Bagaimana cara operator difusi Grover bekerja dan mengapa itu optimal?


15

Dalam jawaban ini , algoritma Grover dijelaskan. Penjelasan menunjukkan bahwa algoritma ini sangat bergantung pada Operator Difusi Grover , tetapi tidak memberikan rincian tentang cara kerja operator ini.

Secara singkat, Operator Difusi Grover menciptakan 'inversi tentang rata-rata' untuk secara iteratif membuat perbedaan kecil dalam langkah-langkah sebelumnya cukup besar untuk dapat diukur.

Pertanyaannya sekarang:

  1. Bagaimana operator difusi Grover mencapai ini?
  2. Mengapa O yang dihasilkan ( O(n)total waktu untuk mencari database unordered yang optimal?

1
Hanya komentar pada pertanyaan kedua. Ada beberapa karya yang menunjukkan bahwa jejak status dalam algoritma Grover mengikuti persis geodesik yang menghubungkan kondisi awal algoritma dan status tujuan. Jadi itu optimal.
XXDD

Jawaban:


5

Karena pertanyaan aslinya adalah tentang deskripsi orang awam, saya menawarkan solusi yang sedikit berbeda yang mungkin lebih mudah dipahami (tergantung latar belakang), berdasarkan evolusi waktu yang berkesinambungan. (Namun saya tidak berpura-pura bahwa itu cocok untuk orang awam.)

Kita mulai dari keadaan awal yang merupakan superposisi seragam semua negara, dan kami bertujuan untuk menemukan sebuah negara| xyang dapat diakui sebagai jawaban yang benar (dengan asumsi ada tepat satu negara tersebut, meskipun hal ini dapat digeneralisasi). Untuk melakukan ini, kami berkembang dalam waktu di bawah tindakan Hamiltonian H=| xx| +| ψψ| . Fitur yang sangat indah dari pencarian Grover adalah bahwa pada titik ini, kita dapat mengurangi matematika menjadi subruang hanya dari dua negara

|ψ=12ny{0,1}n|y
|x
H=|xx|+|ψψ|.
, daripada membutuhkan semua 2 n . Lebih mudah dijelaskan jika kita membuat basis ortonormal dari kondisi ini, { | x , | ψ } mana | ψ = 1{|x,|ψ}2n{|x,|ψ} Menggunakan dasar ini, evolusi waktue-iHt| ψdapat ditulis sebagai e-it(I+2-nZ+
|ψ=12n1y{0,1}n:yx|y.
eiHt|ψ manaXdanZadalah matriks Pauli standar. Hal ini dapat ditulis kembali sebagai e-it(Icos(t
eit(I+2nZ+2n12nX)(12n112n),
XZ Jadi, jika kita berevolusi untuk sementara waktut=π
eit(Icos(t2n/2)i12n/2sin(t2n/2)(Z+X2n1))(12n112n).
, dan mengabaikan fase global, kondisi terakhir adalah 1t=π22n/2
12n/2(Z+X2n1)(12n112n)=(12n2n12n)+(112n2n12n)=(10).
|x

H~=5H, and evolve using H~ and the evolution time would be 5 times shorter. If you wanted to be really radical, replace the 5 with 2n/2, and Grover's search runs in constant time! But you're not allowed to do this arbitrarily. Any given experiment would have a fixed maximum coupling strength (i.e. a fixed multiplier). So, different experiments have different running times, but their scaling is the same, 2n/2. It's just like saying that the gate cost in the circuit model is constant, rather than assuming that if we use a circuit of depth k each gate can be made to run in time 1/k.

The optimality proof essentially involves showing that if you made detection of one possible marked state |x any quicker, it would make detection of a different marked state, |y, slower. Since the algorithm should work equally well whichever state is marked, this solution is the best one.


4

One way of defining the diffusion operator is1 D=HnU0Hn, where U0 is the phase oracle

U0|0n=|0n,U0|x=|xfor|x|0n.

This shows that U0 can also be written as U0=I2|0n0n|, giving

D=2|++|I,
where |+=2n/2(|0+|1)n.

Writing a state |ψ=α|++β|+ where |+ is orthogonal to |+ (i.e. ++=0) gives that D|ψ=α|+β|+.

This gives2 that the diffusion operator is a reflection about |+

As the other part of Grover's algorithm is also a reflection, these combine to rotate the current state closer to the 'searched-for' value x0. This angle decreases linearly with the number of rotations (until it overshoots the searched-for value), giving that the probability of correctly measuring the correct value increases quadratically.

Bennet et. al. showed that this is optimal. By taking a classical solution to an NP-problem, Grover's algorithm can be used to quadratically speed this up. However, taking a language LA={y:xA(x)=y} for a length preserving function A (here, an oracle), any bounded-error oracle based quantum turing machine cannot accept this language in a time T(n)=o(2n/2).

This is achieved by taking a set of oracles where |1n has no inverse (so is not contained in the language). However, this is contained in some new language LAy by definition. The difference in probabilities of a machine accepting LA and a different machine accepting LAy in time T(n) is then less than 1/3 and so neither language is accepted and Grover's algorithm is indeed asymptotically optimal.3

Zalka later showed that Grover's algorithm is exactly optimal.


1 In Grover's algorithm, minus signs can be moved round, so where the minus sign is, is somewhat arbitrary and doesn't necessarily have to be in the definition of the diffusion operator

2 alternatively, defining the diffusion operator without the minus sign gives a reflection about |+

3 Defining the machine using the oracle A as MA and the machine using oracle Ay as MAy, this is a due to the fact that there is a set S of bit strings, where the states of MA and MAy at a time t are ϵ-close4, with a cardinality <2T2/ϵ2. Each oracle where MA correctly decides if |1n is in LA can be mapped to 2nCard(S) oracles where MA fails to correctly decide if |1n is in that oracle's language. However, it must give one of the other 2n1 potential answers and so if T(n)=o(2n/2), the machine is unable to determine membership of LA.

4 Using the Euclidean distance, twice the trace distance

Dengan menggunakan situs kami, Anda mengakui telah membaca dan memahami Kebijakan Cookie dan Kebijakan Privasi kami.
Licensed under cc by-sa 3.0 with attribution required.