust_cursor_roi: newsize = self._optimizer_logic.refocus_XY_size else: viewrange = self.xy_image.getViewBox().viewRange() newsize = np.sqrt(np.sum(np.ptp(viewrange, axis=1)**2)) / 20 self.roi_xy.setSize([newsize, newsize]) self.roi_xy.setPos([hcenter - newsize / 2, vcenter - newsize / 2])
def update_roi_depth_size(self): """ Update the cursor size showing the optimizer scan area for the X-depth image. """ hpos = self.roi_depth.pos()[0] vpos = self.roi_depth.pos()[1] hsize = self.roi_depth.size()[0] vsize = self.roi_depth.size()[1] hcenter = hpos + 0.5 * hsize vcenter = vpos + 0.5 * vsize if self.adjust_cursor_roi: newsize_h = self._optimizer_logic.refocus_XY_size newsize_v = self._optimizer_logic.refocus_Z_size else: viewrange = self.depth_image.getViewBox().viewRange() newsize = np.sqrt(np.sum(np.ptp(viewrange, axis=1)**2)) / 20 newsize_h = newsize newsize_v = newsize self.roi_depth.setSize([newsize_h, newsize_v]) self.roi_depth.setPos([hcenter - newsize_h / 2, vcenter - newsize_v / 2])
def plane_fit(points, tolerance=None): ''' Given a set of points, find an origin and normal using least squares Arguments --------- points: (n,3) tolerance: how non-planar the result can be without raising an error Returns --------- C: (3) point on the plane N: (3) normal vector ''' C = points[0] x = points - C M = np.dot(x.T, x) N = np.linalg.svd(M)[0][:,-1] if not (tolerance is None): normal_range = np.ptp(np.dot(N, points.T)) if normal_range > tol.planar: log.error('Points have peak to peak of %f', normal_range) raise ValueError('Plane outside tolerance!') return C, N
def plot_epipolar_line(p1, p2, F, show_epipole=False): """ Plot the epipole and epipolar line F*x=0 in an image given the corresponding points. F is the fundamental matrix and p2 are the point in the other image. """ lines = np.dot(F, p2) pad = np.ptp(p1, 1) * 0.01 mins = np.min(p1, 1) maxes = np.max(p1, 1) # epipolar line parameter and values xpts = np.linspace(mins[0] - pad[0], maxes[0] + pad[0], 100) for line in lines.T: ypts = np.asarray([(line[2] + line[0] * p) / (-line[1]) for p in xpts]) valid_idx = ((ypts >= mins[1] - pad[1]) & (ypts <= maxes[1] + pad[1])) plt.plot(xpts[valid_idx], ypts[valid_idx], linewidth=1) plt.plot(p1[0], p1[1], 'ro') if show_epipole: epipole = compute_epipole(F) plt.plot(epipole[0] / epipole[2], epipole[1] / epipole[2], 'r*')
def startModulation(self, radiusInMilliRad, frequencyInHz, centerInMilliRad): self._origTargetPosition= centerInMilliRad self.stopModulation() periodInSec= 1./ frequencyInHz assert np.ptp(self._ctrl.getWaveGeneratorTableRate()) == 0, \ "wave generator table rate must be the same for every table" wgtr= self._ctrl.getWaveGeneratorTableRate()[0] timestep= self._ctrl.getServoUpdateTimeInSeconds() * wgtr lengthInPoints= periodInSec/ timestep peakOfTheSineCurve= self._milliRadToGcsUnits( self.getTargetPosition() + radiusInMilliRad) offsetOfTheSineCurve= self._milliRadToGcsUnits( self.getTargetPosition() - radiusInMilliRad) amplitudeOfTheSineCurve= peakOfTheSineCurve - offsetOfTheSineCurve wavelengthOfTheSineCurveInPoints= periodInSec/ timestep startPoint= np.array([0, 0.25])* wavelengthOfTheSineCurveInPoints curveCenterPoint= 0.5* wavelengthOfTheSineCurveInPoints self._ctrl.clearWaveTableData([1, 2, 3]) self._ctrl.setSinusoidalWaveform( 1, WaveformGenerator.CLEAR, lengthInPoints, amplitudeOfTheSineCurve[0], offsetOfTheSineCurve[0], wavelengthOfTheSineCurveInPoints, startPoint[0], curveCenterPoint) self._ctrl.setSinusoidalWaveform( 2, WaveformGenerator.CLEAR, lengthInPoints, amplitudeOfTheSineCurve[1], offsetOfTheSineCurve[1], wavelengthOfTheSineCurveInPoints, startPoint[1], curveCenterPoint) self._ctrl.setConnectionOfWaveTableToWaveGenerator([1, 2], [1, 2]) self._ctrl.setWaveGeneratorStartStopMode([1, 1, 0]) self._modulationEnabled= True
