본문 바로가기

Cryptography/Quantum Information

[양자컴퓨팅 및 정보의 기초] 선형 벡터 공간, 내적, 선형 연산자

https://www.youtube.com/watch?v=K08jdArv0TM

 

 

Linear Vector Spaces

Definition 1

A linear vector space V is a collection of objects $|1\rangle, |2\rangle, ..., |V\rangle, ... |W\rangle, ...$ called vectors, for which the following two operations are well-defined:
-  Vector addition: $|V\rangle + |W\rangle$
-  Multiplication by scalars $a, b, ...$, denoted by $a|V\rangle$
1. Closure: the result of these operation belongs to the space, $|V\rangle + |W\rangle \in V$
2. Scalar multiplication is distributive in both the vectors and scalars:$a(|V\rangle + |W\rangle) = a|V\rangle + a|W\rangle$, $(a + b)|V\rangle = a|V\rangle + b|V\rangle$
3. Scalar multiplication is associative: $a(b|V\rangle) = (ab)|V\rangle$
4. Addition is commutative: $|V\rangle + |W\rangle = |W\rangle + |V\rangle$
5. Addition is associative: $|V\rangle + (|W\rangle + |Z\rangle) = (|V\rangle + |W\rangle) + |Z\rangle$
6. There exists a null vector $|0\rangle$ obeying $|V\rangle + |0\rangle = |V\rangle$
7. For every vector $|V\rangle$, there exists an inverse under addition, $|-V\rangle$ such that $|V\rangle + |-V\rangle = |0\rangle$
8. For every vector $|V\rangle$, $1|V\rangle = |V\rangle$

 

 

Definition 2

The numbers $a,b,...$ are called the field over which the vector space is defined.

 

  • 보통 실수(Real numbers)를 필드로 사용하지만, 양자역학에서는 복소수(Complex numbers)로 확장된다.
  • Field는 스칼라 값의 종류나 행렬/벡터 요소로 허용되는 값의 종류를 의미한다.

 

Definition 3

The set of vectors $(|1\rangle, |2\rangle, ..., |n\rangle)$ is said to be linearly independent if the only set of $a_i$'s satisfying $\displaystyle\sum_{i=1}^n a_i |i\rangle = |0\rangle$ is trivial one with all $a_i = 0$ and none of $|1\rangle, |2\rangle, ..., |n\rangle$ is multiple of $|0\rangle$. If the set of vectors is not linearly independent, we say they are linearly dependent.
  • $a\vec{v} + b\vec{w} = \vec{0}$을 만족하는 유일한 해가 $a=0, b=0$일 때 linearly independent, 아닐 경우 linearly dependent

 

Definition 4

A vector space has dimension n if it can accommodate a maximum of n linearly independent vectors. It will be denoted by $V^n(R)$ if the field is real, and by $V^n(C)$ if the field is complex.
  • 선형 독립이라는 거를 정의를 해야만 이제 벡터 공간의 차원이라는 것을 정의할 수 있게 된다.
  • 벡터 공간의 차원이라고 하는 것은 어떤 벡터 공간이 주어졌을 때 독립이 되는 벡터를 최대한 몇 개까지 찾을 수 있는지를 말하는 것이다.

 

Theorem 1

Any vector $|V\rangle$ in an n-dimensional space can be written as a linear combination of n linearly independent vectors $|1\rangle, |2\rangle, ..., |n\rangle$.
  • n차원 공간의 벡터는 항상 n개의 선형 독립인 벡터의 linear combination으로 나타낼 수 있다.

 

 

Definition 5

A set of n linearly independent vectors in an n-dimensional space is called a basis. We can write $\displaystyle|V\rangle = \sum_{i=1}^n v_i |i\rangle$

 

Definition 6

The coefficients of expansion $v_i$ of a vector in terms of a linearly independent basis ( $|i\rangle$ ) are called the components of the vector in that basis.

 

Theorem 2

The above expansion is unique.

 

  • basis는 선형 독립인 n개의 벡터 집합으로, 임의의 벡터를 표현하는데 사용된다.
  • 직교하는 벡터를 basis로 사용하는 것을 선호하지만, 벡터 공간 정의 자체에는 필수는 아니다.
  • basis가 결정되면, 임의의 벡터는 이 basis들의 linear combination으로 항상 표현할 수 있게 된다.
  • 이러한 linear combination의 앞에 붙는 계수를 components of the vector라고 부른다.
  • 주어진 basis에 대한 component는 unique하게 결정된다.

