Question is like this:
f(x) = A sin(2π * L * x) + B cos(2π * M * x) + C sin(2π * N * x)
and L,M,N are constants integer, 0 <= L,M,N <= 100
and A,B,C can be any possible integers.
Here is the given data:
x = [0,0.01,0.02,0.03,0.04,0.05,0.06,0.07,0.08,0.09,0.1,0.11,0.12,0.13,0.14,0.15,0.16,0.17,0.18,0.19,0.2,0.21,0.22,0.23,0.24,0.25,0.26,0.27,0.28,0.29,0.3,0.31,0.32,0.33,0.34,0.35,0.36,0.37,0.38,0.39,0.4,0.41,0.42,0.43,0.44,0.45,0.46,0.47,0.48,0.49,0.5,0.51,0.52,0.53,0.54,0.55,0.56,0.57,0.58,0.59,0.6,0.61,0.62,0.63,0.64,0.65,0.66,0.67,0.68,0.69,0.7,0.71,0.72,0.73,0.74,0.75,0.76,0.77,0.78,0.79,0.8,0.81,0.82,0.83,0.84,0.85,0.86,0.87,0.88,0.89,0.9,0.91,0.92,0.93,0.94,0.95,0.96,0.97,0.98,0.99]
y = [4,1.240062433,-0.7829654986,-1.332487982,-0.3337640721,1.618033989,3.512512389,4.341307895,3.515268061,1.118929599,-2.097886967,-4.990538967,-6.450324073,-5.831575611,-3.211486891,0.6180339887,4.425660706,6.980842552,7.493970785,5.891593744,2.824429495,-0.5926374511,-3.207870455,-4.263694544,-3.667432785,-2,-0.2617162175,0.5445886005,-0.169441247,-2.323237059,-5.175570505,-7.59471091,-8.488730333,-7.23200463,-3.924327772,0.6180339887,5.138501587,8.38127157,9.532377045,8.495765687,5.902113033,2.849529206,0.4768388529,-0.46697525,0.106795821,1.618033989,3.071952496,3.475795162,2.255463709,-0.4905371745,-4,-7.117914956,-8.727599664,-8.178077181,-5.544088451,-1.618033989,2.365340134,5.169257268,5.995297102,4.758922924,2.097886967,-0.8873135564,-3.06024109,-3.678989552,-2.666365632,-0.6180339887,1.452191817,2.529722611,2.016594378,-0.01374122059,-2.824429495,-5.285215072,-6.302694708,-5.246870619,-2.210419738,2,6.13956874,8.965976562,9.68000641,8.201089581,5.175570505,1.716858387,-1.02183483,-2.278560533,-1.953524751,-0.6180339887,0.7393509358,1.129293593,-0.02181188158,-2.617913164,-5.902113033,-8.727381729,-9.987404016,-9.043589913,-5.984648344,-1.618033989,2.805900027,6.034770001,7.255101454,6.368389697]
enter image description here
How to use Gradient Descent to solve this multiple terms trigonometry function?
Gradient descent is not well suited for optimisation over integers. You can try a navie relaxation where you solve in floats, and hope the rounded solution is still ok.
from autograd import grad, numpy as jnp
import numpy as np
def cast(params):
[A, B, C, L, M, N] = params
L = jnp.minimum(jnp.abs(L), 100)
M = jnp.minimum(jnp.abs(M), 100)
N = jnp.minimum(jnp.abs(N), 100)
return A, B, C, L, M, N
def pred(params, x):
[A, B, C, L, M, N] = cast(params)
return A *jnp.sin(2 * jnp.pi * L * x) + B*jnp.cos(2*jnp.pi * M * x) + C * jnp.sin(2 * jnp.pi * N * x)
x = [0,0.01,0.02,0.03,0.04,0.05,0.06,0.07,0.08,0.09,0.1,0.11,0.12,0.13,0.14,0.15,0.16,0.17,0.18,0.19,0.2,0.21,0.22,0.23,0.24,0.25,0.26,0.27,0.28,0.29,0.3,0.31,0.32,0.33,0.34,0.35,0.36,0.37,0.38,0.39,0.4,0.41,0.42,0.43,0.44,0.45,0.46,0.47,0.48,0.49,0.5,0.51,0.52,0.53,0.54,0.55,0.56,0.57,0.58,0.59,0.6,0.61,0.62,0.63,0.64,0.65,0.66,0.67,0.68,0.69,0.7,0.71,0.72,0.73,0.74,0.75,0.76,0.77,0.78,0.79,0.8,0.81,0.82,0.83,0.84,0.85,0.86,0.87,0.88,0.89,0.9,0.91,0.92,0.93,0.94,0.95,0.96,0.97,0.98,0.99]
