Transformation on a composite system in terms of transformation on the individual basis vectors

26 Views Asked by At

Suppose I have two Hilbert spaces: $\mathcal{H}_1$ with the basis $\{| 0 \rangle_1, | 1 \rangle_1 \}$ and $\mathcal{H}_2$ with the basis $\{| 0 \rangle_2, | 1 \rangle_2 \}$, and I have a unitary $U$ acting on the composite Hilbert space $\mathcal{H}_1 \otimes \mathcal{H}_2$. Can I find a unitary $U^{'}$ acting on the Hilbert space $\mathcal{H}_1 \oplus \mathcal{H}_2$ spanned by the basis vectors $\{ |0 \rangle_1, | 1 \rangle_1, |0 \rangle_2, | 1 \rangle_2 \}$, such that the action on the composite Hilbert space is equivalent?

In other words, given an operator $U$ in $\mathcal{H}_1 \otimes \mathcal{H}_2$, find $U^{'}$ in $\mathcal{H}_1 \oplus \mathcal{H}_2$ such that

$$ \begin{bmatrix} |0 \rangle_1 \\ |1 \rangle_1 \\ |0 \rangle_2 \\ |1 \rangle_2 \end{bmatrix} \longrightarrow U^{'} \begin{bmatrix} |0 \rangle_1 \\ |1 \rangle_1 \\ |0 \rangle_2 \\ |1 \rangle_2 \end{bmatrix} \implies \begin{bmatrix} |0 \rangle_1 |0 \rangle_2 \\ |0 \rangle_1 |1 \rangle_2 \\ |1 \rangle_1 |0 \rangle_2 \\ |1 \rangle_1 |1 \rangle_2 \end{bmatrix} \longrightarrow U \begin{bmatrix} |0 \rangle_1 |0 \rangle_2 \\ |0 \rangle_1 |1 \rangle_2 \\ |1 \rangle_1 |0 \rangle_2 \\ |1 \rangle_1 |1 \rangle_2 \end{bmatrix} $$


Cross-posted on qc.SE

1

There are 1 best solutions below

5
On

After digesting Norbert Schuch's comments plus reading the OP carefully again plus reviewing some old notes I have to convert the previously accepted answer to a long comment:

  • We know that the "one-particle", aka single qubit, Hilbert space ${\cal H}$ describing only spin up/down is spanned by $\{|0\rangle,|1\rangle\}$ and may be identified with $\mathbb C^2$. The two-qubit Hilbert space can therefore be viewed as $\mathbb C^2\otimes\mathbb C^2$ which is spanned by all Kronecker products of two complex two-vectors, i.e., by all matrices $$ \begin{pmatrix}\varphi_1\psi_1&\varphi_1\psi_2\\\varphi_2\psi_1&\varphi_2\psi_2\end{pmatrix}\,,\quad\varphi_1,\varphi_2,\psi_1,\psi_2\in\mathbb C\,. $$

  • As a vector space $\mathbb C^2\otimes\mathbb C^2$ is obviously isomorphic to $\mathbb C^4=\mathbb C^2\color{red}{\oplus}\mathbb C^2\,.$ However, $\mathbb C^4=C^2\color{red}{\oplus}\mathbb C^2$ with its canonical scalar product $$ \langle \mu,\nu\rangle=\overline{\mu}_1\nu_1+\overline{\mu}_2\nu_2+\overline{\mu}_3\nu_3+\overline{\mu}_4\nu_4 $$ is not the right Hilbert space to describe the two-qubit system.

  • The right scalar product for $\mathbb C^2\otimes\mathbb C^2$ starts with the definition $$\tag{1} \langle \varphi\otimes\psi,\mu\otimes\nu\rangle:=\langle\varphi,\mu\rangle\langle\psi,\nu\rangle=(\overline{\varphi}_1\mu_1+\overline{\varphi}_2\mu_2)(\overline{\psi}_1\nu_1+\overline{\psi}_2\nu_2) $$ for product states which can be linearly extended to all states of $\mathbb C^2\otimes\mathbb C^2$ since they are linear combinations of product states. The fact that we norm all single qubit states to one is also important to make the scalar product well defined.

  • To see that $\mathbb C^2\otimes\mathbb C^2$ with (1) is the right Hilbert space, and not $\mathbb C^2\color{red}{\oplus}\mathbb C^2\,,$ consider three qubits. Their Hilbert space is -as we know- $\mathbb C^2\otimes\mathbb C^2\otimes\mathbb C^2$ which is eight-dimensional and not six-dimensional like $\mathbb C^6$.

  • To make a long story short, the coincidence that $\mathbb C^2\otimes\mathbb C^2$ is isomorphic to $\mathbb C^2\color{red}{\oplus}\mathbb C^2$ (as a vector space) led me to formulate the below answer which does not really address your issue.

Previous answer:

If you identify each Hilbert space ${\cal H}_i$ with $\mathbb C^2$ then ${\cal H}_1\otimes {\cal H}_2$ will become isomorphic to $\mathbb C^4$. A unitary $U'$ that acts only on ${\cal H}_1$ would be a matrix $$ U'=\begin{pmatrix}u_{11}&u_{12}& 0 & 0\\ u_{21}&u_{22}&0&0\\ 0&0&1&0\\0&0&0&1 \end{pmatrix}\,. $$ If you compose this with a unitary $U''$ that acts only on ${\cal H}_2$ you will end up with

$$ U''U'=\begin{pmatrix}u_{11}&u_{12}& 0 & 0\\ u_{21}&u_{22}&0&0\\ 0&0&u_{33}&u_{34}\\0&0&u_{43}&u_{44} \end{pmatrix}\,. $$ It is clear that not every unitary $U:\mathbb C^4\to\mathbb C^4$ can have this form.

In the language of tensor products: $U(2)\otimes U(2)$ is only a subgroup of $U(4)$ as mentioned in ZeroTheHero's comment.