 

Inner Product Spaces

Properties of inner product for the case of arrow vector (3차원 공간)

1. $\vec{A} \cdot \vec{B} = A_x B_x + A_y B_y + A_z B_z$
2. $\vec{A} \cdot \vec{B} = \vec{B} \cdot \vec{A}$ (symmetry)
3. $\vec{A} \cdot \vec{A} \geq 0$, $0$ iff $\vec{A} = 0$ (positive semidefinite)
4. $\vec{A} \cdot (b\vec{B} + c\vec{C}) = b\vec{A} \cdot \vec{B} + c\vec{A} \cdot \vec{C}$ (linearity)
  • 선형성(Linearity)은 내적의 중요한 속성으로, 두 가지 조건으로 정의된다.
    1. 스칼라 곱에 대해: $a(\vec{x} \cdot \vec{y}) = (a\vec{x}) \cdot \vec{y}$.
    2. 덧셈에 대해: $(\vec{x}_1 + \vec{x}_2) \cdot \vec{y} = (\vec{x}_1 \cdot \vec{y}) + (\vec{x}_2 \cdot \vec{y})$.

 

Generalized requirement for inner product

1. The result is a number (generally a complex)
2. $\langle V|W \rangle = \langle W|V \rangle^*$ (skew-symmetry)
3. $\langle V|V \rangle \geq 0$, $0$ iff $|V\rangle = |0\rangle$ (positive semidefinite)
4. $\langle V|(a|W\rangle + b|Z\rangle) = a\langle V|W \rangle + b\langle V|Z \rangle$ (linearity in ket)
    • 순서를 바꾸려면 결과에 Complex Conjugation이 취해져야한다.
      • $\langle w \vert v \rangle = (\langle v \vert w \rangle)^*$
      • 이는 복소수 필드에서 자기 자신과의 내적 시 결과가 음수가 되는 모순을 피하기 위해 필요하다.
    • 캣 벡터에 대해서는 선형성이 적용된다 (브라 벡터에 대해서는 반선형성).

 

Definition 7

A vector space with an inner product is called an inner product space.

 

 

Anti-linearity of the Bra

  • 브라 벡터에 대한 연산은 캣 벡터에 대한 선형성과 달리 반선형성(Anti-linearity)을 가진다.
  • 쉽게 말하자면, Ket($| \rangle$) 안의 스칼라는 그대로 나오지만, Bra($\langle |$) 안의 스칼라는 밖으로 나올 때 켤레 복소수($*$)가 되어 나온다는 것이다.
  • 캣 벡터 $\vert v \rangle$를 브라 벡터 $\langle v \vert$로 변환할 때, 계수(스칼라)에도 켤레 복소수 변환이 반드시 적용되어야 한다.

 

 

Definition 8

Two vectors are orthogonal or perpendicular if their inner product vanishes.

 

Definition 9

$\sqrt{\langle V|V \rangle} = |V|$ will be referred as the norm or length of the vector

 

Definition 10

A set of basis vectors, all of which are pairwise orthogonal and have unit norms, will be called an orthonormal basis.

 

  • 내적을 정의 한 후에야 다음 개념들을 정의할 수 있는 것이다.
    • Orthogonal : 두 벡터의 내적이 0일 때
    • Norm : 내적을 취한 후 제곱근을 취한 값.
  • Orthonormal basis : 직교하고 정규화된(길이가 1인) basis

 

 

Theorem 3 (Gram-Schmidt)

For any lineary independent basis, we can always find and orthonormal basis by combining these basis vectors.
  • 꼭 직교하지 않는다 하더라도 linearly independent한 어떤 basis가 주어지면 그걸 가지고 항상 서로 직교하고 정규화된 basis를 찾을 수 있다는 것이다.
If $|1\rangle, |2\rangle, ..., |n\rangle$ are orthonormal basis:
$$\langle i|j\rangle = \begin{cases} 1 & \text{for } i = j \\ 0 & \text{for } i \neq j \end{cases} \equiv \delta_{ij} \text{ (Kronecker delta)}$$
  • Orthonormal basis $\vert i \rangle$를 사용할 경우, 크로네커 델타($\delta_{ij}$) 형태로 내적이 표현되어 계산이 유리하다.
    • $\langle i \vert j \rangle = \delta_{ij}$.

 

We know that the components of a vector are uniquely determined for a given basis, but then how to find components $v_i$ of a given vector $|V\rangle$ for orthonormal basis $|1\rangle, |2\rangle, ..., |n\rangle$?

 

우리가 어떤 벡터 $|V\rangle$를 basis 벡터들($|1\rangle, |2\rangle, ...$)의 합으로 표현하면 아래와 같은 모양이 된다.

