How many parameters are required to specify the density matrix of a $n$-qubit system, and how many parameters are required to specify a quantum operation (completely positive maps between states) on an $n$-qubit system?

For example, when $n=1$, there are 3 parameters required to specify the density matrix and 12 parameters are required to specify a completely positive map.


1 Answer 1


1) Density matrix

Let us admit that the most general quantum description of a system is a "not normed" density matrix (I mean a density matrix with a trace not necessarily equals to $1$).

The density matrix is a hermitian matrix, so in a $N$ dimensional space, it has $N$ real diagonal parameters and $\frac{N(N-1)}{2}$ non-diagonal complex parameters, so the total is : $N + 2 * \frac{N(N-1)}{2} = N^2$ real parameters.

To describe a n-qubit, you have $N = 2^n$.

So you have $2^{2n}$ real parameters to describe the most general n-qbit density matrix.

If you normalize the density matrix ($Tr(\rho)=1$), you need only only $2^{2n}-1$ real parameters, so for $n=1$, you will recover your $3$ parameters.

2) Unitary transformations

A unitary transformation ($U(N)$), in a $N$ dimensional space, requires $N^2$ real parameters, so in our case ($N = 2^n$), it requires $2^{2n}$ real parameters. However, mathematical states which differ only by a global phase correspond to the same physical state, so we may consider, that physically, we have only $2^{2n}-1$ real parameters, which correspond to special unitary transformations ($SU(N)$). I must say I don't understand where you get your "$12$".


3) Completely positive maps

According to this paper, the dimension of completely positive maps seems to be $N^4-N^2$. Here, we have $N=2^n$, so the dimension is $2^{4n}-2^{2n}$.

So for one qubit $(n=1; N=2)$, we would have a dimension $2^4-2^2=16-4=12$.

  • $\begingroup$ My bad, I meant general quantum operations, not just unitary transforms. $\endgroup$
    – ruadath
    Jun 13, 2014 at 14:11
  • $\begingroup$ @ruadath : What do yo mean exactly by "general quantum operations" ? $\endgroup$
    – Trimok
    Jun 14, 2014 at 10:57
  • $\begingroup$ Completely positive maps $\endgroup$
    – ruadath
    Jun 16, 2014 at 12:55
  • $\begingroup$ @ruadath : I update the answer $\endgroup$
    – Trimok
    Jun 17, 2014 at 8:35
  • $\begingroup$ Alright cool, I will vote up once I get the necessary reputation. $\endgroup$
    – ruadath
    Jun 17, 2014 at 11:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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