python - 如何在 axis=1 处追加一维数组

Question

我正在获取维度 (2,1) 的样本。我正在尝试将它们堆叠成列。

我尝试过以下方法:

# My initial state
state=np.array([2,3])
trajectory =state

# the following generate the next samples
class Buck:
  """ The following code simulates a Buck converter """
  def __init__(self,state,control):
    self.control=control
    self.state=state

  def Next_State(self):
    L, C = 1.0, 1.0
    R, G = 1.0, 1.0
    delta = 0.001


    Q = np.array([[-1.0/L,0.0],[0.0,1.0/C]])
    A = Q*np.matmul(Q,np.array([[R,1.0],[1.0,-G]]))

    next_state = state + delta*np.matmul(A,state)

    return next_state

# Here I am appending the new samples to trajectory

for i in range(100000):
  state=Buck.Next_State(state)
  np.append(trajectory,state,axis=1)

这意味着我无法将 (2,) 维数组转换为 (2,2) 维数组。

Answer 1

state 需要是列向量才能进行乘法运算。目前它只是一个一维数组。您可以添加单个维度，或者将 state 设置为单行的二维数组并转置:

state=np.array([2,3])[:,None] 

或者

state=np.array([[2,3]]).T

但是，如果您的任务是将所有状态附加到轨迹中，那么您还需要更改两件事:

1. 您需要将状态复制到轨迹。目前，您仅向其提供一个切片，因此修改轨迹也会修改状态。

2. np.append 输出新附加的数组。您没有捕获该方法的输出，因此实际上没有附加任何内容。

因此:

# My initial state
import numpy as np

state=np.array([2,3])[:,None] # Change
trajectory =state.copy() # Change

# the following generate the next samples
class Buck:
  """ The following code simulates a Buck converter """
  def __init__(self,state,control):
    self.control=control
    self.state=state

  def Next_State(self):
    L, C = 1.0, 1.0
    R, G = 1.0, 1.0
    delta = 0.001


    Q = np.array([[-1.0/L,0.0],[0.0,1.0/C]])
    A = Q*np.matmul(Q,np.array([[R,1.0],[1.0,-G]]))

    next_state = state + delta*np.matmul(A,state)

    return next_state

# Here I am appending the new samples to trajectory

for i in range(100000):
  state=Buck.Next_State(state)
  trajectory = np.append(trajectory,state,axis=1) # Change