Wie kann man die CNOT-Matrix für ein 3-Qbit-System ableiten, bei dem die Kontroll- und Ziel-Qbit nicht benachbart sind?

In einem Drei-Qbit-System ist es einfach, den CNOT-Operator abzuleiten, wenn die Kontroll- und Ziel-Qbit-Werte nebeneinander liegen. Sie müssen lediglich den 2-Bit-CNOT-Operator mit der Identitätsmatrix an der signifikanten Position des unberührten Qbit spannen:

$C_{10}|\phi_2\phi_1\phi_0\rangle = (\mathbb{I}_2 \otimes C_{10})|\phi_2\phi_1\phi_0\rangle$

Es ist jedoch nicht klar, wie der CNOT-Operator abgeleitet werden soll, wenn die Kontroll- und Ziel-QBits nicht benachbart sind:

$C_{20}|\phi_2\phi_1\phi_0\rangle$

Wie geht das?

quantum-gate gate-synthesis

— ahelwer
quelle

Antworten:

Für eine Präsentation nach ersten Grundsätzen mag ich die Antwort von Ryan O'Donnell . Aber für eine etwas höhere algebraische Behandlung würde ich es folgendermaßen machen.

Das Hauptmerkmal einer gesteuerten $U$ Operation für jede unitäre $U$ ist, dass sie (kohärent) eine Operation für einige Qubits ausführt, abhängig vom Wert eines einzelnen Qubits. Die Art und Weise , dass wir diese algebraisch explizit schreiben können (mit der Kontrolle auf der ersten Qubit) ist:

C U = | 0 ⟩ ⟨ 0 | \otimes 1 + | 1 ⟩ ⟨ 1 | \otimes U

$\mathit{CU} \;=\; \def\ket#1{\lvert #1 \rangle}\def\bra#1{\langle #1 \rvert}\ket{0}\!\bra{0} \!\otimes\! \mathbf 1 \,+\, \ket{1}\!\bra{1} \!\otimes\! U$ wobei

1

$\mathbf 1$ eine Identitätsmatrix mit der gleichen Dimension wie

U

$U$ . Hier,

| 0 ⟩ ⟨ 0 |

$\ket{0}\!\bra{0}$ und

sind Projektoren auf die Staaten

und

der Steuer Qubit - aber wir sind mit ihnen nicht hier als Elemente einer Messung, aber die Auswirkungen auf den anderen Qubits aufeinen oder anderen Unterraum des Zustandsraumes des ersten QubitAbhängigkeit zu beschreiben.

| 1 ⟩ ⟨ 1 |

$\ket{1}\!\bra{1}$

| 0 ⟩

$\ket{0}$

| 1 ⟩

$\ket{1}$

Wir können dies verwenden, um die Matrix für das Gatter abzuleiten das eine Operation an Qubit 3 ausführt , die kohärent vom Zustand von Qubit 1 abhängig ist, indem wir dies als gesteuertes $\mathit{CX}_{1,3}$ $X$ Operation auf Qubits 2 und 3: $(\mathbf 1_2 \!\otimes\! X)$

\begin{aligned} {C X}_{1, 3} & = | 0 ⟩ ⟨ 0 | \otimes 1_{4} + | 1 ⟩ ⟨ 1 | \otimes (1_{2} \otimes X) \\ = [\begin{matrix} 1_{4} & 0_{4} \\ 0_{4} & (1_{2} \otimes X) \end{matrix}] = [\begin{matrix} 1_{2} & 0_{2} & 0_{2} & 0_{2} \\ 0_{2} & 1_{2} & 0_{2} & 0_{2} \\ 0_{2} & 0_{2} & X & 0_{2} \\ 0_{2} & 0_{2} & 0_{2} & X \end{matrix}], \end{aligned}

$\begin{aligned} \mathit{CX}_{1,3} \;&=\; \ket{0}\!\bra{0} \otimes \mathbf 1_4 \,+\, \ket{1}\!\bra{1} \otimes (\mathbf 1_2 \otimes X) \\[1ex]&=\; \begin{bmatrix} \mathbf 1_4 & \mathbf 0_4 \\ \mathbf 0_4 & (\mathbf 1_2 \!\otimes\! X) \end{bmatrix} \;=\; \begin{bmatrix} \mathbf 1_2 & \mathbf 0_2 & \mathbf 0_2 & \mathbf 0_2 \\ \mathbf 0_2 & \mathbf 1_2 & \mathbf 0_2 & \mathbf 0_2 \\ \mathbf 0_2 & \mathbf 0_2 & X & \mathbf 0_2 \\ \mathbf 0_2 & \mathbf 0_2 & \mathbf 0_2 & X \end{bmatrix}, \end{aligned}$ wobei die beiden letzteren Blockmatrixdarstellungen sind, um Platz (und Vernunft) zu sparen.