def compute_group(cls, data, scales, **params): labels = ['x', 'y'] X = np.array(data[labels]) res = boxplot_stats(X, whis=params['coef'], labels=labels)[1] try: n = data['weight'].sum() except KeyError: n = len(data['y']) if len(np.unique(data['x'])) > 1: width = np.ptp(data['x']) * 0.9 else: width = params['width'] if pdtypes.is_categorical(data['x']): x = data['x'].iloc[0] else: x = np.mean([data['x'].min(), data['x'].max()]) d = {'ymin': res['whislo'], 'lower': res['q1'], 'middle': [res['med']], 'upper': res['q3'], 'ymax': res['whishi'], 'outliers': [res['fliers']], 'notchupper': res['med']+1.58*res['iqr']/np.sqrt(n), 'notchlower': res['med']-1.58*res['iqr']/np.sqrt(n), 'x': x, 'width': width, 'relvarwidth': np.sqrt(n)} return pd.DataFrame(d)
def test_ptp(self): a = [3, 4, 5, 10, -3, -5, 6.0] assert_equal(np.ptp(a, axis=0), 15.0)
def _phampcheck(self, pha, amp, axis): """Check phase and amplitude values.""" # Shape checking : if pha.ndim != amp.ndim: raise ValueError("pha and amp must have the same number of " "dimensions.") # Force phase / amplitude to be at least (1, N) : if (pha.ndim == 1) and (amp.ndim == 1): pha = pha.reshape(1, -1) amp = amp.reshape(1, -1) axis = 1 # Check if the phase is in radians : if np.ptp(pha) > 2 * np.pi: raise ValueError("Your phase is probably in degrees and should be" " converted in radians using either np.degrees or" " np.deg2rad.") # Check if the phase/amplitude have the same number of points on axis: if pha.shape[axis] != amp.shape[axis]: phan, ampn = pha.shape[axis], amp.shape[axis] raise ValueError("The phase (" + str(phan) + ") and the amplitude " "(" + str(ampn) + ") do not have the same number" " of points on the specified axis (" + str(axis) + ").") # Force the phase to be in [-pi, pi] : pha = (pha + np.pi) % (2 * np.pi) - np.pi return pha, amp, axis ########################################################################### # PROPERTIES ########################################################################### # ----------- IDPAC -----------
def _postprocess_contours(self, index, times, freqs, salience): """Remove contours that are too short. Parameters ---------- index : np.array array of contour numbers times : np.array array of contour times freqs : np.array array of contour frequencies salience : np.array array of contour salience values Returns ------- index_pruned : np.array Pruned array of contour numbers times_pruned : np.array Pruned array of contour times freqs_pruned : np.array Pruned array of contour frequencies salience_pruned : np.array Pruned array of contour salience values """ keep_index = np.ones(times.shape).astype(bool) for i in set(index): this_idx = (index == i) if np.ptp(times[this_idx]) <= self.min_contour_len: keep_index[this_idx] = False return (index[keep_index], times[keep_index], freqs[keep_index], salience[keep_index])
def fit(self, X, y=None): """Fit it.""" if self.n_channels is None or self.n_times is None: raise ValueError('Cannot fit without knowing n_channels' ' and n_times') X = X.reshape(-1, self.n_channels, self.n_times) deltas = np.array([np.ptp(d, axis=1) for d in X]) epoch_deltas = deltas.max(axis=1) keep = epoch_deltas <= self.thresh self.mean_ = _slicemean(X, keep, axis=0) return self
