为什么我不能让这个 Runge-Kutta 求解器随着时间步长的减少而收敛？
Why can't I get this Runge-Kutta solver to converge as the time step decreases?
由于某些原因，我需要在 PyTorch 中实现 Runge-Kutta4 方法（所以不，我不会使用 scipy.odeint）。我尝试并在最简单的测试用例上得到了奇怪的结果，用 x(0)=1 求解 x'=x（解析解：x=exp(t)）。基本上，当我减少时间步长时，我无法降低数值误差。我可以使用更简单的 Euler 方法来完成，但不能使用 Runge-Kutta 4 方法，这让我怀疑这里存在一些浮点问题（也许我遗漏了一些从双精度到单精度的隐藏转换）？
import torch
import numpy as np
import matplotlib.pyplot as plt
def Euler(f, IC, time_grid):
y0 = torch.tensor([IC])
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dy = f(t_i, y_i) * dt
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
def RungeKutta4(f, IC, time_grid):
y0 = torch.tensor([IC])
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dtd2 = 0.5 * dt
f1 = f(t_i, y_i)
f2 = f(t_i + dtd2, y_i + dtd2 * f1)
f3 = f(t_i + dtd2, y_i + dtd2 * f2)
f4 = f(t_next, y_i + dt * f3)
dy = 1/6 * dt * (f1 + 2 * (f2 + f3) +f4)
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
# differential equation
def f(T, X):
return X
# initial condition
IC = 1.
# integration interval
def integration_interval(steps, ND=1):
return torch.linspace(0, ND, steps)
# analytical solution
def analytical_solution(t_range):
return np.exp(t_range)
# test a numerical method
def test_method(method, t_range, analytical_solution):
numerical_solution = method(f, IC, t_range)
L_inf_err = torch.dist(numerical_solution, analytical_solution, float('inf'))
return L_inf_err
if __name__ == '__main__':
Euler_error = np.array([0.,0.,0.])
RungeKutta4_error = np.array([0.,0.,0.])
indices = np.arange(1, Euler_error.shape[0]+1)
n_steps = np.power(10, indices)
for i, n in np.ndenumerate(n_steps):
t_range = integration_interval(steps=n)
solution = analytical_solution(t_range)
Euler_error[i] = test_method(Euler, t_range, solution).numpy()
RungeKutta4_error[i] = test_method(RungeKutta4, t_range, solution).numpy()
plots_path = "./plots"
a = plt.figure()
plt.xscale('log')
plt.yscale('log')
plt.plot(n_steps, Euler_error, label="Euler error", linestyle='-')
plt.plot(n_steps, RungeKutta4_error, label="RungeKutta 4 error", linestyle='-.')
plt.legend()
plt.savefig(plots_path + "/errors.png")
结果：
如您所见，Euler 方法收敛（缓慢，正如一阶方法所预期的那样）。然而，Runge-Kutta4 方法 不会 随着时间步长越来越小而收敛。错误最初下降，然后再次上升。这里有什么问题？
确实是浮点精度问题。 torch 默认为单精度，因此一旦 truncation error becomes small enough, the total error is basically determined by the roundoff error，并通过增加步数 <=> 减少时间步进一步减少截断误差不会导致总误差减少。
要解决此问题，我们需要为所有浮点 torch 张量和 numpy 数组强制执行双精度 64 位浮点数。请注意，正确的做法是分别使用 torch.float64 和 np.float64 而不是 torch.double 和 np.double，因为 the former are fixed-sized float values, (always 64bit) while the latter depend on the machine and/or compiler。这是固定代码：
import torch
import numpy as np
import matplotlib.pyplot as plt
def Euler(f, IC, time_grid):
y0 = torch.tensor([IC], dtype=torch.float64)
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dy = f(t_i, y_i) * dt
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
def RungeKutta4(f, IC, time_grid):
y0 = torch.tensor([IC], dtype=torch.float64)
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dtd2 = 0.5 * dt
f1 = f(t_i, y_i)
f2 = f(t_i + dtd2, y_i + dtd2 * f1)
f3 = f(t_i + dtd2, y_i + dtd2 * f2)
f4 = f(t_next, y_i + dt * f3)
dy = 1/6 * dt * (f1 + 2 * (f2 + f3) +f4)
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
# differential equation
def f(T, X):
return X
# initial condition
IC = 1.