Besser noch: Wir können erkennen, dass - auf einer mathematischen Ebene, auf der wir uns erlauben zu erkennen, dass die Reihenfolge der Tensorfaktoren nicht in einer festen Reihenfolge sein muss - die Kontrolle und das Ziel der Operation auf zwei beliebigen Tensoren liegen können Faktoren, und dass wir die Beschreibung des Operators auf allen anderen Qubits mit ausfüllen können . Dies würde es uns ermöglichen, direkt zur Darstellung zu springen $\mathbf 1_2$

\begin{aligned} {C X}_{1, 3} & = & \underset{control}{\underset{⏟}{| 0 ⟩ ⟨ 0 |}} \otimes \underset{uninvolved}{\underset{⏟}{1_{2}}} \otimes \underset{target}{\underset{⏟}{1_{2}}} & + \underset{control}{\underset{⏟}{| 1 ⟩ ⟨ 1 |}} \otimes \underset{uninvolved}{\underset{⏟}{1_{2}}} \otimes \underset{target}{\underset{⏟}{X}} \\ = & [\begin{matrix} 1_{2} & 0_{2} \\ 0_{2} & 1_{2} \end{matrix}] & + [\begin{matrix} X & 0_{2} \\ 0_{2} & X \end{matrix}] \end{aligned}

$\begin{alignat}{2} \mathit{CX}_{1,3} \;&=&\; \underbrace{\ket{0}\!\bra{0}}_{\text{control}} \otimes \underbrace{\;\mathbf 1_2\;}_{\!\!\!\!\text{uninvolved}\!\!\!\!} \otimes \underbrace{\;\mathbf 1_2\;}_{\!\!\!\!\text{target}\!\!\!\!} &+\, \underbrace{\ket{1}\!\bra{1}}_{\text{control}} \otimes \underbrace{\;\mathbf 1_2\;}_{\!\!\!\!\text{uninvolved}\!\!\!\!} \otimes \underbrace{\; X\;}_{\!\!\!\!\text{target}\!\!\!\!} \\[1ex]&=&\; \begin{bmatrix} \mathbf 1_2 & \mathbf 0_2 & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} \\ \mathbf 0_2 & \mathbf 1_2 & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} \\ \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} \\ \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} \end{bmatrix} \,&+\, \begin{bmatrix} \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} \\ \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} \\ \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & X & \mathbf 0_2 \\ \phantom{\mathbf 0_2} & \phantom{\mathbf 0_2} & {\mathbf 0_2} & X \end{bmatrix} \end{alignat}$ and also allows us to immediately see what to do if the roles of control and target are reversed:

\begin{aligned} {C X}_{3, 1} & = & \underset{target}{\underset{⏟}{1_{2}}} \otimes \underset{uninvolved}{\underset{⏟}{1_{2}}} \otimes \underset{control}{\underset{⏟}{| 0 ⟩ ⟨ 0 |}} & + \underset{target}{\underset{⏟}{X}} \otimes \underset{uninvolved}{\underset{⏟}{1_{2}}} \otimes \underset{control}{\underset{⏟}{| 1 ⟩ ⟨ 1 |}} \\ = & [\begin{matrix} | 0 ⟩ ⟨ 0 | \\ | 0 ⟩ ⟨ 0 | \\ | 0 ⟩ ⟨ 0 | \\ | 0 ⟩ ⟨ 0 | \end{matrix}] & + [\begin{matrix} | 1 ⟩ ⟨ 1 | \\ | 1 ⟩ ⟨ 1 | \\ | 1 ⟩ ⟨ 1 | \\ | 1 ⟩ ⟨ 1 | \end{matrix}] \\ = & [\begin{matrix} 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 \end{matrix} & \begin{matrix} 0 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 \\ 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 0 \end{matrix}] . \end{aligned}

