betaFeedback

WorksPhysics and Astronomy

A general statistical approach to quantum algorithms in a circuit model based on the expectation and standard deviation of each gate separately

Abstract

Recently, there has been a growing literature exploring the generalization of quantum algorithms, such that different quantum algorithms are special examples of a more fundamental structure. In this short paper, we provide a general approach to describe quantum algorithms as a quantum state with amplitudes that are constructed from the expected value and standard deviation of each quantum gate or a sub-sequence of gates in the algorithm. The proposed statistical-based description relies on the celebrated Aharonov-Vaidman identity. We present a more fundamental identity that, unlike the previous one, allows us to switch the basis of the states into a desired form.

Introduction

Quantum algorithms are a cornerstone of quantum computing, exploiting the unique principles of quantum mechanics—such as superposition and entanglement—to solve problems more efficiently than classical algorithms. Famous quantum algorithms, like Shor’s algorithm for factoring and Grover’s search algorithm, offer exponential and quadratic speedups, respectively. As quantum computing evolves, exploring more fundamental structures that unify different quantum algorithms becomes increasingly important, potentially revealing deeper insights into quantum computation. One promising approach is through amplitude estimation algorithms [1, 2, 3], which generalize Grover’s search by efficiently estimating the probability amplitudes of quantum states. This class of algorithms plays a key role in quantum machine learning, optimization, and financial modeling, and it represents a foundational tool in quantum computing by reducing the complexity of tasks such as Monte Carlo simulations. Quantum signal processing (see [4, 5]) is a general framework that is based on applying polynomial functions to eigenvalues of unitary matrices using controlled rotations. Another important framework is the quantum singular value transformation that builds up a unified method in which different quantum algorithms are deduced from a more basic structure [6]. In [7], the authors have shown how such a method embeds major quantum algorithms such as amplitude amplification, quantum linear systems problem, and quantum simulation. Quantum block encoding is a method that allows to represent a given (matrix) operator as a submatrix embedded within a larger unitary matrix, allowing a more flexible way to build up quantum circuits (see, e.g., [8, 9]) and currently, the most efficient way to follow this method is by using the quantum singular value transformation and quantum signal processing (see, [10]). In [11], the authors proposed a unified framework for quantum classification, followed by trainable quantum circuits. In this short paper, we present a general framework for translating the operations on the qubits in a circuit model into quantum states with amplitudes that depend on the expectation and standard deviation of the gates in the algorithm.

Results

Since its appearance about 35 years ago, the Aharonov-Vaidman (AV) identity has been used and well-explored as a fundamental relation in quantum mechanics [12, 13, 14, 15, 16, 17]. This fundamental identity relates the action of an observable operator on a quantum state to its expected value and a projection onto a subspace orthogonal to the state. Mathematically, the AV identity is a universal formula for a Hermitian operator AA acting on a quantum state ∣ψ⟩|\psi\rangle,

A∣ψ⟩=⟨A⟩ψ∣ψ⟩+ΔψA∣ψ⊥⟩(1)A|\psi\rangle= \langle A\rangle_{\psi}|\psi\rangle+ \Delta_{\psi}A|\psi_{\perp}\rangle \tag*{(1)}

where ∣ψ⊥⟩|\psi_{\perp}\rangle is a quantum state that is orthogonal to ∣ψ⟩|\psi\rangle, ⟨A⟩ψ=⟨ψ∣A∣ψ⟩\langle A\rangle^{\psi} = \langle\psi|A|\psi\rangle is the expectation of AA with respect to the state vector, and ΔψA:=⟨ψ∣(A−⟨A⟩)2∣ψ⟩\Delta_{\psi}A := \sqrt{\langle\psi|(A-\langle A\rangle)^2|\psi\rangle} is the associated quantum uncertainty of AA. We see that the operation of AA on the quantum state naturally yields to the statistical measures of expectation and standard deviation, followed by the key role that probability plays in quantum mechanics.

The AV identity (1) has been used for various aspects of quantum mechanics, such as weak measurements and weak values (see, e.g., [15]), and recently, it has been shown that this identity allows one to deduce the Born rule from standard quantum mechanics by postulating that a property of quantum particles remains fixed for a sufficiently short time [17]. The AV identity is at the heart of the proposed results.

We introduce a more general identity that, unlike the AV identity, leads to a quantum state on a desired basis. But, before that, we provide a short introduction to the concept of weak measurements and weak values.

