Efficient and accurate algorithms are necessary to reconstruct particles in the highly granular detectors anticipated at the High-Luminosity Large Hadron Collider and the Future Circular Collider. We study scalable machine learning models for event reconstruction in electron-positron collisions based on a full detector simulation. Particle-flow reconstruction can be formulated as a supervised learning task using tracks and calorimeter clusters. We compare a graph neural network and kernel-based transformer and demonstrate that we can avoid quadratic operations while achieving realistic reconstruction. We show that hyperparameter tuning significantly improves the performance of the models. The best graph neural network model shows improvement in the jet transverse momentum resolution by up to 50% compared to the rule-based algorithm. The resulting model is portable across Nvidia, AMD and Habana hardware. Accurate and fast machine-learning based reconstruction can significantly improve future measurements at colliders.

A hybrid dual-readout calorimeter concept, comprising both electromagnetic and hadronic sections, has recently been proposed to meet the performance requirements of experiments at future e$^{+}$e$^{-}$ colliders. The front compartment consists of a homogeneous electromagnetic calorimeter made of high-density crystals, each coupled to a pair of Silicon Photomultipliers (SiPMs) providing the simultaneous readout of scintillation and Cherenkov light. To efficiently detect Cherenkov photons in the presence of dominant scintillation signals, an optical filter is placed in front of one of the two SiPMs to suppress photons in the wavelength region corresponding to that of scintillation emission. % In this study, PWO, BGO, and BSO crystals with different dimensions were tested to measure their scintillation light yield and decay time, as well as their transmission and emission spectra. A set of $\sim 100~\rm \mu m$-thick optical filters was also characterized by measuring their transmittance curves. The experimental results were used to model and estimate the expected filter performance in attenuating scintillation light for the various crystals. % The performance of each filter was experimentally validated by measuring the crystal light output with and without the filter using a $^{22}$Na radioactive source and a LYSO:Ce crystal, confirming the accuracy of the calculations. % The results show that interference filters are unsuitable for this application because their transmittance strongly depends on the photon incidence angle. Conversely, two absorptive long-pass filters with cutoff wavelengths around 590~nm were found to block more than 99\% of the scintillation light from PWO crystals, satisfying the calorimeter specifications.