def _get_epochs_interpolation(self, epochs, drop_log, ch_type, verbose='progressbar'): """Interpolate the bad epochs.""" # 1: bad segment, # 2: interpolated fix_log = drop_log.copy() ch_names = epochs.ch_names non_picks = np.setdiff1d(range(epochs.info['nchan']), self.picks) interp_channels = list() n_interpolate = self.n_interpolate[ch_type] for epoch_idx in range(len(epochs)): n_bads = drop_log[epoch_idx, self.picks].sum() if n_bads == 0: continue else: if n_bads <= n_interpolate: interp_chs_mask = drop_log[epoch_idx] == 1 else: # get peak-to-peak for channels in that epoch data = epochs[epoch_idx].get_data()[0] peaks = np.ptp(data, axis=-1) peaks[non_picks] = -np.inf # find channels which are bad by rejection threshold interp_chs_mask = drop_log[epoch_idx] == 1 # ignore good channels peaks[~interp_chs_mask] = -np.inf # find the ordering of channels amongst the bad channels sorted_ch_idx_picks = np.argsort(peaks)[::-1] # then select only the worst n_interpolate channels interp_chs_mask[ sorted_ch_idx_picks[n_interpolate:]] = False fix_log[epoch_idx][interp_chs_mask] = 2 interp_chs = np.where(interp_chs_mask)[0] interp_chs = [ch_name for idx, ch_name in enumerate(ch_names) if idx in interp_chs] interp_channels.append(interp_chs) return interp_channels, fix_log
def normalizeData(X): mean = [] data_range = [] mean.append(np.mean(X[:,1])) mean.append(np.mean(X[:,2])) data_range = np.ptp(X,axis=0)[-2:] #print(mean,data_range) for i in range(len(X)): X[:,1][i] = (X[:,1][i] - float(mean[0]))/float(data_range[0]) X[:,2][i] = (X[:,2][i] - float(mean[1]))/float(data_range[1]) return X
def FeatureScaling(X): mean = [] data_range = [] X1 = np.zeros((len(X),X.shape[1])) mean.append(np.mean(X[:,1])) mean.append(np.mean(X[:,2])) data_range = np.ptp(X,axis=0)[-2:] #print(mean) print(data_range) for i in range(len(X)): X1[:,0][i] = (X[:,0][i] - mean[0])/data_range[0] X1[:,1][i] = (X[:,1][i] - mean[1])/data_range[1] return X1
def neighbours(self, effective_temperature, surface_gravity, metallicity, N, scales=None): """ Return indices of the `N`th-nearest neighbours in the grid. The three parameters are scaled by the peak-to-peak range in the grid, unless `scales` are indicates. :param effective_temperature: The effective temperature of the star. :param surface_gravity: The surface gravity of the star. :param metallicity: The metallicity of the star. :param N: The number of neighbouring indices to return. :returns: An array of length `N` that contains the indices of the closest neighbours in the grid. """ point = np.array([effective_temperature, surface_gravity, metallicity]) if scales is None: scales = np.ptp(self._grid, axis=0) distance = np.sum(((self._grid - point)/scales)**2, axis=1) return np.argsort(distance)[:N]
def nearest_neighbours(self, point, n): """ Return the indices of the n nearest neighbours to the point. """ stellar_parameters = _recarray_to_array(self.stellar_parameters) distances = np.sum(((point - stellar_parameters) \ / np.ptp(stellar_parameters, axis=0))**2, axis=1) return distances.argsort()[:n]
def figure_mouse_pick(self, event): """ Trigger for when the mouse is used to select an item in the figure. :param event: The matplotlib event. """ ycol = "abundance" xcol = { self.ax_excitation_twin: "expot", self.ax_line_strength_twin: "reduced_equivalent_width" }[event.inaxes] xscale = np.ptp(event.inaxes.get_xlim()) yscale = np.ptp(event.inaxes.get_ylim()) try: distance = np.sqrt( ((self._state_transitions[ycol] - event.ydata)/yscale)**2 \ + ((self._state_transitions[xcol] - event.xdata)/xscale)**2) except AttributeError: # Stellar parameters have not been measured yet return None index = np.nanargmin(distance) # Because the state transitions are linked to the parent source model of # the table view, we will have to get the proxy index. proxy_index = self.table_view.model().mapFromSource( self.proxy_spectral_models.sourceModel().createIndex(index, 0)).row() self.table_view.selectRow(proxy_index) return None
