One-Way High-Dimensional ANOVA

ANOVA is one of the most important tools in comparing treatment means among different groups repeated measurements. The classical F test routinely used to if are same across groups. However, it inefficient when number or dimension gets large. We propose a smoothing truncation deal with this problem. It shown theoretically and empirically that proposed works regardless dimension. limiting null alternative distributions our statistic established for fixed diverging treatments. Simulations demonstrate superior performance over F settings.