y = [4,1.240062433,-0.7829654986,-1.332487982,-0.3337640721,1.618033989,3.512512389,4.341307895,3.515268061,1.118929599,-2.097886967,-4.990538967,-6.450324073,-5.831575611,-3.211486891,0.6180339887,4.425660706,6.980842552,7.493970785,5.891593744,2.824429495,-0.5926374511,-3.207870455,-4.263694544,-3.667432785,-2,-0.2617162175,0.5445886005,-0.169441247,-2.323237059,-5.175570505,-7.59471091,-8.488730333,-7.23200463,-3.924327772,0.6180339887,5.138501587,8.38127157,9.532377045,8.495765687,5.902113033,2.849529206,0.4768388529,-0.46697525,0.106795821,1.618033989,3.071952496,3.475795162,2.255463709,-0.4905371745,-4,-7.117914956,-8.727599664,-8.178077181,-5.544088451,-1.618033989,2.365340134,5.169257268,5.995297102,4.758922924,2.097886967,-0.8873135564,-3.06024109,-3.678989552,-2.666365632,-0.6180339887,1.452191817,2.529722611,2.016594378,-0.01374122059,-2.824429495,-5.285215072,-6.302694708,-5.246870619,-2.210419738,2,6.13956874,8.965976562,9.68000641,8.201089581,5.175570505,1.716858387,-1.02183483,-2.278560533,-1.953524751,-0.6180339887,0.7393509358,1.129293593,-0.02181188158,-2.617913164,-5.902113033,-8.727381729,-9.987404016,-9.043589913,-5.984648344,-1.618033989,2.805900027,6.034770001,7.255101454,6.368389697]
def loss(params):
p = pred(params, np.array(x))
return jnp.mean((np.array(y)-p)**2)
params = np.array([np.random.random()*100 for _ in range(6)])
for _ in range(10000):
g = grad(loss)
params = params - 0.001*g(params)
print("Relaxed solution", cast(params), "loss=", loss(params))
constrained_params = np.round(cast(params))
print("Integer solution", constrained_params, "loss=", loss(constrained_params))
print()
Since the problem will have a lot of local minima, you might need to run it multiple times.
It's quite hard to use gradient descent to find a solution to this problem, because it tends to get stuck when changing the L, M, or N parameters. The gradients for those can push it away from the right solution, unless it is very close to an optimal solution already.
There are ways to get around this, such as basinhopping or random search, but because of the function you're trying to learn, you have a better alternative.
Since you're trying to learn a sinusoid function, you can use an FFT to find the frequencies of the sine waves. Once you have those frequencies, you can find the amplitudes and phases used to generate the same sine wave.
Pardon the messiness of this code, this is my first time using an FFT.
import scipy.fft
import numpy as np
import math
import matplotlib.pyplot as plt
def get_top_frequencies(x, y, num_freqs):
x = np.array(x)
y = np.array(y)
# Find timestep (assume constant timestep)
dt = abs(x[0] - x[-1]) / (len(x) - 1)
# Take discrete FFT of y
spectral = scipy.fft.fft(y)
freq = scipy.fft.fftfreq(y.shape[0], d=dt)
# Cut off top half of frequencies. Assumes input signal is real, and not complex.
spectral = spectral[:int(spectral.shape[0] / 2)]