$\begin{alignat}{2} \mathit{CX}_{3,1} \;&=&\; \underbrace{\;\mathbf 1_2\;}_{\!\!\!\!\text{target}\!\!\!\!} \otimes \underbrace{\;\mathbf 1_2\;}_{\!\!\!\!\text{uninvolved}\!\!\!\!} \otimes \underbrace{\ket{0}\!\bra{0}}_{\text{control}} \,&+\, \underbrace{\;X\;}_{\!\!\!\!\text{target}\!\!\!\!} \otimes \underbrace{\;\mathbf 1_2\;}_{\!\!\!\!\text{uninvolved}\!\!\!\!} \otimes \underbrace{\ket{1}\!\bra{1}}_{\text{control}} \\[1ex]&=&\; {\scriptstyle\begin{bmatrix} \!\ket{0}\!\bra{0}\!\! & & & \\ & \!\!\ket{0}\!\bra{0}\!\! & & \\ & & \!\!\ket{0}\!\bra{0}\!\! & \\ & & & \!\!\ket{0}\!\bra{0} \end{bmatrix}} \,&+\, {\scriptstyle\begin{bmatrix} & & \!\!\ket{1}\!\bra{1}\!\! & \\ & & & \!\!\ket{1}\!\bra{1} \\ \!\ket{1}\!\bra{1}\!\! & & & \\ & \!\!\ket{1}\!\bra{1} & & \end{bmatrix}} \\[1ex]&=&\; \left[{\scriptstyle\begin{matrix} 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 \end{matrix}}\right.\,\,&\,\,\left.{\scriptstyle\begin{matrix} 0 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 \\ 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 0 \end{matrix}}\right]. \end{alignat}$ But best of all: if you can write down these operators algebraically, you can take the first steps towards dispensing with the giant matrices entirely, instead reasoning about these operators algebraically using expressions such as

{C X}_{1, 3} = | 0 ⟩ ⟨ 0 | \otimes 1_{2} \otimes 1_{2} + | 1 ⟩ ⟨ 1 | \otimes 1_{2} \otimes X

$\mathit{CX}_{1,3} = \ket{0}\!\bra{0} \! \otimes\!\mathbf 1_2\! \otimes\! \mathbf 1_2 + \ket{1}\!\bra{1} \! \otimes\! \mathbf 1_2 \! \otimes\! X$ and

{C X}_{3, 1} = 1_{2} \otimes 1_{2} \otimes | 0 ⟩ ⟨ 0 | + X \otimes 1_{2} \otimes | 1 ⟩ ⟨ 1 |

$\mathit{CX}_{3,1} = \mathbf 1_2 \! \otimes\! \mathbf 1_2 \! \otimes \! \ket{0}\!\bra{0} + X \! \otimes\! \mathbf 1_2 \! \otimes \! \ket{1}\!\bra{1}$ . There will be a limit to how much you can do with these, of course — a simple change in representation is unlikely to make a difficult quantum algorithm efficiently solvable, let alone tractable by manual calculation — but you can reason about simple circuits much more effectively using these expressions than with giant space-eating matrices.

— Niel de Beaudrap
quelle

Oh yeah, I recall projectors from early on in the Mermin book. Projectors and matrix addition are a way of encoding conditional logic in matrices!

— ahelwer

"a simple change in representation is unlikely to make a difficult quantum algorithm efficiently solvable" - what about in the case of wick rotation's?

— meowzz

@meowzz: Every once in a while, such a change in notation allows you to make a conceptual advance and helps you to solve problems more easily. But not often, and probably not in the case of this particular change of notation, which is reasonably well-known. As to the specific case of the Wick rotation, however, the question I would ask is what specific advance it made possible to solve problems, and what problems it was helpful for.

— Niel de Beaudrap

This is a good question; it's one that textbooks seem to sneak around. I reached this exact question when preparing a quantum computing lecture a couple days ago.

As far as I can tell, there's no way of getting the desired 8x8 matrix using the Kronecker product $\otimes$ notation for matrices. All you can really say is: Your operation of applying CNOT to three qubits, with the control being the first and the target being the third, has the following effects:

$\lvert 000\rangle \mapsto \lvert 000\rangle$

$\lvert 001\rangle \mapsto \lvert 001\rangle$

$\lvert 010\rangle \mapsto \lvert 010\rangle$

$\lvert 011\rangle \mapsto \lvert 011\rangle$

$\lvert 100\rangle \mapsto \lvert 101\rangle$

$\lvert 101\rangle \mapsto \lvert 100\rangle$

$\lvert 110 \rangle \mapsto \lvert 111 \rangle$

$\lvert 111 \rangle \mapsto \lvert 110 \rangle$

and therefore it is given by the following matrix:

$U = \begin{bmatrix} 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 \\ 0 & 0 & 0 & 0 & 0 & 0 & 1 & 0 \end{bmatrix}$

This matrix $U$ is indeed neither $I_2 \otimes \mathrm{CNOT}$ nor $\mathrm{CNOT} \otimes I_2$ . There is no succinct Kronecker-product-based notation for it; it just is what it is.

