def newton_raphson(self, ox, oy, direction, step=1e-5): """ Computes an iterate of NR Parameters ---------- ox : float DESCRIPTION. oy : float DESCRIPTION. direction : +/- 1 DESCRIPTION. 1 for positive branch of unstable manifold. -1 for negative step : TYPE, optional DESCRIPTION. The default is 1e-5. Returns ------- PE/PE' : float DESCRIPTION. Error iterate """ bmk = Bmk({'ox': ox, 'oy': oy}) ode = Ode(bmk.ode) PE = ode.homoclinic_dist(direction, 'rotational') bmk_incr = Bmk({'ox': ox, 'oy': oy + step}) ode_incr = Ode(bmk_incr.ode) PE_incr = ode_incr.homoclinic_dist(direction, 'rotational') return PE / ((PE_incr - PE) / step)
def newton_raphson(ox, oy, direction, step=1e-8): bmk = Bmk({'ox': ox, 'oy': oy}) ode = Ode(bmk.ode) PE = ode.homoclinic_dist(direction, 'contractible') bmk_incr = Bmk({'ox': ox + step, 'oy': oy}) ode_incr = Ode(bmk_incr.ode) PE_incr = ode_incr.homoclinic_dist(direction, 'contractible') return PE / ((PE_incr - PE) / step)
def jacobian(PE, ox, oy, step=1e-15): bmk = Bmk({'ox': ox + step, 'oy': oy}) ode = Ode(bmk.ode) PE_x = ode.pontryagin_energy('rotational') bmk = Bmk({'ox': ox, 'oy': oy + step}) ode = Ode(bmk.ode) PE_y = ode.pontryagin_energy('rotational') return np.array([[(PE_x[0] - PE[0]) / step, (PE_y[0] - PE[0]) / step], [(PE_x[1] - PE[1]) / step, (PE_y[1] - PE[1]) / step]])
def periodic_counter(ox, oy, granularity=200): bmk = Bmk({'ox': ox, 'oy': oy}) ode = Ode(bmk.ode) centre = ode.get_centre() saddle = ode.get_saddle_point() if centre[0]['x'] > saddle[0]['x']: theta = np.arctan((centre[0]['y'] - saddle[0]['y']) / (centre[0]['x'] - saddle[0]['x'])) else: theta = -np.pi + np.arctan((centre[0]['y'] - saddle[0]['y']) / (centre[0]['x'] - saddle[0]['x'])) c = np.cos(theta) s = np.sin(theta) granularity = 80 H_axis_x = np.linspace(centre[0]['x'], saddle[0]['x'], granularity) H_axis_y = np.linspace(centre[0]['y'], saddle[0]['y'], granularity) H_axis = list(zip(H_axis_x, H_axis_y)) delta_H = [] for (x_0, y_0) in H_axis: pt_dict = {'x': x_0, 'y': y_0} (x_0, y_0) = transform(x_0, y_0, c, s, centre) curve_x = [x_0] curve_y = [y_0] not_crossed = True looper = 0 while not_crossed and looper < 50000: looper += 1 pt_dict = ode._euler_iter(pt_dict, h=5e-3) pt_x, pt_y = transform(pt_dict['x'], pt_dict['y'], c, s, centre) if curve_y[-1] < -1e-12 and pt_y > 1e-12: not_crossed = False curve_x.append(pt_x) curve_y.append(pt_y) if np.abs(curve_y[-1]) < 1e-1 and curve_x[-1] < 1e-2: delta_H.append(x_0 - pt_x) else: #print('Escaped bounded region') break # plt.plot(curve_x, curve_y) #plt.show() po_count = 0 index = [] for H in range(10, len(delta_H) - 1): if np.sign(delta_H[H]) != np.sign(delta_H[H - 1]): po_count += 1 index.append(H) if po_count == 2: return po_count, delta_H else: return po_count, delta_H
@author: Francesco """ import PyDSTool #from numpy import (sin, cos, pi) #from PyDSTool.Toolbox import phaseplane as pp #import numpy as np import matplotlib.pyplot as plt from rhc_cont import Ode import pickle from bmk import Bmk #from ode import Ode from rhc_cont import RhcCont bmk = Bmk({'ox': 0, 'oy': 0}) ode = bmk.get_ode() #Setting Continuation class PC = PyDSTool.ContClass(ode) ########################################################## # # # ########################################################## def inner_outer_init(PC): print("Initialising boundary") PCargs = PyDSTool.args(name='EQ1', type='EP-C') PCargs.freepars = ['ox']
