Is the var function of Matlab wrong?

Today, the power of the constellation symbol is calculated. In the past, when analyzing the data, the variance was calculated directly by matlab's var function, and no problem was found. The number of constellations calculated today is relatively small, which is strange no matter how it is calculated.

Refer to the formula known by Baidu

That's right. Is it because the mean is not zero?

For simple

A = [- 3-113]; the mean is 0

Let var (a) = 6.6667. W..H..Y..

Help var took a look.

Variance, according to the definition in higher numbers, can be divided into biased estimators (biased estimator) and unbiased estimators (unbiased estimator). The difference lies in whether the denominator of the above formula is divided by n or nmai 1.

The default var for Matlab calculates unbiased estimates. Looking back at my test example, isn't it? the denominator is divided by 3. The solution is to add a parameter 1 and modify it to a biased estimate.

Refer to blog posts:

Http://blog.163.com/wanglei2146073@126/blog/static/90689607201132294156177/