— Ryan O'Donnell
quelle

As a general idea CNOT flips target based on control. I choose to flip the target if control is $\uparrow (= [1\ 0]^T)$ , you may choose it $\downarrow (= [0\ 1]^T)$ too. So assume any general multiparticle state $|\phi\rangle=|\uparrow_1\downarrow_2\downarrow_3....\uparrow_{n-1}\downarrow_n\rangle$ . Now you choose your control and target, lets say $i'th$ is control and $k'th$ is target. Applying CNOT on $|\phi\rangle$ will be just

C N O T | ϕ ⟩ = C N O T | ↑_{1} ↓_{2} . . . ↑_{i} . . . ↑_{k} . . . ↑_{n - 1} ↓_{n} ⟩ = | ↑_{1} ↓_{2} . . . ↑_{i} . . . ↓_{k} . . . ↑_{n - 1} ↓_{n} ⟩

$\begin{equation} CNOT|\phi\rangle=CNOT|\uparrow_1\downarrow_2...\uparrow_i...\uparrow_k...\uparrow_{n-1}\downarrow_n\rangle= |\uparrow_1\downarrow_2...\uparrow_i...\downarrow_k...\uparrow_{n-1}\downarrow_n\rangle \end{equation}$

To construct the matrix of such CNOT gate we apply $\sigma_x$ ( $x$ -Pauli matrix) if $i'th$ state is up and we apply $I$ ( $2\times2$ Identity) if $i'th$ state is down. We apply these matrices at the $k'th$ position, which is our target. Mathematically,

C N O T = [| ↑_{1} . . . ↑_{i} . . . ↑_{k - 1} ⟩ ⟨ ↑_{1} . . . ↑_{i} . . . ↑_{k - 1} | \otimes σ_{x} \otimes | ↑_{k + 1} . . . ↑_{n} ⟩ ⟨ ↑_{k + 1} . . . ↑_{n} | + a l l p e r m u t a t i o n s o f s t a t e s o t h e r t h e n i^{'} t h] + [| ↑_{1} . . . ↓_{i} . . . ↑_{k - 1} ⟩ ⟨ ↑_{1} . . . ↓_{i} . . . ↑_{k - 1} | \otimes I \otimes | ↑_{k + 1} . . . ↑_{n} ⟩ ⟨ ↑_{k + 1} . . . ↑_{n} | + a l l p e r m u t a t i o n s o f s t a t e s o t h e r t h e n i^{'} t h]

$\begin{equation} CNOT = \Big[|\uparrow_1...\uparrow_i...\uparrow_{k-1}\rangle\langle\uparrow_1...\uparrow_i...\uparrow_{k-1}|\otimes\sigma_x\otimes|\uparrow_{k+1}...\uparrow_n\rangle\langle\uparrow_{k+1}...\uparrow_n| + all\ permutations\ of\ states\ other\ then\ i'th\Big] + \Big[|\uparrow_1...\downarrow_i...\uparrow_{k-1}\rangle\langle\uparrow_1...\downarrow_i...\uparrow_{k-1}|\otimes I\otimes|\uparrow_{k+1}...\uparrow_n\rangle\langle\uparrow_{k+1}...\uparrow_n| + all\ permutations\ of\ states\ other\ then\ i'th\Big] \end{equation}$

Note $k'th$ state(target) is excluded while creating the permutation matrix and at $k'th$ position the operator $\sigma_x$ or $I$ is written.

Take an example of five qubits in which $2^{nd}$ qubit is target and $4^{th}$ is control. Lets build the permutation matrix of $CNOT$ . I take, if control is $\uparrow$ flip the target. You can take vice-versa too.