The two-state vector formalism (TSVF) provides a time-symmetric formulation of quantum mechanics, where quantum systems are described by initial and final boundary conditions of the quantum system. It is shown that this framework opens the door for unique measurements that do not collapse the wavefunction, known as weak measurements, where the outcomes of such measurements are known as weak values [18, 19, 20, 21, 22]. Let ∣ψ⟩|\psi\rangle be our initial (pre-selected) state of the quantum system, and let the final (post-selected) state be ⟨ϕ∣\langle\phi|, which is not an orthogonal state to ∣ψ⟩|\psi\rangle. Then, the weak value of an observable AA is defined as

⟨A⟩wϕ,ψ:=⟨ϕ∣A∣ψ⟩⟨ϕ∣ψ⟩,⟨ϕ∣ψ⟩≠0.(2)\langle A\rangle_w^{\phi,\psi}:=\frac{\langle\phi|A|\psi\rangle}{\langle\phi|\psi\rangle},\qquad\langle\phi|\psi\rangle\ne0. \tag*{(2)}

Unlike strong measurements, weak values can take on values outside the range of the eigenvalues of the measured observable AA. Moreover, unlike the standard quantum expectation ⟨A⟩ψ\langle A\rangle^\psi, the weak value ⟨A⟩wϕ,ψ\langle A\rangle_w^{\phi,\psi} is, in general, complex-valued.

In the following, we provide a more fundamental identity for the operation of an operator on the quantum state into a superposition of states in a chosen basis. To obtain the proposed identity, we consider, again, a pair of quantum states ∣ψ⟩|\psi\rangle and ∣ϕ⟩|\phi\rangle that are not orthogonal. We can define an orthogonal state to ∣ϕ⟩|\phi\rangle which we term as ∣ϕ⊥⟩|\phi_\perp\rangle such that

∣ϕ⊥⟩=1CRϕA∣ψ⟩,(3)|\phi_\perp\rangle=\frac{1}{C}R_\phi A|\psi\rangle, \tag*{(3)}

where C≠0C\ne0 is the normalizing constant such that ⟨ϕ⊥∣ϕ⊥⟩=1\langle\phi_\perp|\phi_\perp\rangle=1, and

Rϕ:=qI−K∣ϕ⟩⟨ϕ∣R_\phi:=qI-K|\phi\rangle\langle\phi|

is an orthogonal projection operator, where q∈R>0q\in\mathbb{R}_{>0} is some constant and KK is an operator that satisfies

⟨K⟩ϕ=q.(4)\langle K\rangle^\phi=q. \tag*{(4)}

Follow such a construction, it is then evident that ∣ϕ⟩|\phi\rangle and ∣ϕ⊥⟩|\phi_\perp\rangle are orthogonal to each other

⟨ϕ∣ϕ⊥⟩=1C(q⋅⟨ϕ∣A∣ψ⟩−⟨K⟩ϕ⋅⟨ϕ∣A∣ψ⟩)=0.(5)\langle\phi|\phi_\perp\rangle=\frac{1}{C}\left(q\cdot\langle\phi|A|\psi\rangle-\langle K\rangle^\phi\cdot\langle\phi|A|\psi\rangle\right)=0. \tag*{(5)}

The normalization condition ⟨ϕ⊥∣ϕ⊥⟩=1\langle\phi_\perp|\phi_\perp\rangle=1 leads to the calculation of the normalizing constant, which takes the form

C=qμRϕAψΔψ(RϕA)(6)C=q\mu_{R_\phi A}^{\psi}\Delta_\psi(R_\phi A) \tag*{(6)}

where μRϕAψ:=1+⟨(RϕA)ψ⟩2Δψ(RϕA)2/⟨K⟩ϕ\mu_{R_\phi A}^{\psi}:=\sqrt{1+\frac{\langle(R_\phi A)^\psi\rangle^2}{\Delta_\psi(R_\phi A)^2}/\langle K\rangle^\phi}. We note that in the case q=1q=1 and K=IK=I, when considering that the state ∣ϕ⟩|\phi\rangle is part from a full basis with ∑j=1P∣ϕj⟩⟨ϕj∣=I\sum_{j=1}^{P}|\phi_j\rangle\langle\phi_j|=I such that ∣ϕ⟩|\phi\rangle is taken to be ∣ϕ1⟩|\phi_1\rangle, CC takes the form C=q∑j=2P⟨ψ∣ϕj⟩⟨ϕj∣ψ⟩⋅⟨A†⟩wψ,ϕj⋅⟨A⟩wϕj,ψC=q\sqrt{\sum_{j=2}^{P}\langle\psi|\phi_j\rangle\langle\phi_j|\psi\rangle}\cdot\langle A^\dagger\rangle_{w}^{\psi,\phi_j}\cdot\langle A\rangle_{w}^{\phi_j,\psi}. We further define κ:=⟨ϕ∣ψ⟩/⟨K⟩wϕ,ψ≠0\kappa:=\langle\phi|\psi\rangle/\langle K\rangle_{w}^{\phi,\psi}\ne0.

