Custom Loss Function to maximise the difference

Hi team,
I want to define a loss function as to maximise a-b where b’s value is fixed in the starting of the function while a’s value keep changing at each iteration. Please help me to define a correct loss function so as to maximise a-b.Currently I have defined L as a loss function.Is this correct to represent the above condition.
L = max(0, a-b)
where a = torch.tensor(3.04)
and b =torch.tensor(7.698) at first iteration.
How to represent this L ?Thanks in advance. Please help me to resolve this issue.