1: \begin{abstract}

2: We give a general setting for Cramér's large deviations theorem for the empirical means $\overline{X}_n$ of a sequence $(X_n)_{n \geqslant 1}$ of i.i.d. random vectors, which contains Cramér's theorem in a Banach space and Sanov's theorem. We define the notion of \emph{measurable locally convex space} in which the variables $\overline{X}_n$ are indeed measurable. We obtain the standard weak large deviations principle for the sequence $(\overline{X}_n)_{n \geqslant 1}$ and, having a close look at the notion of convex tension, we prove that the upper bound holds for convex sets. Hence we show that the identification between the entropy and the opposite of the Fenchel-Legendre transform of the pressure, $s = -p^*$, holds in any measurable locally convex space. The proof is based on convex duality and monotone convergence and does not resort to the law of large numbers or any other limit theorem. We also show that, if $\mu$ is the law of $X_1$, then $\overline{\textup{dom}(s)} = \textup{cosupp}(\mu)$.

3: \end{abstract}