Substituting (2) and (6) into (3), after some calculations, we obtain the desired extension of the AV identity

A∣ψ⟩=⟨κA⟩wϕ,ψK∣ϕ⟩+μRϕAψΔψ(RϕA)∣ϕ⊥⟩.(7)A|\psi\rangle=\langle\kappa A\rangle_{w}^{\phi,\psi}K|\phi\rangle+\mu_{R_{\phi}A}^{\psi}\Delta_{\psi}(R_{\phi}A)|\phi_{\perp}\rangle. \tag*{(7)}

We see that, unlike the AV identity, we have decomposed the operation A∣ψ⟩A|\psi\rangle into a superposition of an operation KK acting on a state ∣ϕ⟩|\phi\rangle and its orthogonal state. For instance, if we have a system of NN qubits with the initial state ∣0⟩⊗N|0\rangle^{\otimes N} and a gate G0G_0, we can choose a desired basis, e.g.,

∣ϕ⟩=∣+⟩⊗N=12N/2∏i=1N(∣0⟩i+∣1⟩i),(8)|\phi\rangle=|+\rangle^{\otimes N}=\frac{1}{2^{N/2}}\prod_{i=1}^{N}(|0\rangle_i+|1\rangle_i), \tag*{(8)}

and an operation (a gate) acting on it, denoted by G+G_+, to obtain the statistical-based translation mechanism of the quantum algorithm

G0∣0⟩⊗N=⟨κG0⟩w+,0⋅G+∣+⟩⊗N+μRϕG00Δ0(RϕG0)∣+⊥⟩,(9)G_0|0\rangle^{\otimes N}=\langle\kappa G_0\rangle_{w}^{+,0}\cdot G_+|+\rangle^{\otimes N}+\mu_{R_{\phi}G_0}^{0}\Delta_0(R_{\phi}G_0)|+_{\perp}\rangle, \tag*{(9)}

where ∣+⊥⟩|+_{\perp}\rangle is a state that is orthogonal state to ∣+⟩⊗N|+\rangle^{\otimes N}.

The following Figure 1 illustrates the proposed identity compared with the AV identity.

A geometric illustration of the proposed identity compared with the AV identity

Figure 1. A geometric illustration of the proposed identity for K≡IK\equiv I, compared to the AV identity for real-valued (scaled) weak value ⟨κA⟩wϕ,ψ\langle\kappa A\rangle_{w}^{\phi,\psi}.

As illustrated in Figure 1, the operation of AA acting on ∣ψ⟩|\psi\rangle can be geometrically described as a vector within the orthogonal axes constructed from ∣ϕ⟩|\phi\rangle and ∣ϕ⊥⟩|\phi_{\perp}\rangle where ∣ϕ⟩|\phi\rangle can serve as a new basis for the qubits’ state. In the special case where ∣ϕ⟩:=∣ψ⟩|\phi\rangle:= |\psi\rangle, we reconstruct the AV identity with a similar geometry shown in [23].

Every quantum algorithm in the circuit model can be described as a sequence of quantum gates

GM,GM−1,...,G0(10)G_M,G_{M-1},...,G_0 \tag*{(10)}

acting on an initial quantum state ∣0⟩⊗N|0\rangle^{\otimes N}, which we take as the zero state of NN qubits on the computational basis. Then, we can define our quantum algorithm through the process

UG∣0⟩⊗N=GMGM−1...G0∣0⟩⊗N.(11)U_G |0\rangle^{\otimes N}=G_MG_{M-1}...G_0 |0\rangle^{\otimes N}. \tag*{(11)}

The basic observation of the proposed result is that we can operate the AV identity and its extension (7) on the entire sequence of operations GMGM−1...G0G_MG_{M-1}...G_0 or sub-operations

Gj;p:=Gj+pGj+p−1...Gj,j=0,1,...,M,p=0,1,2,...,M−j.(12)\mathcal{G}_{j;p}:=G_{j+p}G_{j+p-1}...G_j,\quad j=0,1,...,M,\quad p=0,1,2,...,M-j. \tag*{(12)}

By applying the AV identity on Gj;p\mathcal{G}_{j;p}, we have

Gj;p∣ψ⟩=⟨Gj;p⟩ψ∣ψ⟩+ΔψGj;p∣ψ⊥⟩,(13)\mathcal{G}_{j;p}|\psi\rangle=\langle\mathcal{G}_{j;p}\rangle^\psi|\psi\rangle+\Delta_\psi\mathcal{G}_{j;p}|\psi_\perp\rangle, \tag*{(13)}

where ∣ψ⊥⟩|\psi_\perp\rangle is an orthogonal state (of the NN qubits) to ∣ψ⟩|\psi\rangle. When applying (13) on the entire sequence, we have a superposition of only two states, the state ∣0⟩⊗N|0\rangle^{\otimes N} and its associated orthogonal state ∣0⊥⟩|0_\perp\rangle, while their amplitudes ⟨Gj;p⟩ψ=0=⟨0∣⊗NGj;p∣0⟩⊗N\langle\mathcal{G}_{j;p}\rangle^{\psi=0}=\langle0|^{\otimes N}\mathcal{G}_{j;p}|0\rangle^{\otimes N} and Δψ=0Gj;p\Delta_{\psi=0}\mathcal{G}_{j;p}, respectively, are based on the statistical measures of the sequence of gates. We can simplify the calculation of the amplitudes by applying (13) on sub-operations as given in (12). Then, the amplitudes of the desired state consist of combinations of expectations and standard deviations of the sub-sequences, where in case of p=0p=0, the amplitudes of the final state are combinations of expectations and standard deviations of each of the operations separately, i.e., the amplitudes are functions of the pairs

{⟨Gj⟩ψ,ΔψGj}j=0M,(14)\left\{\langle G_j\rangle^\psi,\Delta_\psi G_j\right\}_{j=0}^{M}, \tag*{(14)}

for associated states ψ\psi.

