In an attempt to transform following equivalent definitions of the cartesian product from the first into the second, trying to be rather formal about it:
$$ \forall A : A \in X \times Y \leftrightarrow \exists x \exists y[A = (x, y) \wedge x \in X \wedge y \in Y] \\ \forall x \forall y : (x, y) \in X \times Y \leftrightarrow x \in X \wedge y \in Y $$
My attempts led to the following steps: $$ \begin{align} &\text{0. Suppose} &\Longrightarrow &\forall A : A \in X \times Y \leftrightarrow \exists x \exists y[A = (x, y) \wedge x \in X \wedge y \in Y] \\ &\text{1. Universal elimination} &\Longrightarrow &A_0 \in X \times Y \leftrightarrow \exists x \exists y[A_0 = (x, y) \wedge x \in X \wedge y \in Y] \\ &\text{2. Existential elimination} &\Longrightarrow &A_0 \in X \times Y \leftrightarrow A_0 = (x_0, y_0) \wedge x_0 \in X \wedge y_0 \in Y \\ &\text{3. Substitution} &\Longrightarrow &(x_0, y_0) \in X \times Y \leftrightarrow A_0 = (x_0, y_0) \wedge x_0 \in X \wedge y_0 \in Y \\ &\text{4. Conjunction elimination} &\Longrightarrow &(x_0, y_0) \in X \times Y \leftrightarrow x_0 \in X \wedge y_0 \in Y \\ &\text{5. Universal generalization} &\Longrightarrow &\forall x \forall y: (x, y) \in X \times Y \leftrightarrow x \in X \wedge y \in Y \\ \end{align} \\ \blacksquare . $$
To me these look and feel like valid steps, however, I failed to find or understand why exactly the universal generalization is indeed allowed in this particular case, considering one of the steps involves existential elimination, which does not guarentee the arbitrariness of $x_0, y_0$.
To justify that intuitively, each arbitrary choice of $A_1 \neq A_2 \rightarrow x_1 \neq x_2$ and $y_1 \neq y_2$ by the definition of an ordered pair, so for each arbitrary choice of $A$ there is a unique pair $(x, y)$ for which the definition holds. This ties the choice of the pair entirely to the choice of the arbitrary element, making it just as arbitrary.
What I wish to know, if at all possible, are three things:
- Is this proof even logically correct, or are some of the steps invalid within natural deduction / first order logic?
- What is the formal way to define the dependence of the pair on the arbitrary element $A$ and how do we show it's uniquness?
- What is the formal way to define the applicability of the universal generalization in this particular case?
I would greatly appreciate it if the answers detailed both the general definitions and the special application for this particular case. Thanks in advance!
If any notation makes the intent of some sentence unclear, or if there are any language-related errors, please comment the specific issue so it can be fixed.