{"id":"arxiv/2606.08498","url":"https://arxiv.org/abs/2606.08498","tags":["math.ST","stat.ME"],"score":0,"title":"Tests for Independence of High-Dimensional Nonstationary Time Series","domain":"math.ST","source":"arxiv","authors":["Yunyi Zhang"],"reviews":0,"subject":"math.ST","summary":"This manuscript studies the problem of independence testing between two high-dimensional time series without assuming weak stationarity, that is, allowing their autocovariances to vary over time. To this end, we propose a bimodal weighted-average test statistic that removes the bias induced by temporal dependence under the null hypothesis, thereby avoiding the need to whiten the time series prior to hypothesis testing -- a procedure that is challenging in high-dimensional and nonstationary settings. To facilitate statistical inference, we develop a dependent wild bootstrap procedure. On the…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-06-07","reactions":{}}