Then, the quantum algorithm UG∣0⟩⊗NU_G|0\rangle^{\otimes N} leads to a desired state of the form

UG:∣0⟩⊗N⟼∑x∈{0,1}Ncx({⟨Gj⟩Ψ,ΔΨGj}j=0M)∣x⟩(15)U_G:|0\rangle^{\otimes N}\longmapsto\sum_{x\in\{0,1\}^{N}}c_x\left(\left\{\langle G_j\rangle^\Psi,\Delta_\Psi G_j\right\}_{j=0}^{M}\right)|x\rangle \tag*{(15)}

where ∣x⟩|x\rangle are states in the computational basis of the NN qubits.

As an illustrative example, let us consider the case of a pair of qubits and four quantum gates, where the first gate is the NOT gate acting on the second qubit, G0=X[2]G_0 = X_{[2]}, in which the Deutsch algorithm is a special case. Then, we have

G3G2G1G0∣0⟩⊗2=G3G2G1∣01⟩=⟨G1⟩01⋅G3G2∣01⟩+(Δ01G1)⋅G3G2∣a1⊥b1⊥⟩=⟨G1⟩01⋅(⟨G2⟩01⋅G3∣01⟩+(Δ01G2)G3∣a2⊥b2⊥⟩)+Δ01G1⋅(⟨G2⟩a1⊥b1⊥G3∣a1⊥b1⊥⟩+(Δa1⊥b1⊥G2)G3∣a3⊥b3⊥⟩),(16)\begin{aligned} G_3G_2G_1G_0 |0\rangle^{\otimes2} &= G_3G_2G_1 |01\rangle= \langle G_1\rangle^{01}\cdot G_3G_2 |01\rangle+ (\Delta_{01}G_1)\cdot G_3G_2 |a_{1\perp}b_{1\perp}\rangle\\ &= \langle G_1\rangle^{01}\cdot\left(\langle G_2\rangle^{01}\cdot G_3 |01\rangle+ (\Delta_{01}G_2)G_3 |a_{2\perp}b_{2\perp}\rangle\right) \\ &\quad+ \Delta_{01}G_1\cdot\left(\langle G_2\rangle^{a_{1\perp}b_{1\perp}}G_3 |a_{1\perp}b_{1\perp}\rangle+ (\Delta_{a_{1\perp}b_{1\perp}}G_2)G_3 |a_{3\perp}b_{3\perp}\rangle\right), \tag*{(16)} \end{aligned}

applying, again, the AV identity, and after some algebraic calculations, we conclude that

G3G2G1G0∣0⟩⊗2=c0∣01⟩+∑p=17cp∣ap⊥bp⊥⟩.(17)G_3G_2G_1G_0 |0\rangle^{\otimes2} = c_0 |01\rangle+ \sum_{p=1}^{7} c_p |a_{p\perp}b_{p\perp}\rangle. \tag*{(17)}

Here

c0=⟨G1⟩01⟨G2⟩01⟨G3⟩01,c1=(Δ01G1)⟨G2⟩a1⊥b1⊥⟨G3⟩a1⊥b1⊥,c2=⟨G1⟩01(Δ01G2)⟨G3⟩a2⊥b2⊥,c3=(Δ01G1)(Δa1⊥b1⊥G2)⟨G3⟩a3⊥b3⊥,c4=⟨G1⟩01⟨G2⟩01Δ01G3,c5=⟨G1⟩01(Δ01G2)Δa2⊥b2⊥G3∣a5⊥b5⊥⟩,c6=(Δ01G1)⟨G2⟩a1⊥b1⊥Δa1⊥b1⊥G3,c7=(Δ01G1)(Δa1⊥b1⊥G2)Δa3⊥b3⊥G3.(18)\begin{aligned} c_0 &= \langle G_1\rangle^{01}\langle G_2\rangle^{01}\langle G_3\rangle^{01},\quad c_1 = (\Delta_{01}G_1)\langle G_2\rangle^{a_{1\perp}b_{1\perp}}\langle G_3\rangle^{a_{1\perp}b_{1\perp}},\\ c_2 &= \langle G_1\rangle^{01}(\Delta_{01}G_2)\langle G_3\rangle^{a_{2\perp}b_{2\perp}},\quad c_3 = (\Delta_{01}G_1)(\Delta_{a_{1\perp}b_{1\perp}}G_2)\langle G_3\rangle^{a_{3\perp}b_{3\perp}},\\ c_4 &= \langle G_1\rangle^{01}\langle G_2\rangle^{01}\Delta_{01}G_3,\quad c_5 = \langle G_1\rangle^{01}(\Delta_{01}G_2)\Delta_{a_{2\perp}b_{2\perp}}G_3 |a_{5\perp}b_{5\perp}\rangle,\\ c_6 &= (\Delta_{01}G_1)\langle G_2\rangle^{a_{1\perp}b_{1\perp}}\Delta_{a_{1\perp}b_{1\perp}}G_3,\quad c_7 = (\Delta_{01}G_1)(\Delta_{a_{1\perp}b_{1\perp}}G_2)\Delta_{a_{3\perp}b_{3\perp}}G_3. \tag*{(18)} \end{aligned}