# integration interval
def integration_interval(steps, ND=1):
return torch.linspace(0, ND, steps, dtype=torch.float64)
# analytical solution
def analytical_solution(t_range):
return np.exp(t_range, dtype=np.float64)
# test a numerical method
def test_method(method, t_range, analytical_solution):
numerical_solution = method(f, IC, t_range)
L_inf_err = torch.dist(numerical_solution, analytical_solution, float('inf'))
return L_inf_err
if __name__ == '__main__':
Euler_error = np.array([0.,0.,0.], dtype=np.float64)
RungeKutta4_error = np.array([0.,0.,0.], dtype=np.float64)
indices = np.arange(1, Euler_error.shape[0]+1)
n_steps = np.power(10, indices)
for i, n in np.ndenumerate(n_steps):
t_range = integration_interval(steps=n)
solution = analytical_solution(t_range)
Euler_error[i] = test_method(Euler, t_range, solution).numpy()
RungeKutta4_error[i] = test_method(RungeKutta4, t_range, solution).numpy()
plots_path = "./plots"
a = plt.figure()
plt.xscale('log')
plt.yscale('log')
plt.plot(n_steps, Euler_error, label="Euler error", linestyle='-')
plt.plot(n_steps, RungeKutta4_error, label="RungeKutta 4 error", linestyle='-.')
plt.legend()
plt.savefig(plots_path + "/errors.png")
结果：
现在，随着我们减小时间步长，RungeKutta4 近似的误差会随着正确率而减小。
由于某些原因，我需要在 PyTorch 中实现 Runge-Kutta4 方法（所以不，我不会使用 scipy.odeint）。我尝试并在最简单的测试用例上得到了奇怪的结果，用 x(0)=1 求解 x'=x（解析解：x=exp(t)）。基本上，当我减少时间步长时，我无法降低数值误差。我可以使用更简单的 Euler 方法来完成，但不能使用 Runge-Kutta 4 方法，这让我怀疑这里存在一些浮点问题（也许我遗漏了一些从双精度到单精度的隐藏转换）？
import torch
import numpy as np
import matplotlib.pyplot as plt
def Euler(f, IC, time_grid):
y0 = torch.tensor([IC])
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dy = f(t_i, y_i) * dt
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
def RungeKutta4(f, IC, time_grid):
y0 = torch.tensor([IC])
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dtd2 = 0.5 * dt
f1 = f(t_i, y_i)
f2 = f(t_i + dtd2, y_i + dtd2 * f1)
f3 = f(t_i + dtd2, y_i + dtd2 * f2)
f4 = f(t_next, y_i + dt * f3)
dy = 1/6 * dt * (f1 + 2 * (f2 + f3) +f4)
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
# differential equation
def f(T, X):
return X
# initial condition
IC = 1.
# integration interval
def integration_interval(steps, ND=1):
return torch.linspace(0, ND, steps)
# analytical solution
def analytical_solution(t_range):
return np.exp(t_range)
# test a numerical method
def test_method(method, t_range, analytical_solution):
numerical_solution = method(f, IC, t_range)
L_inf_err = torch.dist(numerical_solution, analytical_solution, float('inf'))
return L_inf_err
if __name__ == '__main__':
Euler_error = np.array([0.,0.,0.])