PE_x = ode.pontryagin_energy('rotational') bmk = Bmk({'ox': ox, 'oy': oy + step}) ode = Ode(bmk.ode) PE_y = ode.pontryagin_energy('rotational') return np.array([[(PE_x[0] - PE[0]) / step, (PE_y[0] - PE[0]) / step], [(PE_x[1] - PE[1]) / step, (PE_y[1] - PE[1]) / step]]) par = [-0.00964967, 0.92704312] progress = [tuple(par)] PE_progress = [] for i in range(100): bmk = Bmk({'ox': par[0], 'oy': par[1]}) ode = Ode(bmk.ode) PE = ode.pontryagin_energy('rotational') jac = jacobian(PE, par[0], par[1]) par -= 0.1 * np.matmul(np.linalg.inv(jac), np.array(PE)) progress.append(tuple(par)) PE_progress.append(sum(PE)) print("Epoch:", i, " PE:", round(PE_progress[-1], 8)) #pickle.dump(progress, open('progress_08.p','wb')) #plt.scatter(*zip(*progress)) plt.plot(PE_progress) plt.show()
# -*- coding: utf-8 -*- """ Created on Fri May 29 18:43:12 2020 @author: Francesco """ import matplotlib.pyplot as plt from ode import Ode from bmk import Bmk import numpy as np bmk = Bmk({'ox': -0.009, 'oy': 0.927}) ode = Ode(bmk.ode) fp = ode.get_fixed_points() if ode.fp_eigen(fp[0]) in ['hyperbolic repeller', 'hyperbolic attractor']: centre = fp[0] saddle = fp[1] else: centre = fp[1] saddle = fp[0] H_axis_x = np.linspace(centre['x'],saddle['x']+1, 20) H_axis_y = np.linspace(centre['y'],centre['y'], 20) H_axis = list(zip(H_axis_x, H_axis_y))
for index, val_pair in enumerate(zip(xs, ys)): if index != 0: if np.abs(xs[index] - xs[index - 1]) > break_threshold: xs_.append(None) ys_.append(None) if np.abs(ys[index] - ys[index - 1]) > break_threshold: xs_.append(None) ys_.append(None) xs_.append(val_pair[0]) ys_.append(val_pair[1]) return np.array(xs_), np.array(ys_) bmk = Bmk({'ox': -0.16, 'oy': 1}) ode = Ode(bmk.ode) P1, _ = ode.unstable_manifold(1) P2, _ = ode.unstable_manifold(-1) x_plt = [] y_plt = [] for i in range(len(P1)): (x, y) = ode._point_dict_getter(P1[i]) x_plt.append(x) y_plt.append(y) x_plt, y_plt = para_plot(np.array(x_plt) % 1, np.array(y_plt) % 1, 0.5) x_plt_lift = []
Created on Sat Apr 25 21:57:51 2020 @author: Francesco """ import matplotlib.pyplot as plt from ode import Ode from bmk import Bmk import numpy as np ox = np.linspace(-1e-02, 0, 100) tr_saddle = [] tr_node = [] #-0.009649961793522382], [0.9270432434144844 for i in ox: bmk = Bmk({'ox': i, 'oy': 0.9270432434144844}) ode = Ode(bmk.ode) fp = ode.get_fixed_points() fp_type = ode.fp_eigen(fp[0])[0] if fp_type == 'saddle point': print('case saddle {}'.format(i)) tr_saddle.append(ode.jacobian(fp[0])[0][0] + ode.jacobian(fp[0])[1][1]) tr_node.append(ode.jacobian(fp[1])[0][0] + ode.jacobian(fp[1])[1][1]) else: print('case node {}'.format(i)) tr_saddle.append(ode.jacobian(fp[1])[0][0] + ode.jacobian(fp[1])[1][1]) tr_node.append(ode.jacobian(fp[0])[0][0] + ode.jacobian(fp[0])[1][1]) #figure, axes = plt.subplots(nrows=2, ncols=1) plt.plot(ox, tr_saddle) #plt.plot(ox, [0]*len(ox))