I have an expression for the varaince of measurement noise obtained from an estimator. The measurement noise is additive white gaussian having values in complex domain. I have squared the term $(.)^2$, should I be squaring or just implement without squaring? But if I dont square the term, then I am getting a complex valued variance which is incorrect as variance is always real valued.

Kindly suggest what to do.


1 Answer 1


You must put:


that is, replacing your code by:

for i = 1: N
   dy(i,1) = (y(i)-yhat(i))*conj(y(i)-yhat(i));
sigma2 = 1/(N-1)*sum(dy);

The variance is the second moment of a probability distribution, which can be defined for any metric space in the sense of the definition over the random variable (integral of a density), or in any vector space, in the sense of the estimator of the random variable (sum of n samples).

For $X$ a random variable $\in \mathbb{R}^n$, with probability density function $p(x):I_X \subset \mathbb{R}^n \rightarrow \mathbb{R}$, the second moment $\in \mathbb{R}$ is:

$$\sigma^2=\mathbf{E}\{(X-\mu)^2\}=\int_{I_X} (x-\mu)^2 p(x)dx$$

which is the variance for $n=1$ and the covariance matrix in general.

For other fields $\mathbb{F}$, complex and even real numbers, the metric $d(x,y): \mathbb{F}x\mathbb{F}\rightarrow \mathbb{R}$ can redefine the moment definition

$$\sigma^2_{\mathbb{F}}=\mathbf{E}_{\mathbb{F}}\{(X-\mu)^2\}=\int_{I_X} d(x,\mu)^2 p(x)dx$$

For example, for the complex field, the metric function can be the standard distance from the hermitian inner product, and thus we have the variance definition:

$$d(x,y)=\sqrt{(x-y)\overline{(x-y)}}$$ $$\sigma^2_{\mathbb{C}}=\mathbf{E}_{\mathbb{C}}\{(X-\mu)^2\}=\int_{I_X} (x-y)\overline{(x-y)} p(x)dx$$

But the standard product, without the conjugated term, though can be a valuable figure (named as the "complementary variance") is not a second moment:

$${\sigma^2}_{\mathbb{C}comp}=\int_{I_X} (x-y)^2 p(x)dx$$

Of course, you always could be more "reasonable" and prefer to drop the complex analysis and move onto a $\mathbb{C}^2$ cartesian plane, and start taking covariances instead of a single number which dont dive you the full distribution information of your stochastics.

  • 1
    $\begingroup$ Thank you for your reply, I have taken the complex conjugate of the second term if you may kindly see the code in my question. Moreover, there is a multiplication term with $y$ in the expression $(y(y-yhat))$. I took the conjugate for that term and I think you have not mentioned this part which is how to handle multiplication with iteself for scalar complex valued number. What I have done is the same as your answer except for the doubt in the term $y_n \times y_n$. I plugged you code and still I am getting very large value for the variance like 2 to three digit number. $\endgroup$
    – SKM
    Dec 4, 2016 at 19:06
  • $\begingroup$ I checked your code and you are calculating a different expression which dont make any known symmetry to me. You should take A=y(i) and B=yhat(i). The square root of the result should preserve the dimensional symmetry. Could you plot the point cloud in the z plane and the variance result you are having? $\endgroup$
    – Brethlosze
    Dec 5, 2016 at 2:19
  • $\begingroup$ Besides, you can made two different calculations with A1=real(y(i)) B1=real(yhat(i)) and for A2=imag(y(i)) B2=imag(yhat(i)). sigma1 and sigma2 should trivially preserve the dimensionality...... $\endgroup$
    – Brethlosze
    Dec 5, 2016 at 2:22
  • $\begingroup$ Regarding your comment, the complex product is handled as ussualy defined in the $\mathbb{C}$ field, and there are no y(i)*y(i) terms, only y(i)*conj(y(i)).Do you remember the bracket notation? $<a|b>=a\overline{b}$ $\endgroup$
    – Brethlosze
    Dec 5, 2016 at 2:24
  • 1
    $\begingroup$ Ive just checked your questions.... I will address them and kill your doubts one by one.... :) $\endgroup$
    – Brethlosze
    Dec 7, 2016 at 0:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.