 

$|V\rangle = v_1|1\rangle + v_2|2\rangle + \dots + v_n|n\rangle$

 

여기서 $v_1, v_2$ 같은 숫자들을 components라고 한다. 이러한 $v_i$ 값들을 알아내려면 어떤 계산식을 써야 할까?

 

바로 내적을 이용해서 구할 수 있다. basis들이 서로 수직이로 길이가 1인 Orthonormal한 상태이기 때문에 계산이 아주 간단해진다. 구하고 싶은 component $v_i$는 해당 basis $|i\rangle$와 원래 벡터 $|V\rangle$를 내적하면 나온다.

 

$v_i = \langle i | V \rangle$

 

원래 식의 양변에 $\langle i |$를 내적해 보면,

$\langle i | V \rangle = v_1\underbrace{\langle i | 1 \rangle}_{0} + \dots + v_i\underbrace{\langle i | i \rangle}_{1} + \dots$

 

다른 basis들과의 내적은 모두 0이 되고 자기 자신과의 내적은 1이 되므로(Kronecker delta 성질), 결국 $v_i$만 남게 되기 때문이다.

 

결론적으로, Orthonormal basis를 선호하는 이유는 임의의 벡터의 component를 얻기 매우 쉽기 때문이다.

 

If two vectors $|V\rangle, |W\rangle$ are expanded in terms of orthonormal basis $|1\rangle, |2\rangle, ..., |n\rangle$,
$|V\rangle = \sum_i v_i |i\rangle$
$|W\rangle = \sum_j w_j |j\rangle$
$\langle V | W \rangle = \sum_i \sum_j v_i^* w_j \langle i | j \rangle = \sum_i v_i^* w_i$
  • 또한 이렇게 orthonormal basis로 전개된 두 벡터 $\vert V \rangle, \vert  W \rangle$의 내적을 간단하게 계산할 수 있다.
    • 주의할 점은 캣 벡터를 브라 벡터를 바꿀 때 계수에 Complex conjugation을 적용해야한다는 것이다.
  • 두 벡터를 내적($\langle V | W \rangle$)하면 이중 시그마($\sum_i \sum_j$)가 나오지만, basis들이 Orthonormal하기 때문에 $\langle i | j \rangle$$i=j$일 때만 1이 되고 나머지는 0이 되어 사라진다.
  • 결과적으로 같은 인덱스의 성분끼리 곱한 것들의 합($\sum_i v_i^* w_i$)만 남게 된다.

 

같은 벡터를 나타내는 열벡터 표현은 basis가 달라지면 unique하지 않다. 만약 basis가 고정이 되었다면 그때는 값이 unique하게 결정이 되는 것이다. 항상 일정한 것은 벡터 그 자체. 벡터는 일정하지만 그것을 행렬 형태, 열벡터 형태로 나타낼 때 벡터를 나타내는 표현이 바뀌는 것이다. 
이것이 중요한 이유는, 양자 계산에서 많이 하는 것이 바로 basis 변환이다. basis가 변화하면 같은 양자 상태가 완전히 다르게 나타나게 되기 때문에 이러한 개념을 명확하게 알고 있는 것이 중요하다.

 

 

Vectors $|V\rangle, |W\rangle$ are uniquely specified by their components in a given basis $\rightarrow$
Can be written as column vectors:
$|V\rangle \rightarrow \begin{bmatrix} v_1 \\ \vdots \\ v_n \end{bmatrix}$ in this basis
$|W\rangle \rightarrow \begin{bmatrix} w_1 \\ \vdots \\ w_n \end{bmatrix}$ in this basis
$\langle V|W\rangle = \sum_i v_i^* w_i = \begin{bmatrix} v_1^* & \dots & v_n^* \end{bmatrix} \begin{bmatrix} w_1 \\ \vdots \\ w_n \end{bmatrix}$
Bra vector can be represented as a row vector with complex conjugation (transpose conjugation or adjoint operation).
$\langle V| = \begin{bmatrix} v_1^* & \dots & v_n^* \end{bmatrix}$
  • Basis가 결정되면 임의의 벡터 $\vert v \rangle$는 열벡터로 나타낼 수 있다.
  • 두 벡터 $\vert v \rangle, \vert w \rangle$의 내적은 행렬 연산과 유사하게 표현될 수 있다.
  • 한 연산이 간단하게 가능한 것은 두 벡터가 같은 basis에 대해 표현되었을 때만 해당된다.