def normalize(vec): """ Given an input vector normalize the vector Parameters ========== vec : array_like input vector to normalize Returns ======= out : array_like normalized vector Examples ======== >>> import spacepy.toolbox as tb >>> tb.normalize([1,2,3]) [0.0, 0.5, 1.0] """ # check to see if vec is numpy array, this is fastest if isinstance(vec, np.ndarray): out = (vec - vec.min())/np.ptp(vec) else: vecmin = np.min(vec) ptp = np.ptp(vec) out = [(val - vecmin)/ptp for val in vec] return out
def test_ptp(self): N = 1000 arr = np.random.randn(N) ser = Series(arr) self.assertEqual(np.ptp(ser), np.ptp(arr)) # GH11163 s = Series([3, 5, np.nan, -3, 10]) self.assertEqual(s.ptp(), 13) self.assertTrue(pd.isnull(s.ptp(skipna=False))) mi = pd.MultiIndex.from_product([['a', 'b'], [1, 2, 3]]) s = pd.Series([1, np.nan, 7, 3, 5, np.nan], index=mi) expected = pd.Series([6, 2], index=['a', 'b'], dtype=np.float64) self.assert_series_equal(s.ptp(level=0), expected) expected = pd.Series([np.nan, np.nan], index=['a', 'b']) self.assert_series_equal(s.ptp(level=0, skipna=False), expected) with self.assertRaises(ValueError): s.ptp(axis=1) s = pd.Series(['a', 'b', 'c', 'd', 'e']) with self.assertRaises(TypeError): s.ptp() with self.assertRaises(NotImplementedError): s.ptp(numeric_only=True)
def _get_indice(cls, w, flux, blue, red, band=None, unit='ew', degree=1, **kwargs): """ compute spectral index after continuum subtraction Parameters ---------- w: ndarray (nw, ) array of wavelengths in AA flux: ndarray (N, nw) array of flux values for different spectra in the series blue: tuple(2) selection for blue continuum estimate red: tuple(2) selection for red continuum estimate band: tuple(2), optional select region in this band only. default is band = (min(blue), max(red)) unit: str `ew` or `mag` wether equivalent width or magnitude degree: int (default 1) degree of the polynomial fit to the continuum Returns ------- ew: ndarray (N,) equivalent width array """ wi, fi = cls.continuum_normalized_region_around_line(w, flux, blue, red, band=band, degree=degree) if unit in (0, 'ew', 'EW'): return np.trapz(1. - fi, wi, axis=-1) else: m = np.trapz(fi, wi, axis=-1) m = -2.5 * np.log10(m / np.ptp(wi)) return m
def test_basic(self): a = [3, 4, 5, 10, -3, -5, 6.0] assert_equal(np.ptp(a, axis=0), 15.0) b = [[3, 6.0, 9.0], [4, 10.0, 5.0], [8, 3.0, 2.0]] assert_equal(np.ptp(b, axis=0), [5.0, 7.0, 7.0]) assert_equal(np.ptp(b, axis=-1), [6.0, 6.0, 6.0])
def plot_cdf(x, copy=True, fractional=True, **kwargs): """ Add a log-log CCDF plot to the current axes. Arguments --------- x : array_like The data to plot copy : boolean copy input array in a new object before sorting it. If data is a *very* large, the copy can avoided by passing False to this parameter. fractional : boolean compress the data by means of fractional ranking. This collapses the ranks from multiple, identical observations into their midpoint, thus producing smaller figures. Note that the resulting plot will NOT be the exact CCDF function, but an approximation. Additional keyword arguments are passed to `matplotlib.pyplot.loglog`. Returns a matplotlib axes object. """ N = float(len(x)) if copy: x = x.copy() x.sort() if fractional: t = [] for x, chunk in groupby(enumerate(x, 1), itemgetter(1)): xranks, _ = zip(*list(chunk)) t.append((float(x), xranks[0] + np.ptp(xranks) / 2.0)) t = np.asarray(t) else: t = np.c_[np.asfarray(x), np.arange(N) + 1] if 'ax' not in kwargs: ax = plt.gca() else: ax = kwargs.pop('ax') ax.loglog(t[:, 0], (N - t[:, 1]) / N, 'ow', **kwargs) return ax
