列表和 numpy arrays python

Question

I actually have multiple questions that are related.

I have the following class:

class environment_step:
    def __init__(self,solar,battery,allocation,G1,G2):
        self.solar = solar
        self.battery = battery
        self.allocation = allocation
        self.G1 = G1
        self.G2 = G2

    def step(self):
        self.residual = self.solar+self.battery-self.allocation
        self.reward = min(0,max(self.residual*self.G1,self.residual*self.G2))
        self.cost = -self.reward
        self.battery = max(0,self.residual)
        return self.reward, self.cost, self.battery

I use it as follows

reward = []
B = np.zeros(1)
allocation = np.ones(1)*12
G1 = np.ones(24)
G2 = np.ones(24)*2
solar = np.random.rand(24)
for t in range(24):
    environment = environment_step(solar[t], B, allocation, G1[t], G2[t])
    R, C, B = environment.step()
    reward.append(R)

Where solar, battery, allocation, G1, and G2 are numpy arrays. When I check R, if it is 0, then I just get 0, but for something different than zero, I get [[-2.114]] or whatever the number is. Why am I getting that instead of [-2.114] ?

I run the environment multiple times with different values for the entries and append R to the list reward. That list is going to look something like the following: [0, 0, array([[-2.114]], dtype = float32), 0, array([[-1.324]], dtype = float32)] . I convert that list reward to a pytorch tensor by reward = torch.tensor(reward, dtype=torch.float32) . When the list begins with 0, as in the example list I showed, it works perfectly. However, if the list starts with non-zero, for instance [array([[-2.114]], dtype =float32) 0, 0, 0, array([[-1.324]], dtype = float32)] , then I get the error TypeError: not a sequence . I suspect the problem is the double brackets of array, but I am not sure. That is why I'd like to fix that before (the above paragraph).

Even though, I am working with torch tensors, I think the problem is happening before with the arrays. So, what is the difference between let's say [1.2, 1.42, 3.13] and [array([1.2], dtype=float32), array([1.42], dtype=float32), array([3.13], dtype=float32)] ? or even the same but with double brackets (my case). I know that lists are general and admit texts, whereas arrays only admit numbers, but from this there seem to be more to it, I am not sure though. I'd appreciate any help!

Edit: I have added a minimum working example.

Answer 1

I was able to fix the problem by doing

R[0], B[0], C[0] = environment.step()

as opposed to what I described initially. I am writing this in case it is useful to someone. My list looks now like this: [array([-1.23]), array([-2.32]), array([0.])] , which makes more sense.