# Unary notation representation in Turing Machine

What can be a good convention to represent positive and negative integers in unary notation on a Turing machine? I want to design a Turing machine which acts as a subtractor.So should I partition the integer range to represent positive and negative integers in unary notation or is there some other way?

A reasonable convention would be to take any enumeration of the integers and use it. For example, this enumeration $$0,1,-1,2,-2,3,-3,\ldots$$ corresponds to this encoding: \begin{align*} 0 &\to \epsilon \\ 1 &\to 1 \\ -1 &\to 11 \\ 2 &\to 111 \\ -2 &\to 1111 \\ 3 &\to 11111 \\ -3 &\to 111111 \end{align*}