We note that, in general, it is possible that ∣ap⊥bp⊥⟩=∣ap′⊥bp′⊥⟩|a_{p\perp}b_{p\perp}\rangle= |a_{p'\perp}b_{p'\perp}\rangle for p≠p′p \ne p'.

We have thus translated the quantum algorithm into a desired state that is constructed from the expectation and standard deviation of each quantum gate separately.

We can find the orthogonal state ∣ap⊥bp⊥⟩|a_{p\perp}b_{p\perp}\rangle via ∣a(p−1)⊥b(p−1)⊥⟩|a_{(p-1)\perp}b_{(p-1)\perp}\rangle using the formula

∣ap⊥bp⊥⟩=1Cp(I−∣a(p−1)⊥b(p−1)⊥⟩⟨a(p−1)⊥b(p−1)⊥∣)Gj∣a(p−1)⊥b(p−1)⊥⟩.(19)|a_{p\perp}b_{p\perp}\rangle= \frac{1}{C_p}\left(I-|a_{(p-1)\perp}b_{(p-1)\perp}\rangle\langle a_{(p-1)\perp}b_{(p-1)\perp}| \right)G_j|a_{(p-1)\perp}b_{(p-1)\perp}\rangle. \tag*{(19)}

where ∣a(−1)⊥b(−1)⊥⟩:=∣01⟩|a_{(-1)\perp}b_{(-1)\perp}\rangle:= |01\rangle. In the case of Deutsch algorithm, we have a pair of qubits and the quantum gates G0=X[2]G_0 = X_{[2]}, G1=H[2]H[1]G_1 = H_{[2]}H_{[1]}, G2=UfG_2 = U_f, G3=H[1]G_3 = H_{[1]}. By measuring the averages and standard deviations of each of the quantum gates separately, we are able to compute all of the components of the amplitudes cjc_j.

The Grover operator, a key sub-routine in quantum algorithms such as Grover’s quantum search algorithm [24]. This iterative transformation amplifies the probability amplitude of the desired solution in an unstructured search space, providing a quadratic speedup over classical algorithms. The Grover operator takes the form Q=AS0A†SfQ = AS_0A^\dagger S_f where SfS_f is called the phase oracle, S0S_0 is the zero phase-shift (zero reflection) operator, and AA is the state preparation, which in the Grover algorithm consists of the Hadamard gates, A=H⊗NA = H^{\otimes N}.

We can decompose the computation of the Grover operator by applying the AV identity of the quantum algorithm on the sub-routines Gj;p[1]=A†SfG_{j;p}^{[1]} = A^\dagger S_f and Gj;p[2]=AS0G_{j;p}^{[2]} = AS_0. Starting with some quantum state ∣ψ⟩|\psi\rangle of the NN qubits, we can decompose the sub-routine QQ into a computation of each part of the sub-routine into

Gj;p[2]Gj;p[1]∣ψ⟩=c0∣ψ⟩+c1∣ψ1⊥⟩+c2∣ψ2⊥⟩+c3∣ψ3⊥⟩(20)G^{[2]}_{j;p}G^{[1]}_{j;p}|\psi\rangle= c_0|\psi\rangle+ c_1|\psi_{1\perp}\rangle+ c_2|\psi_{2\perp}\rangle+ c_3|\psi_{3\perp}\rangle \tag*{(20)}

where

c0=⟨Gj;p[1]⟩ψ⟨Gj;p[2]⟩ψ,c1=⟨ΔψGj;p[1]⟩ψ⟨Gj;p[2]⟩ψ1⊥(21)c_0 = \frac{\langle G^{[1]}_{j;p}\rangle^\psi}{\langle G^{[2]}_{j;p}\rangle^\psi},\quad c_1 = \frac{\langle\Delta_\psi G^{[1]}_{j;p}\rangle^\psi}{\langle G^{[2]}_{j;p}\rangle^{\psi_{1\perp}}} \tag*{(21)}
c2=⟨Gj;p[1]⟩ψΔψGj;p[2],c3=(ΔψGj;p[1])Δψ1⊥Gj;p[2].c_2 = \langle G^{[1]}_{j;p}\rangle^\psi\Delta_\psi G^{[2]}_{j;p},\quad c_3 = \left(\Delta_\psi G^{[1]}_{j;p}\right)\Delta_{\psi_{1\perp}}G^{[2]}_{j;p}.

