Short-timescale variation in the submillimeter flux of Sagittarius A*

We present a study of short-timescale $340$ GHz flux-density variability of Sagittarius A* (Sgr A*) using ALMA Cycle 3 observations. Careful self-calibration enabled snapshot imaging with $10$ s integrations, achieving an effective image-domain SNR exceeding 10⁸ and allowing high-cadence monitoring of major Galactic Center sources. To suppress common-mode atmospheric and instrumental fluctuations, we measured the flux of Sgr A* relative to multiple non-variable sources within the same field of view. We further quantified and corrected apparent variability induced by time-dependent u–v coverage and the associated PSF changes by performing imaging simulations with a static input model. These procedures isolate the intrinsic intensity variations of Sgr A* with substantially reduced non-source contamination. We searched for characteristic variability timescales over 20\, s< τ ≤ Tobs/3 using structure-function analysis, the Lomb–Scargle method, and state-space-model-based autoregressive spectral analysis. No dominant narrow periodic component is detected. Instead, we identify a distinct short-timescale, flat (white-noise-like) regime at τ 2.3–6.3\,min, followed by red-noise-like behavior at longer timescales. The short-timescale white-noise regime appears in both active and quiescent phases, indicating statistically independent fluctuations on these timescales. We interpret the upper boundary of the white-noise regime as an empirical transition timescale, below which fluctuations remain effectively decorrelated and above which temporally correlated variability emerges. Existing theoretical and numerical studies of black-hole accretion flows have not, to our knowledge, explicitly predicted a physically white power spectrum at the shortest timescales, although they do not necessarily exclude such behavior. Reported results have more commonly shown red-noise-like or broken-power-law variability, and the physical origin of the flat short-timescale component identified here therefore remains uncertain.