Let $X$ be a vector space, $Y$ a subspace of $X$, and $A \subset X$.
Then algebraic interior of $A$ with respect to $Y$ is defined as $$ \operatorname{aint}_{Y} (A) := \{x \in X \mid \forall y\in Y, \exists t >0: [x, x+t y] \subset A \}. $$ Clearly, $\operatorname{aint}_{Y} (A) \subset A$. If $A$ is convex, then the above definition reduces to $$ \operatorname{aint}_{Y} (A) = \{x \in X \mid \forall y\in Y, \exists t >0: x+t y \in A \}. $$
The algebraic interior of $A$ is defined as $$ \operatorname{aint} (A) := \operatorname{aint}_{X} (A). $$
Then I'm reading below result and its proof from this lecture note
Observation 0.7. Let $A$ be a subset of a vector space $X$ such that $\operatorname{aint} (A) \neq \emptyset$. Let $\alpha \in \mathbb R$ and $\ell: X \to \mathbb R$ be a nontrivial linear functional. Then the following statements are equivalent.
- (i) $A \subset \{\ell \le \alpha\}$.
- (ii) $\operatorname{aint} (A) \subset \{\ell \le \alpha\}$.
- (iii) $\operatorname{aint} (A) \subset \{\ell < \alpha\}$.
Proof.
The implication (i) $\implies$ (ii) is obvious.
(ii) $\implies$ (iii) If (iii) is false, there exists $a \in \operatorname{aint} (A)$ such that $\ell(a)=\alpha$. Take a vector $v \in X$ such that $\ell(v)>0$. There exists $t>0$ with $a+t v \in A$, but then necessarily $a+\frac{t}{2} v \in \operatorname{aint} (A)$ and $\ell\left(a+\frac{t}{2} v\right)>\alpha$; thus (ii) is false.
(iii) $\implies$ (i) Assume that (iii) holds while (i) is false. Fix $a \in \operatorname{aint} (A)$ and $x \in A$ with $\ell(x)>\alpha$. Then the relatively open segment $(a, x)$ is contained in $\operatorname{aint} (A)$ and intersects the hyperplane $\ell^{-1}(\alpha)$, which contradicts (iii).
I suspect the the proof only holds with an additional assumption that $A$ is convex. Could you confirm if my understanding is correct?