RungeKutta4_error = np.array([0.,0.,0.])
indices = np.arange(1, Euler_error.shape[0]+1)
n_steps = np.power(10, indices)
for i, n in np.ndenumerate(n_steps):
t_range = integration_interval(steps=n)
solution = analytical_solution(t_range)
Euler_error[i] = test_method(Euler, t_range, solution).numpy()
RungeKutta4_error[i] = test_method(RungeKutta4, t_range, solution).numpy()
plots_path = "./plots"
a = plt.figure()
plt.xscale('log')
plt.yscale('log')
plt.plot(n_steps, Euler_error, label="Euler error", linestyle='-')
plt.plot(n_steps, RungeKutta4_error, label="RungeKutta 4 error", linestyle='-.')
plt.legend()
plt.savefig(plots_path + "/errors.png")
结果：
如您所见，Euler 方法收敛（缓慢，正如一阶方法所预期的那样）。然而，Runge-Kutta4 方法 不会 随着时间步长越来越小而收敛。错误最初下降，然后再次上升。这里有什么问题？
确实是浮点精度问题。 torch 默认为单精度，因此一旦 truncation error becomes small enough, the total error is basically determined by the roundoff error，并通过增加步数 <=> 减少时间步进一步减少截断误差不会导致总误差减少。
要解决此问题，我们需要为所有浮点 torch 张量和 numpy 数组强制执行双精度 64 位浮点数。请注意，正确的做法是分别使用 torch.float64 和 np.float64 而不是 torch.double 和 np.double，因为 the former are fixed-sized float values, (always 64bit) while the latter depend on the machine and/or compiler。这是固定代码：
import torch
import numpy as np
import matplotlib.pyplot as plt
def Euler(f, IC, time_grid):
y0 = torch.tensor([IC], dtype=torch.float64)
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dy = f(t_i, y_i) * dt
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
def RungeKutta4(f, IC, time_grid):
y0 = torch.tensor([IC], dtype=torch.float64)
time_grid = time_grid.to(y0[0])
values = y0
for i in range(0, time_grid.shape[0] - 1):
t_i = time_grid[i]
t_next = time_grid[i+1]
y_i = values[i]
dt = t_next - t_i
dtd2 = 0.5 * dt
f1 = f(t_i, y_i)
f2 = f(t_i + dtd2, y_i + dtd2 * f1)
f3 = f(t_i + dtd2, y_i + dtd2 * f2)
f4 = f(t_next, y_i + dt * f3)
dy = 1/6 * dt * (f1 + 2 * (f2 + f3) +f4)
y_next = y_i + dy
y_next = y_next.unsqueeze(0)
values = torch.cat((values, y_next), dim=0)
return values
# differential equation
def f(T, X):
return X
# initial condition
IC = 1.
# integration interval
def integration_interval(steps, ND=1):
return torch.linspace(0, ND, steps, dtype=torch.float64)
# analytical solution
def analytical_solution(t_range):
return np.exp(t_range, dtype=np.float64)
# test a numerical method
def test_method(method, t_range, analytical_solution):
numerical_solution = method(f, IC, t_range)
L_inf_err = torch.dist(numerical_solution, analytical_solution, float('inf'))
return L_inf_err
if __name__ == '__main__':
Euler_error = np.array([0.,0.,0.], dtype=np.float64)
RungeKutta4_error = np.array([0.,0.,0.], dtype=np.float64)
indices = np.arange(1, Euler_error.shape[0]+1)
n_steps = np.power(10, indices)
for i, n in np.ndenumerate(n_steps):
t_range = integration_interval(steps=n)
solution = analytical_solution(t_range)
Euler_error[i] = test_method(Euler, t_range, solution).numpy()
RungeKutta4_error[i] = test_method(RungeKutta4, t_range, solution).numpy()
plots_path = "./plots"
a = plt.figure()
plt.xscale('log')
plt.yscale('log')
plt.plot(n_steps, Euler_error, label="Euler error", linestyle='-')
plt.plot(n_steps, RungeKutta4_error, label="RungeKutta 4 error", linestyle='-.')
plt.legend()
plt.savefig(plots_path + "/errors.png")
结果：
现在，随着我们减小时间步长，RungeKutta4 近似的误差会随着正确率而减小。