Let $A$ is a commutative ring with unit 1. Let $\alpha,\beta$ be ideals of $A$.
I have proved that $\alpha \cap \beta=\alpha \beta$ if $\alpha+\beta=(1)$,
I wonder if the converse is true or not?
Let $A$ is a commutative ring with unit 1. Let $\alpha,\beta$ be ideals of $A$.
I have proved that $\alpha \cap \beta=\alpha \beta$ if $\alpha+\beta=(1)$,
I wonder if the converse is true or not?
Concerning your question, it may be interesting to see counterexamples not involving the zero ideal. So here are some:
Let $A = \mathbb{Z}[x]$, $\alpha = (2)$ and $\beta = (x)$. Then one easily checks that $\alpha \beta = \alpha \cap \beta$, but $ \alpha + \beta = (2,x)$ is a maximal ideal and therefore proper.
Analoguous examples can be constructed in any factorial domain $A$ that has two distinct prime ideals of height greater than $1$ containing two distinct height $1$ primes.
A phenomenon from which we also can construct examples is the following:
Let $A$ be a commutative ring with unity. There is a general notion of ideals $\alpha \subseteq A$ with the property that for all ideals $\beta \subseteq A$, we have $\alpha \cap \beta = \alpha \beta$. Such ideals are called pure. We have the following characterization of pure ideals, that you can find in any book on multiplicative ideal theory:
Theorem: The following assertions are equivalent for an ideal $\alpha \subseteq A$:
Pure ideals $\alpha$ are very special in the sense that they satisfy the property you were asking, no matter which ideal $\beta$ you take. We can easily construct examples of non-zero pure ideals, using properties 3. and 4. from above:
Examples:
Let $I = (x^2 - x)$ the ideal of $\mathbb{Z}[x]$ generated by $x^2 - x$ and let $A = \mathbb{Z}[x]/I$. The ideal $\alpha = (x + I)$ in $A$ is generated by an the idempotent element $x + I$ and is therefore idempotent. As a principal ideal it is locally principal, hence it is pure by property 3. in the theorem above.
Let $A$ be an infinite product of fields and $M \subseteq A$ a maximal ideal. Then the canonical map $A \to A_M$ with $a \mapsto \frac{a}{1}$ has kernel $M$ (you can find a proof of this here). Therefore $M_M = (0)_M$. For every other maximal ideal $N \subseteq A$, we clearly have $M_N = A_N$. By property 4. of the theorem, we have that $M$ is pure.