Second input must be a scalar or non-empty vector of real values between 0 and 100. When dim is not 1 (non-tall dimension), exact computation is more efficient. Try setting the 'Method' parameter to 'exact' instead. 只有 tall 数组为 tall 列向量时，才支持计算其第一个维度的百分位数或分位数。 The value of {0} must be real. The 'approximate' method does not support object inputs such as gpuArray. The first input must be numeric. First input must be a numeric array for the 'approximate' method.