तरंग विश्लेषण। भाग 3

import pywt import numpy as np import matplotlib.pyplot as plt t = np.linspace(-1, 1, 200, endpoint=False) sig = np.cos(2 * np.pi * 7 * t) + np.real(np.exp(-7*(t-0.4)**2)*np.exp(1j*2*np.pi*2*(t-0.4))) widths = np.arange(1, 31) cwtmatr, freqs = pywt.cwt(sig, widths, 'cmor1-1.5') plt.imshow(cwtmatr, extent=[-1, 1, 1, 31], cmap='PRGn', aspect='auto', vmax=abs(cwtmatr).max(), vmin=-abs(cwtmatr).max()) # doctest: +SKIP plt.show() # doctest: +SKIP 

 from numpy import* from pylab import* import scaleogram as scg import pywt #  ,        #       'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') #    1  /  ns = 1024 time =arange(ns) scales = scg.periods2scales( arange(1, 40) ) p1=10; periodic1 = cos(2*pi/p1*time) * exp(-((time-ns/2)/200)**2) fig1, ax1 = subplots(1, 1, figsize=(6.9,2.9)); lines = ax1.plot(periodic1); ax1.set_xlim(0, len(time)) ax1.set_title("   ") fig1.tight_layout() ax2 = scg.cws(periodic1, scales=scales, figsize=(6.9,2.9)); txt = ax2.annotate("p1=%s"%p1, xy=(100, 10), bbox=dict(boxstyle="round4", fc="w")) tight_layout() print("    :", scg.get_default_wavelet(), "(", pywt.ContinuousWavelet(scg.get_default_wavelet()).family_name, ")") show() 

 from numpy import* import pandas as pd from pylab import* import scaleogram as scg #          #       'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') ##    1  /  ns = 1024 time = arange(ns) scales = scg.periods2scales(arange(1, 40)) pulses = zeros(ns, dtype=float32) steps = linspace(0, ns, 8) periods = [10, 20, 40] for i in range(0,3): step_mask = (time > steps[i*2+1]) & (time < steps[i*2+2]) pulses += cos(2*pi/periods[i]*time) * step_mask fig1, ax1 = subplots(1, 1, figsize=(7,3)); lines = ax1.plot(pulses); ax1.set_xlim(0, len(time)); ax1.set_title("     "); ax1.set_xlabel("Time") fig1.tight_layout() ax2 = scg.cws(pulses, scales=scales, figsize=(7,3)) for i in range(0, 3): txt = ax2.annotate("p%d=%ds"%(i+1,periods[i]), xy=(steps[i*2]+20, periods[i]), bbox=dict(boxstyle="round4", fc="w")) ax2.plot(steps[i*2+1]*np.ones(2), ax2.get_ylim(), '-w', alpha=0.5) ax2.plot(steps[i*2+2]*np.ones(2), ax2.get_ylim(), '-w', alpha=0.5) tight_layout() show() 

 from numpy import* import pandas as pd from pylab import* import scaleogram as scg scg.set_default_wavelet('cmor1-1.5') #          #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') ##    1  /  ns = 1024 time = arange(ns) scales = scg.periods2scales(arange(1, 40)) allwaves = np.zeros(ns, dtype=np.float32) periods = [10, 20, 40] for i in range(0,3): allwaves += cos(2*np.pi/periods[i]*time) fig1, ax1 = subplots(1, 1, figsize=(6.5,2)); lines = ax1.plot(allwaves); ax1.set_title("  ") ax1.set_xlim(0, len(time)); ax1.set_xlabel("") fig1.tight_layout() ax2 = scg.cws(allwaves, scales=scales, figsize=(7,2)) tight_layout() show() 

 from numpy import* from pylab import* import scipy.signal import scaleogram as scg scg.set_default_wavelet('cmor1-1.5') #          #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') ##    1  /  ns = 1024 time = arange(ns) scales = scg.periods2scales(arange(1, 40)) ptri = 30.0 # period in samples raising = 0.8 # fraction of the period raising triangle = scipy.signal.sawtooth(time/ptri*2*pi, raising) # plot the signal fig1, ax1 = subplots(1, 1, figsize=(7,3)); lines = ax1.plot(triangle); ax1.set_xlim(0, len(time)) ax1.set_title("triangle wave increasing 80% of the time with a 30s period") fig1.tight_layout() # and the scaleogram ax2 = scg.cws(triangle, scales=scales, figsize=(7,3)); txt = ax2.annotate("first harmonic", xy=(100, 30), bbox=dict(boxstyle="round4", fc="w")) txt = ax2.annotate("second harmonic", xy=(100, 15), bbox=dict(boxstyle="round4", fc="w")) tight_layout() show() 