# Double amplitudes to correct for cutting off top half.
spectral *= 2
# Adjust amplitude by sampling timestep
spectral *= dt
# Get ampitudes for all frequencies. This is taking the magnitude of the complex number
spectral_amplitude = np.abs(spectral)
# Pick frequencies with highest amplitudes
highest_idx = np.argsort(spectral_amplitude)[::-1][:num_freqs]
# Find amplitude, frequency, and phase components of each term
highest_amplitude = spectral_amplitude[highest_idx]
highest_freq = freq[highest_idx]
highest_phase = np.angle(spectral[highest_idx]) / math.pi
# Convert it into a Python function
function = ["def func(x):", "return ("]
for i, components in enumerate(zip(highest_amplitude, highest_freq, highest_phase)):
amplitude, freq, phase = components
plus_sign = " +" if i != (num_freqs - 1) else ""
term = f"{amplitude:.2f} * math.cos(2 * math.pi * {freq:.2f} * x + math.pi * {phase:.2f}){plus_sign}"
function.append(" " + term)
function.append(")")
return "\n ".join(function)
x = [0,0.01,0.02,0.03,0.04,0.05,0.06,0.07,0.08,0.09,0.1,0.11,0.12,0.13,0.14,0.15,0.16,0.17,0.18,0.19,0.2,0.21,0.22,0.23,0.24,0.25,0.26,0.27,0.28,0.29,0.3,0.31,0.32,0.33,0.34,0.35,0.36,0.37,0.38,0.39,0.4,0.41,0.42,0.43,0.44,0.45,0.46,0.47,0.48,0.49,0.5,0.51,0.52,0.53,0.54,0.55,0.56,0.57,0.58,0.59,0.6,0.61,0.62,0.63,0.64,0.65,0.66,0.67,0.68,0.69,0.7,0.71,0.72,0.73,0.74,0.75,0.76,0.77,0.78,0.79,0.8,0.81,0.82,0.83,0.84,0.85,0.86,0.87,0.88,0.89,0.9,0.91,0.92,0.93,0.94,0.95,0.96,0.97,0.98,0.99]
y = [4,1.240062433,-0.7829654986,-1.332487982,-0.3337640721,1.618033989,3.512512389,4.341307895,3.515268061,1.118929599,-2.097886967,-4.990538967,-6.450324073,-5.831575611,-3.211486891,0.6180339887,4.425660706,6.980842552,7.493970785,5.891593744,2.824429495,-0.5926374511,-3.207870455,-4.263694544,-3.667432785,-2,-0.2617162175,0.5445886005,-0.169441247,-2.323237059,-5.175570505,-7.59471091,-8.488730333,-7.23200463,-3.924327772,0.6180339887,5.138501587,8.38127157,9.532377045,8.495765687,5.902113033,2.849529206,0.4768388529,-0.46697525,0.106795821,1.618033989,3.071952496,3.475795162,2.255463709,-0.4905371745,-4,-7.117914956,-8.727599664,-8.178077181,-5.544088451,-1.618033989,2.365340134,5.169257268,5.995297102,4.758922924,2.097886967,-0.8873135564,-3.06024109,-3.678989552,-2.666365632,-0.6180339887,1.452191817,2.529722611,2.016594378,-0.01374122059,-2.824429495,-5.285215072,-6.302694708,-5.246870619,-2.210419738,2,6.13956874,8.965976562,9.68000641,8.201089581,5.175570505,1.716858387,-1.02183483,-2.278560533,-1.953524751,-0.6180339887,0.7393509358,1.129293593,-0.02181188158,-2.617913164,-5.902113033,-8.727381729,-9.987404016,-9.043589913,-5.984648344,-1.618033989,2.805900027,6.034770001,7.255101454,6.368389697]
print(get_top_frequencies(x, y, 3))
That produces this function:
def func(x):
return (
5.00 * math.cos(2 * math.pi * 10.00 * x + math.pi * 0.50) +
4.00 * math.cos(2 * math.pi * 5.00 * x + math.pi * -0.00) +
2.00 * math.cos(2 * math.pi * 3.00 * x + math.pi * -0.50)
)
Which is not quite the format you specified - you asked for two sins and one cos, and for no phase parameter. However, using the trigonometric identity cos(x) = sin(pi/2 - x), you can convert this into an equivalent expression that matches what you want:
def func(x):
return (
5.00 * math.sin(2 * math.pi * -10.00 * x) +
4.00 * math.cos(2 * math.pi * 5.00 * x) +
2.00 * math.sin(2 * math.pi * 3.00 * x)
)
And there's the original function!
I would like to draw cycloid that is going on other cycloid but I don't know exactly how to do this. Here is my code.