def test_integrate(): subslice = slice(100,200) wvln = np.linspace(1000., 4000., 1024) flux = np.zeros_like(wvln) flux[subslice] = 1./np.ptp(wvln[subslice]) # so the integral is 1 s = Spectrum(wvln*u.angstrom, flux*u.erg/u.cm**2/u.angstrom) # the integration grid is a sub-section of the full wavelength array wvln_grid = s.wavelength[subslice] i_flux = s.integrate(wvln_grid) assert np.allclose(i_flux.value, 1.) # "close" because this is float comparison
def is_circle(points, scale, verbose=True): ''' Given a set of points, quickly determine if they represent a circle or not. ''' # make sure input is a numpy array points = np.asanyarray(points) scale = float(scale) # can only be a circle if the first and last point are the # same (AKA is a closed path) if np.linalg.norm(points[0] - points[-1]) > tol.merge: return None box = points.ptp(axis=0) # the bounding box size of the points # check aspect ratio as an early exit if the path is not a circle aspect = np.divide(*box) if np.abs(aspect - 1.0) > tol.aspect_frac: return None # fit a circle with tolerance checks CR = fit_circle_check(points, scale=scale) if CR is None: return None # return the circle as three control points control = angles_to_threepoint([0,np.pi*.5], *CR) return control
def print_confidence_interval(ci, tabs=''): """Pretty print confidence interval information""" ci = list(ci) ci += [np.ptp(ci)] print(tabs + 'Value: {1:.04f}'.format(*ci)) print(tabs + '95% Confidence Interval: ({0:.04f}, {2:.04f})'.format(*ci)) print(tabs + '\tCI Width: {3:.05f}'.format(*ci))
def ptp(a, axis=None, out=None): """ Range of values (maximum - minimum) along an axis. The name of the function comes from the acronym for 'peak to peak'. Parameters ---------- a : array_like Input values. axis : int, optional Axis along which to find the peaks. By default, flatten the array. out : array_like Alternative output array in which to place the result. It must have the same shape and buffer length as the expected output, but the type of the output values will be cast if necessary. Returns ------- ptp : ndarray A new array holding the result, unless `out` was specified, in which case a reference to `out` is returned. Examples -------- >>> x = np.arange(4).reshape((2,2)) >>> x array([[0, 1], [2, 3]]) >>> np.ptp(x, axis=0) array([2, 2]) >>> np.ptp(x, axis=1) array([1, 1]) """ return _wrapfunc(a, 'ptp', axis=axis, out=out)
def test_scalar(self): """ Should return 0 for all scalar """ x = scalar('x') p = ptp(x) f = theano.function([x], p) y = numpy.asarray(rand() * 2000 - 1000, dtype=config.floatX) result = f(y) numpyResult = numpy.ptp(y) self.assertTrue(numpy.array_equal(result, numpyResult))
def test_vector(self): x = vector('x') p = ptp(x, 0) f = theano.function([x], p) y = rand_ranged(-1000, 1000, [100]) result = f(y) numpyResult = numpy.ptp(y, 0) self.assertTrue(numpy.array_equal(result, numpyResult))
def test_matrix_first_axis(self): x = matrix('x') p = ptp(x, 1) f = theano.function([x], p) y = rand_ranged(-1000, 1000, [100, 100]) result = f(y) numpyResult = numpy.ptp(y, 1) self.assertTrue(numpy.array_equal(result, numpyResult))
def test_matrix_second_axis(self): x = matrix('x') p = ptp(x, 0) f = theano.function([x], p) y = rand_ranged(-1000, 1000, [100, 100]) result = f(y) numpyResult = numpy.ptp(y, 0) self.assertTrue(numpy.array_equal(result, numpyResult))
def test_matrix_neg_axis(self): x = matrix('x') p = ptp(x, -1) f = theano.function([x], p) y = rand_ranged(-1000, 1000, [100, 100]) result = f(y) numpyResult = numpy.ptp(y, -1) self.assertTrue(numpy.array_equal(result, numpyResult))
def test_interface(self): x = matrix('x') p = x.ptp(1) f = theano.function([x], p) y = rand_ranged(-1000, 1000, [100, 100]) result = f(y) numpyResult = numpy.ptp(y, 1) self.assertTrue(numpy.array_equal(result, numpyResult))
