Latest results from Double Chooz

Double Chooz (DC) is a reactor neutrino experiment running at the Chooz nuclear power plant in France. In 2011, DC first reported indication of non-zero $\theta_{13}$ neutrino mixing angle with the far detector (FD) located at the maximum of oscillation effects. A robust observation of $\theta_{13}$ followed in 2012 by other reactor experiments with multiple detector configurations. In the DC single detector configuration, the precision of $\theta_{13}$ was totally dominated by the reactor flux uncertainty. Since 2015 DC runs in a multi-detector configuration reducing strongly the impact of several otherwise dominating systematics. Due to the unique almost iso-flux site configuration, the DC near detector (ND) becomes a direct accurate non-oscillation reference to the FD. The most recent DC $\theta_{13}$ measurement with two detectors uses a novel analysis method aimed to maximize the accuracy; i.e. the goodness of the central value of $\theta_{13}$. Despite a major increase of statistical power, a conservative approach on systematics has been adopted for now. The latest multi-detector DC $\theta_{13}$ value presents a deviation with respect to the world average. The combined "reactor-$\theta_{13}$" measurement is expected to remain as the world reference for decades, relying on systematic uncertainties in the per mil level. Thus, the redundancy of multiple experiments is critical to ensure the accuracy and precision of the measurement, which is reference for current and future projects sensitive to CP-violation and atmospheric mass hierarchy observables. Beyond oscillation physics, DC has articulated one of the most precise single-detector setups, allowing high precision reactor spectrum characterization, including both shape and normalization. This is a particularly hot topic since DC provided the first evidence of a sizeable reactor neutrino spectral distortion in May 2014, confirmed later on by other experiments, questioning the accuracy and precision of today's reactor neutrino predictions technology.