 from numpy import* from pylab import* import scaleogram as scg scg.set_default_wavelet('cmor1-1.5') #          #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') ##    1  /  ns = 1024 time = arange(ns) scales = scg.periods2scales(arange(1, 40)) #       s1=1; s2=5; s3=10 events = [ns*0.1, ns*0.3, ns*0.7] bumps = exp(-((time-events[0])/s1)**2) + exp(-((time-events[1])/s2)**2) + \ exp(-((time-events[2])/s3)**2) #      w1=1; w2=5; w3= 50 steps = ((time > events[0]-w1) & (time < events[0]+w1)) + \ ((time > events[1]-w2) & (time < events[1]+w2)) + \ ((time > events[2]-w3) & (time < events[2]+w3)) #   fig1, (ax1, ax2) = subplots(1, 2, figsize=(11, 2)); ax1.set_title("Bumps"); ax2.set_title("Steps") lines = ax1.plot(bumps); ax1.set_xlim(0, len(time)) lines = ax2.plot(steps); ax2.set_xlim(0, len(time)) fig1.tight_layout() #   scales_bumps = scg.periods2scales(arange(1, 120, 2) ) fig2, (ax3, ax4) = subplots(1, 2, figsize=(14,3)); ax3 = scg.cws(bumps, scales=scales_bumps, ax=ax3) ax4 = scg.cws(steps, scales=scales_bumps, ax=ax4) for bmpw, stepw in [(2*s1, 2*w1), (2*s2,2*w2), (2*s3, 2*w3)]: ax3.plot(ax3.get_xlim(), bmpw*ones(2), 'w-', alpha=0.5) ax4.plot(ax4.get_xlim(), stepw*ones(2), 'w-', alpha=0.5) fig2.tight_layout() show() 

 from numpy import* from pylab import* import scaleogram as scg import random scg.set_default_wavelet('cmor1-1.5') #          #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') #    1  /  ns = 1024 time =arange(ns) scales = scg.periods2scales( arange(1, 40) ) noise =random.sample(range(ns), ns) fig1, ax1 = plt.subplots(1, 1, figsize=(6.2,2)); ax1.set_title("  ") lines = ax1.plot(time, noise) ax1.set_xlim(0, ns) fig1.tight_layout() ax2 = scg.cws(noise, scales=scales, figsize=(7,2)) tight_layout() show() 

