Orchestrating resistance trajectories in TaO _x memristors: the synergy of pulse stimulus and initial filament states

/Pt memristor, employing pulsed operations rather than voltage sweeps. Our results reveal that resistance evolution is primarily governed by the co-modulation between pulse stimuli and initial filament states. While previous studies have considered these variables separately, we establish the first comprehensive framework that systematically elucidates their combined influence. Contrary to the conventional view that the reset process corresponds to a purely monotonic resistance increase, we identify four distinct resistance evolutions under reset pulses: quasi-stationary, valley-switching, abrupt-switching, and deep-dip rebound. These characteristics are further leveraged as a programming strategy, enabling gradual resistance modulation and reliable access to intermediate resistances. By integrating these physical insights into a deterministic evolution framework, this study provides a blueprint for state-dependent pulse programming, offering new avenues for precise resistance control in next generation in-memory computing architectures.