def has_constant(x): """ Parameters ---------- x: ndarray Array to be checked for a constant (n,k) Returns ------- const : bool Flag indicating whether x contains a constant or has column span with a constant loc : int Column location of constant """ if np.any(np.all(x == 1, axis=0)): loc = np.argwhere(np.all(x == 1, axis=0)) return True, int(loc) if np.any((np.ptp(x, axis=0) == 0) & ~np.all(x == 0, axis=0)): loc = np.any((np.ptp(x, axis=0) == 0) & ~np.all(x == 0, axis=0)) loc = np.argwhere(loc) return True, int(loc) n = x.shape[0] aug_rank = matrix_rank(np.c_[np.ones((n, 1)), x]) rank = matrix_rank(x) has_const = bool(aug_rank == rank) loc = None if has_const: out = np.linalg.lstsq(x, np.ones((n, 1))) beta = out[0].ravel() loc = np.argmax(np.abs(beta) * x.var(0)) return has_const, loc
def test_ids(panel): data = PanelData(panel) eids = data.entity_ids assert eids.shape == (77, 1) assert len(np.unique(eids)) == 11 for i in range(0, len(eids), 7): assert np.ptp(eids[i:i + 7]) == 0 assert np.all((eids[i + 8:] - eids[i]) != 0) tids = data.time_ids assert tids.shape == (77, 1) assert len(np.unique(tids)) == 7 for i in range(0, 11): assert np.ptp(tids[i::7]) == 0
def test_neighbors_accuracy_with_n_candidates(): # Checks whether accuracy increases as `n_candidates` increases. n_candidates_values = np.array([.1, 50, 500]) n_samples = 100 n_features = 10 n_iter = 10 n_points = 5 rng = np.random.RandomState(42) accuracies = np.zeros(n_candidates_values.shape[0], dtype=float) X = rng.rand(n_samples, n_features) for i, n_candidates in enumerate(n_candidates_values): lshf = LSHForest(n_candidates=n_candidates) ignore_warnings(lshf.fit)(X) for j in range(n_iter): query = X[rng.randint(0, n_samples)].reshape(1, -1) neighbors = lshf.kneighbors(query, n_neighbors=n_points, return_distance=False) distances = pairwise_distances(query, X, metric='cosine') ranks = np.argsort(distances)[0, :n_points] intersection = np.intersect1d(ranks, neighbors).shape[0] ratio = intersection / float(n_points) accuracies[i] = accuracies[i] + ratio accuracies[i] = accuracies[i] / float(n_iter) # Sorted accuracies should be equal to original accuracies assert_true(np.all(np.diff(accuracies) >= 0), msg="Accuracies are not non-decreasing.") # Highest accuracy should be strictly greater than the lowest assert_true(np.ptp(accuracies) > 0, msg="Highest accuracy is not strictly greater than lowest.")
def test_neighbors_accuracy_with_n_estimators(): # Checks whether accuracy increases as `n_estimators` increases. n_estimators = np.array([1, 10, 100]) n_samples = 100 n_features = 10 n_iter = 10 n_points = 5 rng = np.random.RandomState(42) accuracies = np.zeros(n_estimators.shape[0], dtype=float) X = rng.rand(n_samples, n_features) for i, t in enumerate(n_estimators): lshf = LSHForest(n_candidates=500, n_estimators=t) ignore_warnings(lshf.fit)(X) for j in range(n_iter): query = X[rng.randint(0, n_samples)].reshape(1, -1) neighbors = lshf.kneighbors(query, n_neighbors=n_points, return_distance=False) distances = pairwise_distances(query, X, metric='cosine') ranks = np.argsort(distances)[0, :n_points] intersection = np.intersect1d(ranks, neighbors).shape[0] ratio = intersection / float(n_points) accuracies[i] = accuracies[i] + ratio accuracies[i] = accuracies[i] / float(n_iter) # Sorted accuracies should be equal to original accuracies assert_true(np.all(np.diff(accuracies) >= 0), msg="Accuracies are not non-decreasing.") # Highest accuracy should be strictly greater than the lowest assert_true(np.ptp(accuracies) > 0, msg="Highest accuracy is not strictly greater than lowest.")