Using the proposed identity (7), we can derive a similar decomposition but on a computational basis and with respect to the zero state ∣0⟩⊗N|0\rangle^{\otimes N},

Gj;p[2]Gj;p[1]∣ψ⟩=c1′∣0⟩⊗N+c2′∣0⊥⟩+c3′∣0(2)⊥⟩+c4′∣0(3)⊥⟩,(22)G^{[2]}_{j;p}G^{[1]}_{j;p}|\psi\rangle= c'_1|0\rangle^{\otimes N} + c'_2|0_{\perp}\rangle+ c'_3|0_{(2)\perp}\rangle+ c'_4|0_{(3)\perp}\rangle, \tag*{(22)}

for K=IK = I. By applying the zero state in the computational basis, ∣ϕ⟩=∣0⟩⊗N|\phi\rangle= |0\rangle^{\otimes N}, we get that all other states, ∣0(p)⊥⟩|0_{(p)\perp}\rangle, are also in the same computational basis. So, we needed to use only the proposed identity and then apply standard AV identity.

Quantum phase estimation is a fundamental sub-routine in quantum computing used to estimate the eigenvalue, or phase, associated with an eigenstate of a unitary operator (see, [25, 26, 27]). By preparing a quantum state and applying controlled unitary operations, the algorithm leverages the power of quantum interference to extract phase information with high precision. This process is essential for many quantum algorithms, such as Shor’s algorithm for factoring large numbers, quantum simulations of physical systems, and solving eigenvalue problems. Suppose we have a quantum system of NN qubits starting at the state ∣0⟩⊗N|0\rangle^{\otimes N} and a quantum state ∣v⟩|v\rangle constructed from N′N' qubits that is an eigenstate of a unitary operator UU such that U∣v⟩=eiϕ∣v⟩U|v\rangle= e^{i\phi}|v\rangle for a (real-valued) scalar ϕ\phi. Then, the (sub-routine) quantum algorithm is given by UG∣0⟩⊗N∣v⟩U_G|0\rangle^{\otimes N}|v\rangle where UG=GM...G0U_G = G_M...G_0 where

G0=H⊗N, G1=C1U, G2=C2U2, …, GN=CNU2N−1, GM=N+1=IQFT(23)G_0 = H^{\otimes N},\ G_1 = C_1U,\ G_2 = C_2U^2,\ \ldots,\ G_N = C_NU^{2^{N-1}},\ G_{M=N+1} = IQFT \tag*{(23)}

and so that UG∣0⟩⊗N∣v⟩=GM...G1∣+⟩⊗N∣v⟩U_G|0\rangle^{\otimes N}|v\rangle= G_M...G_1|+\rangle^{\otimes N}|v\rangle for plus states ∣+⟩i=12(∣0⟩i+∣1⟩i)|+\rangle_i = \frac{1}{\sqrt{2}}(|0\rangle_i + |1\rangle_i). We can decompose the process into sub-routines

G1;N−1:=GM−1...G1,(24)\mathcal{G}_{1;N-1} := G_{M-1}...G_1, \tag*{(24)}
GN+1;0=IQFT,\mathcal{G}_{N+1;0} = IQFT,

to obtain the desired statistical-based decomposition

UG∣0⟩⊗N∣v⟩=c0∣Ψ⟩+c1∣Ψ1⊥⟩+c2∣Ψ2⊥⟩+c3∣Ψ3⊥⟩(25)U_G|0\rangle^{\otimes N}|v\rangle= c_0|\Psi\rangle+ c_1|\Psi_{1\perp}\rangle+ c_2|\Psi_{2\perp}\rangle+ c_3|\Psi_{3\perp}\rangle \tag*{(25)}

where ∣Ψ⟩=∣+⟩⊗N∣v⟩|\Psi\rangle= |+\rangle^{\otimes N}|v\rangle, with coefficients that depend on the expectation of standard deviations of the sub-routines

c1=⟨G1;N−1∣+⟩⊗N∣v⟩⟨IQFT∣+⟩⊗N∣v⟩,c1=(Δ∣+⟩⊗N∣v⟩G1;N−1)⟨IQFT∣Ψ⊥⟩,c2=⟨G1;N−1∣+⟩⊗N∣v⟩(Δ∣+⟩⊗N∣v⟩IQFT),c3=(Δ∣+⟩⊗N∣v⟩G1;N−1)(Δ∣Ψ⊥⟩IQFT)(26)\begin{aligned} c_1 &= \langle\mathcal{G}_{1;N-1}|+\rangle^{\otimes N}|v\rangle\langle IQFT|+\rangle^{\otimes N}|v\rangle,\\ c_1 &= \left(\Delta_{|+\rangle^{\otimes N}|v\rangle}\mathcal{G}_{1;N-1}\right)\langle IQFT|\Psi^\perp\rangle,\\ c_2 &= \langle\mathcal{G}_{1;N-1}|+\rangle^{\otimes N}|v\rangle\left(\Delta_{|+\rangle^{\otimes N}|v\rangle}IQFT\right),\\ c_3 &= \left(\Delta_{|+\rangle^{\otimes N}|v\rangle}\mathcal{G}_{1;N-1}\right)\left(\Delta_{|\Psi_\perp\rangle}IQFT\right) \tag*{(26)} \end{aligned}

