关于getActionNbghs函数
tangarfff opened this issue · 2 comments
你好，关于getActionNbghs这个函数的定义有些不太确定的地方。
梳理了一下getActionNbghs函数输入输出：
输入：矩阵，整数N
输出：整数N,整数N右邻居对应的值（邻居 >= 0）。
整数N,整数N（邻居 < 0）
getActionNbghs函数在JSSP_Env中的调用代码：
# adj matrix
precd, succd = self.getNghbs(action, self.opIDsOnMchs)
self.adj[action] = 0
self.adj[action, action] = 1
if action not in self.first_col:
self.adj[action, action - 1] = 1
self.adj[action, precd] = 1
self.adj[succd, action] = 1
所以getActionNbghs函数是想要获得下一个operation的值是吗？如果下一个operation<0, 说明下一个operation不能被开始对吗？为什么[action, precd] 和[succd, action]都要设置成1呢？
(A little tip for you that please ask the question in English such that others in the community can be benefit from your questions)
If I recall correctly, the functioning of getActionNbghs() is to compute the machine successor and predecessor of a node, say 16.
From the example I have provided in the code base:
if __name__ == '__main__':
opIDsOnMchs = np.array([[7, 29, 33, 16, -6, -6],
[6, 18, 28, 34, 2, -6],
[26, 31, 14, 21, 11, 1],
[30, 19, 27, 13, 10, -6],
[25, 20, 9, 15, -6, -6],
[24, 12, 8, 32, 0, -6]])
action = 16
precd, succd = getActionNbghs(action, opIDsOnMchs)
print(precd, succd)
If you run the program you will get:
33 16
That is, the machine predecesor of 16 is 33, and the machine successor is itself 16 since there is no successor yet (refer to the matrix input opIDsOnMchs, next number on the right of 16 is -1, and -1 means no oeprations here)
The rest of the code:
precd, succd = self.getNghbs(action, self.opIDsOnMchs)
self.adj[action] = 0
self.adj[action, action] = 1
if action not in self.first_col:
self.adj[action, action - 1] = 1
self.adj[action, precd] = 1
self.adj[succd, action] = 1
they are just for adjusting the adjacent matrix that connects action with the computed machine predecessor precd and machine successor succd.
Can reopen it if there is further questions regarding this issue.