def ptp(a, axis=None, out=None): """ Range of values (maximum - minimum) along an axis. The name of the function comes from the acronym for 'peak to peak'. Parameters ---------- a : array_like Input values. axis : int, optional Axis along which to find the peaks. By default, flatten the array. out : array_like Alternative output array in which to place the result. It must have the same shape and buffer length as the expected output, but the type of the output values will be cast if necessary. Returns ------- ptp : ndarray A new array holding the result, unless `out` was specified, in which case a reference to `out` is returned. Examples -------- >>> x = np.arange(4).reshape((2,2)) >>> x array([[0, 1], [2, 3]]) >>> np.ptp(x, axis=0) array([2, 2]) >>> np.ptp(x, axis=1) array([1, 1]) """ try: ptp = a.ptp except AttributeError: return _wrapit(a, 'ptp', axis, out) return ptp(axis, out)
def _plot_histogram(params): """Function for plotting histogram of peak-to-peak values.""" import matplotlib.pyplot as plt epochs = params['epochs'] p2p = np.ptp(epochs.get_data(), axis=2) types = list() data = list() if 'eeg' in params['types']: eegs = np.array([p2p.T[i] for i, x in enumerate(params['types']) if x == 'eeg']) data.append(eegs.ravel()) types.append('eeg') if 'mag' in params['types']: mags = np.array([p2p.T[i] for i, x in enumerate(params['types']) if x == 'mag']) data.append(mags.ravel()) types.append('mag') if 'grad' in params['types']: grads = np.array([p2p.T[i] for i, x in enumerate(params['types']) if x == 'grad']) data.append(grads.ravel()) types.append('grad') params['histogram'] = plt.figure() scalings = _handle_default('scalings') units = _handle_default('units') titles = _handle_default('titles') colors = _handle_default('color') for idx in range(len(types)): ax = plt.subplot(len(types), 1, idx + 1) plt.xlabel(units[types[idx]]) plt.ylabel('count') color = colors[types[idx]] rej = None if epochs.reject is not None and types[idx] in epochs.reject.keys(): rej = epochs.reject[types[idx]] * scalings[types[idx]] rng = [0., rej * 1.1] else: rng = None plt.hist(data[idx] * scalings[types[idx]], bins=100, color=color, range=rng) if rej is not None: ax.plot((rej, rej), (0, ax.get_ylim()[1]), color='r') plt.title(titles[types[idx]]) params['histogram'].suptitle('Peak-to-peak histogram', y=0.99) params['histogram'].subplots_adjust(hspace=0.6) try: params['histogram'].show(warn=False) except Exception: pass if params['fig_proj'] is not None: params['fig_proj'].canvas.draw()
def relim_axes(axes, percent=20): """ Generate new axes for a matplotlib axes based on the collections present. :param axes: The matplotlib axes. :param percent: [optional] The percent of the data to extend past the minimum and maximum data points. :returns: A two-length tuple containing the lower and upper limits in the x- and y-axis, respectively. """ data = np.vstack([item.get_offsets() for item in axes.collections \ if isinstance(item, PathCollection)]) if data.size == 0: return (None, None) data = data.reshape(-1, 2) x, y = data[:,0], data[:, 1] # Only use finite values. finite = np.isfinite(x*y) x, y = x[finite], y[finite] if x.size > 1: xlim = [ np.min(x) - np.ptp(x) * percent/100., np.max(x) + np.ptp(x) * percent/100., ] elif x.size == 0: xlim = None else: xlim = (x[0] - 1, x[0] + 1) if y.size > 1: ylim = [ np.min(y) - np.ptp(y) * percent/100., np.max(y) + np.ptp(y) * percent/100. ] elif y.size == 0: ylim = None else: ylim = (y[0] - 1, y[0] + 1) axes.set_xlim(xlim) axes.set_ylim(ylim) return (xlim, ylim)