 

 

Theorem 4 : The Schwarz inequality

$|\langle V|W\rangle| \leq |V||W|$

 

Theorem 5 : The triangular inequality

$|V + W| \leq |V| + |W|$

 

 

Subspace

Definition 11

Given a vector space $\mathcal{V}$, a subset of its elements that form a vector space among themselves is called a subspace. A particular subspace $i$ of dimensionality $n_i$ will be denoted by $\mathcal{V}_i^{n_i}$.

Example: orthogonal subspace with respective some vector $|W\rangle : \mathcal{V}_{\perp W}^{n-1}$
  • 예를 들어서, 3차원 공간에서 Z축 성분을 0으로 설정하여 얻은 XY 평면은 3차원 공간의 2차원 부분 공간이라고 할 수 있다.
  • 양자 상태를 다룰 때, 더 높은 차원의 공간 내에서 특정 저차원 평면(부분 공간)에 국한하여 상태를 분석할 수 있게 된다.

 

 

Linear Operator

An operator $\Omega$ is an instruction for transforming any given vector $|V\rangle$ into another vector $|V'\rangle$ and this relation is written as $|V'\rangle = \Omega|V\rangle$.

In this class, we will consider only the linear operators $\Omega$ that do not take us out of the vector space. $\Leftrightarrow$ If $|V\rangle \in \mathcal{V}$, $\Omega|V\rangle \in \mathcal{V}$.

Linear operator acting on bra is written as $\langle V|\Omega = \langle V'|$

Linear operator should obey the following rules:
-  $\Omega\alpha|V_i\rangle = \alpha\Omega|V_i\rangle$
-  $\Omega(\alpha|V_i\rangle + \beta|V_j\rangle) = \alpha\Omega|V_i\rangle + \beta\Omega|V_j\rangle$
-  Same for the bra vectors
  • Operator는 정의역(vector space)의 원소를 치역(다른 vector space)의 원소로 매핑하는 함수와 같은 역할을 한다.
  • 양자역학에서 다루는 linear operator는 보통 같은 공간 내에서 매핑을 수행한다.
  • 캣 벡터는 연산자를 왼쪽에, 브라 벡터는 오른쪽에 표기한다.
  • Linearity(선형성)는 Operator가 만족해야 하는 핵심 성질이다.

 

Once the action of the linear operator $\Omega$ for all the basis vectors $|1\rangle, |2\rangle, ..., |n\rangle$ is known, its action on any arbitrary vector is determined.
-  When $\Omega|i\rangle = |i'\rangle$ is known for all $i = 1 ... n$, and an arbitrary vector $|V\rangle = \sum_i v_i|i\rangle$ is given,
-  $\Omega|V\rangle = \sum_i \Omega v_i|i\rangle = \sum_i v_i \Omega|i\rangle = \sum_i v_i |i'\rangle$
  • Linear operator에 관심을 가지는 이유는, Linear라는 조건이 주어지면 예측할 수 있는 것이 많아지기 때문이다.
  • Linear라는 조건을 강요하는 순간, 즉 vector space에서 임의의 벡터를 뽑아서 그것이 뭐에 맵핑되는지만 결정을 해놓으면, 나머지 임의의 벡터들을 아무데나 매핑이 되는 것이 아니라, 이미 그 basis가 매핑되는 관계 때문에 맵핑될 수 있는 것이 무조건적으로 결정된다는 것이다.
  • 이렇게 Linear operator는 basis 벡터들이 어디로 매핑되는지만 결정되면, 임의의 벡터가 어디로 매핑될 지가 결정적으로 결정된다는 가장 큰 특징을 가진다.

 

Product of two operators

$\Lambda\Omega|V\rangle \equiv \Lambda(\Omega|V\rangle) = \Lambda|\Omega V\rangle \rightarrow$ we will use $|\Omega V\rangle$ notation to represent $\Omega|V\rangle$
    • 연산자의 곱은 연산을 순차적으로 적용하는 것으로 정의된다 (예: $\Omega \Lambda$는 $\Lambda$를 먼저 적용하고 $\Omega$를 적용).
    • 연산자는 행렬과 동일하게 이해할 수 있으며, 행렬을 이해하면 선형 연산자의 모든 동작을 알 수 있다.

 

 

Commutator

$[\Omega, \Lambda] = \Omega\Lambda - \Lambda\Omega$.
  • The order of the operators in a product is very important, and generally $[\Omega, \Lambda] \neq 0$
$[\Omega, \Lambda\theta] = \Lambda[\Omega, \theta] + [\Omega, \Lambda]\theta$
$[\Lambda\Omega, \theta] = \Lambda[\Omega, \theta] + [\Lambda, \theta]\Omega$