import numpy as np
import matplotlib.pyplot as plt
import math
from matplotlib import animation
#r = float(input('write r\n'))
#R = float(input('write R\n'))
r = 1
R = 1
x = []
y = []
x2 = []
y2 = []
x3 = []
y3 = []
length=[0]
fig, ax = plt.subplots()
ln, = plt.plot([], [], 'r', animated=True)
f = np.linspace(0, 10*r*math.pi, 1000)
def init():
ax.set_xlim(-r, 12*r*math.pi)
ax.set_ylim(-4*r, 4*r)
return ln,
def update2(frame):
#parametric equations of cycloid
x0 = r * (frame - math.sin(frame))
y0 = r * (1 - math.cos(frame))
x.append(x0)
y.append(y0)
#derivative of cycloid
dx = r * (1 - math.cos(frame))
dy = r * math.sin(frame)
#center of circle
a = dy * dy + dx * dx
b = (-2 * x0 * dy) - (2 * frame * dy * dy) + (2 * y0 * dx) - (2 * frame * dx * dx)
c = (x0 * x0) + (2 * frame * x0 * dy) + (frame * frame * dy * dy) + (y0 * y0) - (2 * frame * y0 * dx) + (frame * frame * dx * dx) -1
t1 = (-b - math.sqrt(b * b - 4 * a * c)) / (2 * a)
#t2 = (-b + math.sqrt(b * b - 4 * a * c)) / (2 * a)
center1x=(x0-dy*(t1-x0))*R
center1y=(y0+dx*(t1-x0))*R
#center2x=(x0-dy*(t2-x0))*R
#center2y=(y0+dx*(t2-x0))*R
#length of cycloid
length.append(math.sqrt(x0*x0 + y0*y0))
dl=sum(length)
param = dl / R
W1x = center1x + R * math.cos(-param)
W1y = center1y + R * math.sin(-param)
#W2x = center2x + R * math.cos(-param)
#W2y = center2y + R * math.sin(-param)
x2.append(W1x)
y2.append(W1y)
#x3.append(W2x)
#y3.append(W2y)
ln.set_data([x, x2], [y, y2])
return ln,
ani = animation.FuncAnimation(fig, update2, frames=f,init_func=init, blit=True, interval = 0.1, repeat = False)
plt.show()
In my function update2 I created parametric equations of first cycloid and then tried to obtain co-ordinates of points of second cycloid that should go on the first one.
My idea is based on that that typical cycloid is moving on straight line, and cycloid that is moving on other curve must moving on tangent of that curve, so center of circle that's creating this cycloid is always placed on normal of curve. From parametric equations of normal I have tried to obtain center of circle that creating cycloid but I think that isn't good way.
My goal is to get something like this:
Here is one way. Calculus gives us the formulas to find the direction angles at any point on the cycloid and the arc lengths along the cycloid. Analytic Geometry tells us how to use that information to find your desired points.
By the way, a plot made by rolling a figure along another figure is called a roulette. My code is fairly simple and could be optimized, but it works now, can be used for other problems, and is broken up to make the math and algorithm easier to understand. To understand my code, use this diagram. The cycloid is the blue curve, the black circles are the rolling circle on the cycloid, point A is an "anchor point" (a point where the rim point touches the cycloid--I wanted to make this code general), and point F is the moving rim point. The two red arcs are the same length, which is what we mean by rolling the circle along the cycloid.
And here is my code. Ask if you need help with the source of the various formulas, but the direction angles and arc lengths use calculus.
"""Numpy-compatible routines for a standard cycloid (one caused by a
circle of radius r above the y-axis rolling along the positive x-axis
starting from the origin).
"""
import numpy as np
def x(t, r):
"""Return the x-coordinate of a point on the cycloid with parameter t."""
return r * (t - np.sin(t))
def y(t, r):
"""Return the y-coordinate of a point on the cycloid with parameter t."""
return r * (1.0 - np.cos(t))
def dir_angle_norm_in(t, r):
"""Return the direction angle of the vector normal to the cycloid at
the point with parameter t that points into the cycloid."""
return -t / 2.0
def dir_angle_norm_out(t, r):
"""Return the direction angle of the vector normal to the cycloid at
the point with parameter t that points out of the cycloid."""
return np.pi - t / 2.0
def arclen(t, r):
"""Return the arc length of the cycloid between the origin and the
point on the cycloid with parameter t."""
return 4.0 * r * (1.0 - np.cos(t / 2.0))
# Roulette problem
def xy_roulette(t, r, T, R):
"""Return the x-y coordinates of a rim point on a circle of radius
R rolling on a cycloid of radius r starting at the anchor point
with parameter T currently at the point with parameter t. (Such a
rolling curve on another curve is called a roulette.)
"""
# Find the coordinates of the contact point P between circle and cycloid
px, py = x(t, r), y(t, r)
# Find the direction angle of PC from the contact point to circle's center
a1 = dir_angle_norm_out(t, r)
# Find the coordinates of the center C of the circle
cx, cy = px + R * np.cos(a1), py + R * np.sin(a1)
# Find cycloid's arc distance AP between anchor and current contact points
d = arclen(t, r) - arclen(T, r) # equals arc PF
# Find the angle φ the circle turned while rolling from the anchor pt
phi = d / R
# Find the direction angle of CF from circle's center to rim point
a2 = dir_angle_norm_in(t, r) - phi # subtract: circle rolls clockwise
# Find the coordinates of the final point F
fx, fy = cx + R * np.cos(a2), cy + R * np.sin(a2)
# Return those coordinates
return fx, fy
import matplotlib.pyplot as plt
r = 1
R = 0.75
T = np.pi / 3
t_array = np.linspace(0, 2*np.pi, 201)
cycloid_x = x(t_array, r)
cycloid_y = y(t_array, r)
roulette_x, roulette_y = xy_roulette(t_array, r, T, R)
fig, ax = plt.subplots()
ax.set_aspect('equal')
ax.axhline(y=0, color='k')
ax.axvline(x=0, color='k')
ax.plot(cycloid_x, cycloid_y)
ax.plot(roulette_x, roulette_y)
plt.show()
And here is the resulting graphic. You can pretty this up as you choose. Note that this only has the circle rolling along one arch of the cycloid. If you clarify what should happen at the cusps, this could be extended.