 #!/usr/bin/env python # -*- coding: utf-8 -*- import numpy as np from pylab import* import scaleogram as scg import pywt signals = pywt.data.demo_signal('list') signal_length=1024 for signal in signals: if signal in ['Gabor', 'sineoneoverx']: x = pywt.data.demo_signal(signal) else: x = pywt.data.demo_signal(signal, signal_length) times=[] signals=[] for time, sig in enumerate(x.real): times.append(time) signals.append(sig) scg.set_default_wavelet('cmor1-1.5') scales = scg.periods2scales( arange(1, 40) ) fig1, ax1 = subplots(1, 1, figsize=(6.9,2.9)); lines = ax1.plot(signals); ax1.set_xlim(0, len(times)) ax1.set_title("%s"%signal) fig1.tight_layout() ax2 = scg.cws(signals, scales=scales, figsize=(6.9,2.9)); tight_layout() show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) fig = figure(figsize=(12,2)) lines = plot(births.index, births.values/1000, '-') xlim(datetime_lim) ylabel("Nb of birthes [k]"); title("Total births per day in the US (CDC)"); xlim = xlim() ax = scg.cws(births, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day") set_x_yearly(ax, len(births)) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() ax = scg.cws(births_normed, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day") set_x_yearly(ax, len(births)) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() scales = scg.periods2scales(np.arange(1, 365+100, 3)) ax = scg.cws(births_normed, scales=scales, figsize=(13.2, 5), xlabel="Year", ylabel="Nb of Day", clim=(0, 2500)) set_x_yearly(ax, len(births)) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() scales = scg.periods2scales(np.arange(1, 365+100, 3)) ax = scg.cws(births_normed, scales=scales, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day", clim=(0,2500), yscale='log') set_x_yearly(ax, len(births)) print("Number of available days:", len(births_normed)) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() #        log-Y-axis :-) scales = scg.periods2scales(np.logspace(np.log10(2), np.log10(365*3), 200)) #  CWT  ,       cwt = scg.CWT(births_normed, scales=scales) ax = scg.cws(cwt, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day", clim=(0,2500), yscale='log') set_x_yearly(ax, len(births)) bboxf = dict(boxstyle="round", facecolor="y", edgecolor="0.5") arrowpropsf = dict(facecolor='yellow', shrink=0.05) text = ax.annotate("unusual\nfeature", xy=(365*6, 15), xytext=(365*3, 50), bbox=bboxf, arrowprops=arrowpropsf) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg import pywt from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() #        log-Y-axis :-) scales = scg.periods2scales(np.logspace(np.log10(2), np.log10(365*3), 200)) #        log-Y-axis :-) cwt = scg.CWT(births_normed, scales=scales) #    : 1974-1976 fig = figure(figsize=(12,4)) lines =plot(births.index, births.values/1000, '-') xlim(pd.to_datetime("1974-01-01"), pd.to_datetime("1976-01-01")) ylabel("Nb of birthes [k]"); title("Total births per day bw 1974 and 1976"); #      ax = scg.cws(cwt, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day", clim=(0,2500), yscale='log') set_x_yearly(ax, len(births)) xlim = ax.set_xlim( (1974-1969)*365, (1976-1969)*365 ) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg import pywt from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() #        log-Y-axis :-) scales = scg.periods2scales(np.logspace(np.log10(2), np.log10(365*3), 200)) #        log-Y-axis :-) cwt = scg.CWT(births_normed, scales=scales) # :  1975    1976  fig = figure(figsize=(12,4)) lines = plot(births.index, births.values/1000, '.-') xlim(pd.to_datetime("1974-12-01"), pd.to_datetime("1975-02-01")) ylabel("Nb of birthes [k]"); title("Total births per day : 2 monthes zoom"); #      ax = scg.cws(cwt, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day", clim=(0,2500), yscale='log') set_x_yearly(ax, len(births)) xlim = ax.set_xlim( (1975-1969)*365-30, (1975-1969)*365+30 ) show() 

 import pandas as pd import numpy as np from pylab import* import scaleogram as scg import pywt from pandas.plotting import register_matplotlib_converters register_matplotlib_converters() #          #    'cmor1-1.5' #scg.set_default_wavelet('cmor1-1.5') #scg.set_default_wavelet('cgau5') #scg.set_default_wavelet('cgau1') #scg.set_default_wavelet('shan0.5-2') #scg.set_default_wavelet('mexh') df = pd.read_csv("births.csv") #   df = df[(df.day>=1) & (df.day<=31) & (df.births.values > 1000)] #   datetime = pd.to_datetime(df[['year', 'month', 'day']], errors='coerce') df.insert(0, 'datetime', datetime) datetime_lim = [ df.datetime.min(), df.datetime.max() ] years_lim = [ df.datetime.min().year, df.datetime.max().year ] births = df[['datetime', 'births']].groupby('datetime').sum().squeeze() def set_x_yearly(ax, days, start_year=1969): xlim = (np.round([0, days]) / 365).astype(np.int32) ticks = np.arange(xlim[0], xlim[1]) ax.set_xticks(ticks*365) ax.set_xticklabels(start_year + ticks) births_normed = births-births.mean() #        log-Y-axis :-) scales = scg.periods2scales(np.logspace(np.log10(2), np.log10(365*3), 200)) #        log-Y-axis :-) cwt = scg.CWT(births_normed, scales=scales) ax = scg.cws(cwt, figsize=(13.2, 4), xlabel="Year", ylabel="Nb of Day", clim=(0,2500), yscale='log') set_x_yearly(ax, len(births)) bbox = dict(boxstyle="round", facecolor="w", edgecolor="0.5") bbox2 = dict(boxstyle="round", facecolor="y", edgecolor="0.5") arrowprops = dict(facecolor='white', shrink=0.05) arrowprops2 = dict(facecolor='yellow', shrink=0.05) for period, label in [ (3.5, "half a week"), (7, "one week"), (365/2, "half a year"), (365, "one year"), (365*2.8, "long trends")]: text = ax.annotate(label, xy=(365*5.5, period), xytext=(365*2, period), bbox=bbox, arrowprops=arrowprops) text = ax.annotate("end of year\neffect", xy=(365*6, 15), xytext=(365*3, 50), bbox=bbox2, arrowprops=arrowprops2) text = ax.annotate("increase in\nvariations *amplitude*", xy=(365*18.5, 7), xytext=(365*14, 50), bbox=bbox2, arrowprops=arrowprops2) show()