Using the proposed identity (7) we can convert the base of quantum algorithm into

UG∣0⟩⊗N∣v⟩=⟨κGN+1;0G1;N−1⟩w0,+∣+⟩⊗N∣v⟩K∣0⟩⊗(N+M)+μR0GN+1;0G1;N−1∣+⟩⊗N∣v⟩Δ∣+⟩⊗N∣v⟩(R0GN+1;0G1;N−1)∣0⊥⟩.(27)\begin{aligned} U_G|0\rangle^{\otimes N}|v\rangle &= \langle\kappa\mathcal{G}_{N+1;0}\mathcal{G}_{1;N-1}\rangle_{w}^{0,+}|+\rangle^{\otimes N}|v\rangle K|0\rangle^{\otimes(N+M)}\\ &\quad+\mu_{R_0\mathcal{G}_{N+1;0}\mathcal{G}_{1;N-1}}^{|+\rangle^{\otimes N}|v\rangle}\Delta_{|+\rangle^{\otimes N}|v\rangle}(R_0\mathcal{G}_{N+1;0}\mathcal{G}_{1;N-1})|0_\perp\rangle. \tag*{(27)} \end{aligned}

where κ=⟨0∣⊗(N+M)∣+⟩⊗N∣v⟩/⟨K⟩0\kappa=\langle0|^{\otimes(N+M)}|+\rangle^{\otimes N}|v\rangle/\langle K\rangle^{0}, for a chosen quantum gate KK.

Discussion

We have provided a general framework that allows the translation of quantum algorithms into a quantum state in which the amplitudes are functions of the expectation and standard deviation of the gates in the algorithm. Thus, instead of operating each of the gates in the system, we can start with constructing such quantum states to achieve the desired outcomes by measuring such desired states. We further introduced a more fundamental version of the AV identity that allows the translation of one quantum basis into another, naturally giving rise to weak values. The construction of the desired quantum state for a quantum algorithm with a sequence of gates UG=GMGM−1…G0U_G=G_MG_{M-1}\ldots G_0 can be made by first computing the expectation and standard deviation of each gate in the algorithm (or a sub-sequence of gates) and then construct the states. The best-known upper bound for the number of gates that allows us to construct the desired state is O(2N)O(2^N) for a quantum algorithm that consists of NN qubits (see, [28]). The connection between statistical measures and quantum algorithms is fundamental, following the probabilistic nature of quantum mechanics, followed by the Born rule, and the implicit quantum uncertainties. Thus, the proposed framework, which combines statistical measures of each gate for obtaining a different representation of the quantum algorithm, provides leads to a new understanding of the relations between the quantum algorithms and the statistics of each of the gates in the algorithm, which in our case, only relies on the two statistical measures, the expectation and standard deviation of the gates. For future research, we suggest investigating how the proposed framework can optimize existing quantum algorithms. By focusing on the expectation and standard deviation of the quantum gates, one could develop techniques to minimize gate errors or maximize algorithmic efficiency, leading to more stable quantum computations. Another avenue for future research is to explore whether the proposed representation may improve quantum error correction protocols, followed by the proposed partitions of translations of the quantum gates (see, (12) with (13)). There are, of course, additional aspects of quantum algorithms that can be explored within the proposed general framework.