Or, if you want a smaller circle and a curve that ends at the cusps (here r = 1, T = 0 n = 6 (the number of little arches), and R = 4 * r / np.pi / n),
You can generate coordinates of the center of rolling circle using parallel curve definition. Parametric equations for this center are rather simple (if I did not make mistakes):
for big cycloid:
X = R(t - sin(t))
Y = R(1 - cos(t))
X' = R(1 - cos(t))
Y' = R*sin(t)
parallel curve (center of small circle):
Sqrt(X'^2+Y'^2)=R*Sqrt(1-2*cos(t)+cos^2(t)+sin^2(t)) =
R*Sqrt(2-2*cos(t))=
R*Sqrt(4*sin^2(t/2))=
2*R*sin(t/2)
x(t) = X(t) + r*R*sin(t)/(2R*sin(t/2)) =
R(t - sin(t)) + r*2*sin(t/2)*cos(t/2) / (2*sin(t/2)) =
R(t - sin(t)) + r*cos(t/2)
y(t) = Y(t) - r*R*(1-cos(t))/(2*R*sin(t/2)) =
R(1 - cos(t)) - r*(2*sin^2(t/2)/(2*sin(t/2)) =
R(1 - cos(t)) - r*sin(t/2)
But trajectory of point on the circumference is superposition of center position and rotation around it with angular velocity that depends on length of main cycloid plus rotation of main tangent.
Added from dicussion in comments:
cycloid arc length
L(t) = 4R*(1-cos(t/2))
to use it for small circle rotation, divide by r
tangent rotation derivation
fi(t) = atan(Y'/X') = atan(sin(t)/(1-cos(t)) =
atan(2*sin(t/2)*cos(t/2)/(2(sin^2(t/2))) =
atan(ctg(t/2)) = Pi/2 - t/2
so tangent direction change is proportional to big cycloid parameter
and final result is (perhaps some signs are not correct)
theta(t) = L(t)/r + t/2 + Phase
ox(t) = x(t) + r * cos(theta(t))
oy(t) = y(t) + r * sin(theta(t))
Thanks everyone. Somehow I have managed to accomplish that. Solution is maybe ugly but sufficient for me.
import numpy as np
import matplotlib.pyplot as plt
import math
from matplotlib import animation
r = float(input('write r\n'))
R = float(input('write R\n'))
#r=1
#R=0.1
x = []
y = []
x2 = []
y2 = []
x_1=0
x_2=0
lengthX=[0]
lengthY=[0]
lengthabs=[0]
fig, ax = plt.subplots()
ln, = plt.plot([], [], 'r', animated=True)
f = np.linspace(0, 2*math.pi, 1000)
def init():
ax.set_xlim(-r, 4*r*math.pi)
ax.set_ylim(0, 4*r)
return ln,
def update2(frame):
#cycloid's equations
x0 = r * (frame - math.sin(frame))
y0 = r * (1 - math.cos(frame))
x.append(r * (frame - math.sin(frame)))
y.append(r * (1 - math.cos(frame)))
#arc's length
lengthabs.append(math.sqrt((x0-lengthX[-1])*(x0-lengthX[-1])+(y0-lengthY[-1])*(y0-lengthY[-1])))
lengthX.append(x0)
lengthY.append(y0)
dl=sum(lengthabs)
param = dl / R
#center of circle
center1x = r * (frame - math.sin(frame)) + R * math.cos((frame+2*math.pi) / 2)
center1y = r * (1 - math.cos(frame)) - R * math.sin((frame+2*math.pi) / 2)
if(frame<2*math.pi):
W1x = center1x + R * math.cos(-param)
W1y = center1y + R * math.sin(-param)
else:
W1x = center1x + R * math.cos(param)
W1y = center1y + R * math.sin(param)
x2.append(W1x)
y2.append(W1y)
ln.set_data([x,x2], [y,y2])
return ln,
ani = animation.FuncAnimation(fig, update2, frames=f,init_func=init, blit=True, interval = 0.1, repeat = False)
plt.show()