\begin{aligned} C N O T & = | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes σ_{x} \otimes | ↑_{3} ↑_{4} ↑_{5} ⟩ ⟨ ↑_{3} ↑_{4} ↑_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes σ_{x} \otimes | ↑_{3} ↑_{4} ↓_{5} ⟩ ⟨ ↑_{3} ↑_{4} ↓_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes σ_{x} \otimes | ↓_{3} ↑_{4} ↑_{5} ⟩ ⟨ ↓_{3} ↑_{4} ↑_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes σ_{x} \otimes | ↓_{3} ↑_{4} ↓_{5} ⟩ ⟨ ↓_{3} ↑_{4} ↓_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes σ_{x} \otimes | ↑_{3} ↑_{4} ↑_{5} ⟩ ⟨ ↑_{3} ↑_{4} ↑_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes σ_{x} \otimes | ↑_{3} ↑_{4} ↓_{5} ⟩ ⟨ ↑_{3} ↑_{4} ↓_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes σ_{x} \otimes | ↓_{3} ↑_{4} ↑_{5} ⟩ ⟨ ↓_{3} ↑_{4} ↑_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes σ_{x} \otimes | ↓_{3} ↑_{4} ↓_{5} ⟩ ⟨ ↓_{3} ↑_{4} ↓_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes I \otimes | ↑_{3} ↓_{4} ↑_{5} ⟩ ⟨ ↑_{3} ↓_{4} ↑_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes I \otimes | ↑_{3} ↓_{4} ↓_{5} ⟩ ⟨ ↑_{3} ↓_{4} ↓_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes I \otimes | ↓_{3} ↓_{4} ↑_{5} ⟩ ⟨ ↓_{3} ↓_{4} ↑_{5} | \\ + | ↑_{1} ⟩ ⟨ ↑_{1} | \otimes I \otimes | ↓_{3} ↓_{4} ↓_{5} ⟩ ⟨ ↓_{3} ↓_{4} ↓_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes I \otimes | ↑_{3} ↓_{4} ↑_{5} ⟩ ⟨ ↑_{3} ↑_{4} ↓_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes I \otimes | ↑_{3} ↓_{4} ↓_{5} ⟩ ⟨ ↑_{3} ↓_{4} ↓_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes I \otimes | ↓_{3} ↓_{4} ↑_{5} ⟩ ⟨ ↓_{3} ↓_{4} ↑_{5} | \\ + | ↓_{1} ⟩ ⟨ ↓_{1} | \otimes I \otimes | ↓_{3} ↓_{4} ↓_{5} ⟩ ⟨ ↓_{3} ↓_{4} ↓_{5} | \end{aligned}

$\begin{align} CNOT & = |\uparrow_1\rangle\langle\uparrow_1|\otimes\sigma_x\otimes|\uparrow_3\uparrow_4\uparrow_5\rangle\langle\uparrow_3\uparrow_4\uparrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes\sigma_x\otimes|\uparrow_3\uparrow_4\downarrow_5\rangle\langle\uparrow_3\uparrow_4\downarrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes\sigma_x\otimes|\downarrow_3\uparrow_4\uparrow_5\rangle\langle\downarrow_3\uparrow_4\uparrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes\sigma_x\otimes|\downarrow_3\uparrow_4\downarrow_5\rangle\langle\downarrow_3\uparrow_4\downarrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes\sigma_x\otimes|\uparrow_3\uparrow_4\uparrow_5\rangle\langle\uparrow_3\uparrow_4\uparrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes\sigma_x\otimes|\uparrow_3\uparrow_4\downarrow_5\rangle\langle\uparrow_3\uparrow_4\downarrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes\sigma_x\otimes|\downarrow_3\uparrow_4\uparrow_5\rangle\langle\downarrow_3\uparrow_4\uparrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes\sigma_x\otimes|\downarrow_3\uparrow_4\downarrow_5\rangle\langle\downarrow_3\uparrow_4\downarrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes I\otimes|\uparrow_3\downarrow_4\uparrow_5\rangle\langle\uparrow_3\downarrow_4\uparrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes I\otimes|\uparrow_3\downarrow_4\downarrow_5\rangle\langle\uparrow_3\downarrow_4\downarrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes I\otimes|\downarrow_3\downarrow_4\uparrow_5\rangle\langle\downarrow_3\downarrow_4\uparrow_5|\\ & +|\uparrow_1\rangle\langle\uparrow_1|\otimes I\otimes|\downarrow_3\downarrow_4\downarrow_5\rangle\langle\downarrow_3\downarrow_4\downarrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes I\otimes|\uparrow_3\downarrow_4\uparrow_5\rangle\langle\uparrow_3\uparrow_4\downarrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes I\otimes|\uparrow_3\downarrow_4\downarrow_5\rangle\langle\uparrow_3\downarrow_4\downarrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes I\otimes|\downarrow_3\downarrow_4\uparrow_5\rangle\langle\downarrow_3\downarrow_4\uparrow_5|\\ & +|\downarrow_1\rangle\langle\downarrow_1|\otimes I\otimes|\downarrow_3\downarrow_4\downarrow_5\rangle\langle\downarrow_3\downarrow_4\downarrow_5| \end{align}$

— Jitendra
quelle