References

  1. [1]Grinko, D., Gacon, J., Zoufal, C., & Woerner, S. (2021). Iterative quantum amplitude estimation. npj Quantum Information, 7, 52.
  2. [2]Rao, P., Yu, K., Lim, H., Jin, D., & Choi, D. (2020). Quantum amplitude estimation algorithms on IBM quantum devices. Quantum Communications and Quantum Imaging XVIII, 11507, 49-60.arxiv.org/abs/2008.02102
  3. [3]Suzuki, Y., Uno, S., Raymond, R., Tanaka, T., Onodera, T., & Yamamoto, N. (2020). Amplitude estimation without phase estimation. Quantum Information Processing, 19, 1-17.DOI
  4. [4]Low, G. H., Yoder, T. J., & Chuang, I. L. (2016). Methodology of resonant equiangular composite quantum gates. Physical Review X, 6, 041067.DOI
  5. [5]Low, G. H., & Chuang, I. L. (2019). Hamiltonian simulation by qubitization. Quantum, 3, 163.
  6. [6]Gilyén, A., Su, Y., Low, G. H., & Wiebe, N. (2019). Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics. Proceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing, 193-204.DOI
  7. [7]Martyn, J. M., Rossi, Z. M., Tan, A. K., & Chuang, I. L. (2021). Grand unification of quantum algorithms. PRX quantum, 2, 040203.DOI
  8. [8]Rall, P. (2020). Quantum algorithms for estimating physical quantities using block encodings. Physical Review A, 102, 022408.DOI
  9. [9]Wan, L. C., Yu, C. H., Pan, S. J., Qin, S. J., Gao, F., & Wen, Q. Y. (2021). Block-encoding-based quantum algorithm for linear systems with displacement structures. Physical Review A, 104, 062414.arxiv.org/abs/1912.12212
  10. [10]Motlagh, D., & Wiebe, N. (2024). Generalized quantum signal processing. PRX Quantum, 5, 020368.arxiv.org/abs/2308.01501
  11. [11]Nghiem, N. A., Chen, S. Y. C., & Wei, T. C. (2021). Unified framework for quantum classification. Physical Review Research, 3, 033056.DOI
  12. [12]Aharonov, Y., & Vaidman, L. (1990). Properties of a quantum system during the time interval between two measurements. Physical Review A, 41, 11.pubmed.ncbi.nlm.nih.gov/9902834
  13. [13]Goldenberg, L., Vaidman, L. (1996). Applications of a simple quantum mechanical formula. American Journal of Physics, 64, 1059.DOI
  14. [14]Aharonov, Y., & Reznik, B. (2002). How macroscopic properties dictate microscopic probabilities. Physical Review A, 65, 052116.DOI
  15. [15]Cohen, E., & Aharonov, Y. (2017). Quantum to classical transitions via weak measurements and post-selection. Quantum Structural Studies: Classical Emergence from the Quantum Level, 401-425.DOI
  16. [16]Leifer, M. S. (2023). Uncertainty from the Aharonov–Vaidman identity. Quantum Studies: Mathematics and Foundations, 10, 373-397.DOI
  17. [17]Aharonov, Y., & Shushi, T. (2024). Improving the proof of the Born rule using a physical requirement on the dynamics of quantum particles. Quantum Studies: Mathematics and Foundations, 1-4.arxiv.org/abs/2308.10057
  18. [18]Aharonov, Y., Albert, D. Z., & Vaidman, L. (1988). How the result of a measurement of a component of the spin of a spin-1/2 particle can turn out to be 100. Physical Review Letters, 60, 1351.pubmed.ncbi.nlm.nih.gov/10038016
  19. [19]Hallaji, M., Feizpour, A., Dmochowski, G., Sinclair, J., & Steinberg, A. M. (2017). Weak-value amplification of the nonlinear effect of a single photon. Nature Physics, 13, 540-544.DOI
  20. [20]Lundeen, J. S., & Bamber, C. (2012). Procedure for direct measurement of general quantum states using weak measurement. Physical Review Letters, 108, 070402.pubmed.ncbi.nlm.nih.gov/22401180
  21. [21]Pusey, M. F. (2014). Anomalous weak values are proofs of contextuality. Physical Review Letters, 113, 200401.DOI
  22. [22]Aharonov, Y., & Shushi, T. (2022). Complex-valued classical behavior from the correspondence limit of quantum mechanics with two boundary conditions. Foundations of Physics, 52, 56.DOI
  23. [23]Tollaksen, J., & Aharonov, Y. (2009). The deterministic set of operators, quantum interference phenomena, and quantum reality. Journal of Physics: Conference Series, 196, 012006.DOI
  24. [24]Grover, L. K. (1996, July). A fast quantum mechanical algorithm for database search. Proceedings of the twenty-eighth annual ACM symposium on Theory of computing, 212-219.arxiv.org/abs/quant-ph/9605043
  25. [25]Kitaev, A. Y. (1995). Quantum measurements and the Abelian stabilizer problem. arXiv preprint quant-ph/9511026.
  26. [26]Dorner, U., Demkowicz-Dobrzanski, R., Smith, B. J., Lundeen, J. S., Wasilewski, W., Banaszek, K., & Walmsley, I. A. (2009). Optimal quantum phase estimation. Physical Review Letters, 102, 040403.pubmed.ncbi.nlm.nih.gov/19257407
  27. [27]Moore, A. J., Wang, Y., Hu, Z., Kais, S., & Weiner, A. M. (2021). Statistical approach to quantum phase estimation. New Journal of Physics, 23, 113027.DOI
  28. [28]Mottonen, M., Vartiainen, J. J., Bergholm, V., & Salomaa, M. M. (2004). Transformation of quantum states using uniformly controlled rotations. arXiv preprint quant-ph/0407010.arxiv.org/abs/quant-ph/0407010

Paper details

vlnc:W_4zyfxzzt
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents