Review of Linear Algebra II
-
Upload
khangminh22 -
Category
Documents
-
view
0 -
download
0
Transcript of Review of Linear Algebra II
18-819F: Introduction to Quantum Computing 47-779/47-785: Quantum Integer Programming
& Quantum Machine Learning
Review of Linear Algebra II
Lecture 02
2021.09.07.
โข Dirac notation for complex linear algebra
โ State vectors
โ Linear operators โ matrices
โ Hermitian operators
โ Unitary operators
โ Projection operators and measurement
โข Tensors
โ Basis vectors for tensor product space
โ Composite quantum systems
โ Conception of tensor products as quantum entanglement
2
Agenda
Dirac Notation for Linear Algebra โข We previously defined the inner (dot) product of two vectors ๐ข and ๐ฃ as
๐ข. ๐ฃ = (๐ข, ๐ฃ),
where it was understood that ๐ข is a row vector, which is better written as the transpose, ๐ข๐.
โข This bracket notation was specialized by P.A.M Dirac, using angle brackets, to
๐ข. ๐ฃ = ๐ข๐ . ๐ฃ =< ๐ข| ๐ฃ > = < ๐ข ๐ฃ >;
โข An ordinary row vector is now defined as: < ๐ข| = ๐ข1 ๐ข2 โฆ ๐ข๐ , where < | is the
โbraโ and an ordinary column vector is |๐ฃ > =
๐ฃ1๐ฃ2โฎ๐ฃ๐
, where | > is the โketโ so that the
inner (dot) product is formed by the โbra-ketโ < ๐ข|๐ฃ >;
โข The โbraโ implicitly tells us to take the complex conjugate and the transpose of ๐ข.
3
Dirac Notation for Linear Algebra
โข Having defined the notation, one can now use it to write most of ordinary vector and
matrix algebra;
โข The outer product of two vectors, which we previously wrote as
๐ข๐ฃ๐ =
๐ข1๐ข2โฎ๐ข๐
๐ฃ1 ๐ฃ2 โฆ ๐ฃ๐ =
๐ข1๐ฃ1 โฆ ๐ข1๐ฃ๐โฎ โฆ โฎ
๐ข๐๐ฃ1 โฆ ๐ข๐๐ฃ๐Eqn. (3.1), is a matrix that can be
written succinctly as ๐ข๐ฃ๐ = |๐ข >< ๐ฃ| (we prove this later) Eqn. (3.2);
โข The outer product is often called a dyad and is a linear operator, i.e., a matrix.
4
โข In 3D linear algebra, we said we can write a vector เดฑ๐ฃ as a linear combination of the basis
vectors ฦธ๐1, ฦธ๐2, ฦธ๐3, where ๐ฃ1, ๐ฃ2, ๐ฃ3 are the scaling coefficients
แ
เดฑ๐ฃ = ๐ฃ1 ฦธ๐1 + ๐ฃ2 ฦธ๐2 + ๐ฃ3 ฦธ๐3เดฑ๐ฃ = ฦธ๐1 ฦธ๐1
๐ . เดฑ๐ฃ + ฦธ๐2 ๐2๐ . เดฑ๐ฃ
เดฑ๐ฃ = ฦธ๐1 ฦธ๐1๐ + ฦธ๐2 ฦธ๐2
๐ + ฦธ๐3 ฦธ๐3๐ เดฑ๐ฃ
+ ฦธ๐3 ๐3๐ . เดฑ๐ฃ
โข From the 2nd equation above we have: ๐ฃ1 = ฦธ๐1๐ . เดฑ๐ฃ , ๐ฃ2 = ฦธ๐2
๐ . เดฑ๐ฃ, ๐ฃ3 = ๐3๐ . เดฑ๐ฃ;
โข From the 3rd equation above we have 1 = ฦธ๐1 ฦธ๐1๐ + ฦธ๐2 ฦธ๐2
๐ + ฦธ๐3 ฦธ๐3๐ = ฯ๐=1
3 ฦธ๐๐ ฦธ๐๐๐;
โข So ฯ๐=13 ๐๐๐๐
๐ = 1 is the closure (completeness) relationship for the 3D vector space.
5
Dyad in 3D Vector Space
โข In Hilbert space, a state vector |๐ฃ > can be written as a linear combination of the basis
vectors |1 >, |2 >,โฆ |๐ > with complex coefficients ๐1, ๐2, โฆ ๐๐
แ
|๐ฃ > = ๐1 1 > +๐2 2 > +โฏ+ ๐๐|๐ >
|๐ฃ > = |1 > < 1|๐ฃ > + |2 > < 2 ๐ฃ > +โฏ+ |๐ > (< ๐|๐ฃ >
|๐ฃ > = 1 >< 1 + 2 >< 2 +โฏ ๐ >< ๐ |๐ฃ >
โข As in ordinary vector space, the coefficients ๐๐ are โhow muchโ of each basis vector is
in the state vector |๐ฃ >, which is given by the inner product (or the projection of the
state vector onto each basis vector);
โข The left-hand side of the last equation is equal to the right-hand side if
1 = 1 >< 1 + 2 >< 2 +โฏ+ ๐ >< ๐ =
๐=1
๐
|๐ >< ๐|
6
Relationship of Dyad Operator to ๐
โข We know that the closure relationship for a basis set of vectors |๐ > for a vector space is
๐
๐ >< ๐ = 1
โข Any operator แ๐ด in this vector space can therefore be written as
แ๐ด = 1 แ๐ด1 =
๐
๐
|๐ >< ๐ แ๐ด ๐ >< ๐| =
๐๐
< ๐ แ๐ด ๐ > |๐ >< ๐| =
๐๐
๐ด๐๐|๐ >< ๐|
โข Note that < ๐| แ๐ด|๐ = ๐ด๐๐;
โข Suppose the basis set is the standard basis: |1 > =
10โฎ0
, |2 > =
01โฎ0
|3 > =
001โฎ
โฆ |๐ > =
00โฎ1
;
โข It follows then that แ๐ด = ฯ๐๐ ๐ด๐๐ ๐ >< ๐ =
๐ด11 ๐ด12 โฆ
๐ด21 ๐ด22โฎ โฑ
is a matrix (QED).
7
Operators as Matrices
โข Every quantum system has a Hamiltonian operator, ๐ป, representing the total system energy.
โข Suppose the state space of the system is spanned by the basis set |๐ข๐ >, then
๐ป ๐ข๐ >= ๐๐ ๐ข๐ >, with ๐๐ being the eigenvalues;
โข We can therefore write
โข
๐ป = 1๐ป1 = ฯ๐ |๐ข๐ >< ๐ข๐| ๐ป ฯ๐ |๐ข๐ >< ๐ข๐|
= ฯ๐๐ ๐ข๐ >< ๐ข๐ ๐ข๐ > ๐๐ < ๐ข๐|
= ฯ๐๐ |๐ข๐ > ๐ฟ๐๐๐๐ < ๐ข๐|
โข The last equation tells us that ๐ป = ฯ๐ ๐๐|๐ข๐ >< ๐ข๐|, which is a matrix as we saw previously.
โข Hence ๐ป =๐1 0 โฆ
0 ๐2โฎ โฑ
; apparently an operator is diagonal in its own eigen basis.
8
Example of a Special Operator: the Hamiltonian
โข Given the state vector |๐ > = ฯ๐ ๐๐|๐๐ >, where ๐๐ are basis vectors, one can use the
inner product to determine the length of the state vector state|๐ > , thus
< ๐|๐ > = ฯ๐ฯ๐ ๐๐โ ๐๐ < ๐๐|๐๐ >= ฯ๐ ๐๐
2;
โข The length of the state vector is therefore ๐ = < ๐|๐ >;
โข For two state vectors |๐ > = ฯ๐ ๐๐|๐ข๐ > and |๐ > = ฯ๐ ๐๐|๐ฃ๐ >, the inner product
is
< ๐|๐ > =
๐
๐
๐๐โ ๐๐ < ๐ข๐|๐ฃ๐ > =
๐
๐๐โ๐๐
9
Inner Product and the Length of a Vector
Linear operators
โข A linear operator ๐ช can transform vectors (states) in the following manner
๐ช ๐ผ ๐ > +๐ฝ ๐ > = ๐ผ ๐ช ๐ > +๐ฝ ๐ช ๐ > Eqn. (3.3);
โข The expression above is simply following ordinary rules of linear algebra (for addition,
multiplication, and distributivity);
โข Note that |๐ > and |๐ > are โstateโ vectors and ๐ผ and ๐ฝ are complex (scalars);
โข Remember that the operator ๐ช is simply a matrix that transforms one vector to another as
we discussed before.
10
Average (Expectation) Value
โข Given an operator ๐ด and the state ๐ that is a linear combination of a set of basis vectors,
๐ข๐, we can write
|๐ > =
๐1๐2โฎ๐๐
โน |๐ >= ๐1|๐ข1 > +๐2|๐ข2 > +โฏ+ ๐๐|๐ข๐ > Eqn. (3.4);
โข We can compute the average (expectation) result of operating on the state as
๐ด = ๐ ๐ด ๐ = ฯ๐ฯ๐ ๐๐โ๐๐ < ๐ข๐ ๐ด ๐ข๐ >;
โข Define < ๐ข๐ แ๐ด ๐ข๐ >โก ๐๐๐ as matrix elements; since we must have ๐ = ๐
แ๐ด = ๐ ๐ด ๐ =
๐
๐๐โ๐๐๐๐๐ =
๐
๐๐ ๐๐2 =
๐
๐๐๐๐ Eqn. (3.5).
11
Matrix Elementsโข For the state vector |๐ >, which we said could be written as a linear combination of
basis vectors in a Hilbert space,
|๐ >= ๐1|๐ข1 > +๐2|๐ข2 > +โฏ+ ๐๐|๐ข๐> Eqn. (3.6),
โข We can perform the operation แ๐ด|๐ >, to get another state vector; the inner product of
this resulting vector with another basis vector |๐ข๐ > is given by
< ๐ข๐| แ๐ด| ฯ๐ ๐๐|๐ข๐ >= ฯ๐ < ๐ข๐ แ๐ด ๐ข๐ > ๐๐ Eqn. (3.7);
โข The expression< ๐ข๐ แ๐ด ๐ข๐ >= ๐๐๐is a number called the matrix element that expresses
the coupling of state |๐ข๐ > to state |๐ข๐ >; (see previous slide on expectation);
โข When ๐ = ๐, then < ๐ข๐ แ๐ด ๐ข๐ >= แ๐ด is just the expectation value of the operator แ๐ด.
12
Hermitian Operators โข A Hermitian conjugate of an operator ๐ด is obtained by complex conjugating and then
taking the transpose of the operator
๐ดโ = ๐ดโ ๐;
โข Some properties of Hermitian conjugates
๐ด๐ต โ = ๐ตโ ๐ดโ
|๐ > โ = < ๐|
๐ด|๐ > โ = < ๐|๐ดโ
๐ด๐ต|๐ > โ = < ๐|๐ตโ ๐ดโ
๐ผ๐ด โ = ๐ผโ๐ดโ
โข An operator is Hermitian if ๐ด = ๐ดโ ; this is the complex analog of the symmetric matrix;
โข The eigenvalues of a Hermitian operator are always real.
13
Unitary Operators
โข Recall that we defined an inverse of an operator (matrix) ๐ด through ๐ด๐ดโ1 = ๐ผ, where ๐ผis the identity matrix;
โข An operator ๐ด is unitary if its adjoint (transpose) is equal to its inverse: ๐ดโ = ๐ดโ1;
โข Unitary operators are usually denoted by the symbol ๐; from the definition above we
get
๐๐โ = ๐โ ๐ = ๐ผ Eqn. (3.8);
โข An operator ๐ด is normal if ๐ด๐ดโ = ๐ดโ ๐ด Eqn. (3.9);
โข Hermitian and unitary operators are normal.
14
Trace of an Operator
โข Given an operator ๐ด as the matrix ๐ด =๐11 ๐12๐21 ๐22
,
we define the trace as the sum of the diagonal elements, ๐๐ ๐ด = ๐11 + ๐22;
โข For a general operator ๐ด and a basis vectors ๐ข๐, we can write
๐๐ ๐ด = ฯ๐ < ๐ข๐ ๐ด ๐ข๐ > = ฯ๐ ๐๐๐ Eqn. (3.10);
Some properties of the trace:
โข Trace is cyclic, i.e., ๐๐ ๐ด๐ต๐ถ = ๐๐ ๐ถ๐ด๐ต = ๐๐ ๐ต๐ถ๐ด ;
โข Trace is independent of basis vectors, ๐๐ ๐ด = ฯ๐ < ๐ข๐ ๐ด ๐ข๐ >= ฯ๐ < ๐ฃ๐ ๐ด ๐ฃ๐ >;
โข If operator ๐ด has eigenvalues ๐๐ , then ๐๐ ๐ด = ฯ๐ ๐๐ ;
โข Trace is linear, meaning that ๐๐ ๐ผ๐ด = ๐ผ๐๐ ๐ด ; and ๐๐ ๐ด + ๐ต = ๐๐ ๐ด + ๐๐ ๐ต ;
15
Projection Operators
โข We defined earlier that the outer product of two vectors, |๐ > and |๐ > , is an operator,
๐ด (matrix),
๐ >< ๐ = ๐ด Eqn. (3.11);
โข When the outer product is between a normalized vector with itself, for example, |๐ >and |๐ >, we get a special operator called the projection operator,
๐ = |๐ >< ๐| Eqn. (3.12);
โข The projection operator in the example above projects any vector it operates on to the
direction of |๐ >;
16
Properties of Projection Operators
โข For any normalized vector, |๐ข1 >, we have < ๐ข1|๐ข1 >= 1 โน ๐ = |๐ข1 >< ๐ข1|;
โข ๐โ = ๐ข1 >< ๐ข1 = ๐;
โข ๐ is Hermitian since ๐|๐ข1 >= ๐ข1 >< ๐ข1 ๐ข1 >= |๐ข1 >;
โข ๐2 = ๐ since we can show that:
โข ๐2 = ๐ ๐ข1 >< ๐ข1 = ๐ข1 >< ๐ข1 ๐ข1 >< ๐ข1| = ๐ข1 >< ๐ข1 = ๐;
โข The identity ๐ผ is a projection operator: ๐ผโ = ๐ผ, ๐ผ2 = ๐ผ;
โข For ๐ >= ฯ๐ ๐๐๐ข๐ โนฯ๐ ๐๐ ๐ >= ฯ๐ ๐ข๐ >< ๐ข๐ ๐ >= ฯ๐ ๐๐|๐ข๐ >= |๐ >;
โข If ๐๐ are eigenvalues of ๐ด with eigenvectors, |๐๐ >, then ๐ด ๐๐ >= ๐๐ ๐๐ >;we can
therefore write the operator ๐ด = ฯ๐๐ด ๐๐ >< ๐๐ = ฯ๐ ๐๐|๐๐ >< ๐๐|, where we have
used the spectral decomposition theorem discussed earlier;
17
Measurement โข Measurement is an important concept in quantum mechanics and is best understood through the
projection operator, ๐;
โข Assume a quantum mechanical system is initially in the state described by the vector |ฮจ >;
โข A measurement of the quantity ๐ for the system is represented by the operator แ๐ด whose
eigenvectors are given by
แ๐ด ๐๐ >= ๐๐ ๐๐ > Eqn. (3.13);
โข After a measurement, the system is projected onto the direction of the eigenvector |๐๐ >,
๐๐ ฮจ >= ๐๐ >< ๐๐ ฮจ >=< ๐๐ ฮจ > |๐๐ > Eqn.( 3.14);
โข The probability of the outcome of the measurement is given by < ๐๐|ฮจ > 2
18
โข Imagine a quantum system is in state |๐ > = ฯ๐ ๐๐|๐ข๐ > . If a measurement of an observable ๐ช associated
with the system is made, what is the result of the measurement?
โข An observable ๐ช is represented by an (Hermitian) operator ๐ช; therefore, a measurement of the system
could result in any one of the eigenvalues of the system, given by
๐ช ๐ข๐ >= ๐๐ ๐ข๐ >;
The question of interest is: which one? We need to compute the probability. Since the sum of all probabilities
of measuring any one of the eigenvalues is 1, we can write
1 = < ๐ ๐ > = < ๐ 1 ๐ > = < ๐
๐
๐ข๐ >< ๐ข๐ ๐ > =
๐
< ๐ ๐ข๐ >< ๐ข๐ ๐ >
โข The last expression simply says that ฯ๐ < ๐ ๐ข๐ >< ๐ข๐ ๐ > = ฯ๐ ๐๐2 = 1;
โข The probability of the measurement yielding one of the ๐๐ eigenvalues is therefore
๐๐ = < ๐ข๐|๐ > 2 = ๐๐2.
19
Another Perspective on Quantum Measurement
Generalization of the Gram-Schmidt Process for Hilbert space
โข The Gram-Schmidt orthogonalization process for 3๐ท space we discussed previously can
be generalized to ๐๐ท space following the same steps as before;
โข For a vector space ๐ spanned by the basis vectors ๐ข1, ๐ข2, โฆ ๐ข๐ with a defined inner
product, we can calculate an orthogonal basis ๐ฃ๐ by following the (algorithm) steps:
1. |๐ฃ1 >= |๐ข1 > Eqn. (3.15);
2. |๐ฃ2 > = ๐ข2 > โ<๐ฃ1|๐ข2>
<๐ฃ1|๐ฃ1>๐ข1 > Eqn. (3.16);
โฎ
๐. ๐ฃ๐ > = ๐ข๐ > โ<๐ฃ1|๐ข๐>
<๐ข1|๐ฃ1>๐ฃ1 > โ
<๐ฃ2|๐ข๐>
<๐ฃ2|๐ฃ2>๐ฃ2 > โโฏโ
<๐ฃ๐โ1|๐ข๐>
<๐ฃ๐โ1|๐ฃ๐โ1>| ๐ฃ๐โ1 > (3.17).
20
Generalized Vectors and Matrices: Tensors
โข We can combine two vectors ๐ข in โ๐ and ๐ฃ in โ๐ to create another vector in the
combined โ๐๐ space by the operation of multiplication; the new vector is called a
tensor product;
โข Supposed ๐ข =12
and ๐ฃ =345
โน ๐ขโจ๐ฃ =
1.31.41.52.32.42.5
=
3456810
;
โข We have produced a new vector ๐ขโจ๐ฃ in ๐โจ๐ in the space โ2โจโ3;
โข NB: only the tensor (Kronecker) product is of interest to us at this time.
21
Basis Vectors for Tensor Product Space
โข Any vector in ๐โจ๐ is a weighted sum of basis vectors;
note that ๐ is in โ๐ and ๐ is in โ๐;
โข The basis for ๐โจ๐ is the set of all vectors of the form
๐ฃ๐โจ๐ข๐ for ๐ = 1 to ๐ and ๐ = 1 to ๐;
โข The basis set for ๐โจ๐ is illustrated on the figure on the
right with 6 basis vectors;
โข Observe that ๐ฃโจ๐ข is the outer product of ๐ฃ and ๐ข:
๐ฃโจ๐ข = ๐ฃ๐ข๐;
โข We now learn apparently that any ๐ ร ๐ matrix can be
reshaped into an ๐๐ ร 1 vector and vice versa!
22
Tensors and Matrix Reshaping
โข Suppose we have the vectors ๐ฃ =123
and ๐ข =45, we can form the tensor product
๐ฃโจ๐ข = ๐ฃ๐ข๐ =123
4 5 =1.4 1.52.4 2.53.4 3.5
=4 58 1012 15
;
โข We can reshape the matrix into a vector or the vector into a matrix as below
4 58 1012 15
โบ
458101215
.
23
Tensors and Composite States
โข Suppose we have the two basis vectors: |0 > in ๐ป1and |1 > in ๐ป2, where ๐ป1 and ๐ป2are the respective Hilbert spaces. What is the basis for the combined Hilbert space
๐ป = ๐ป1โจ๐ป2?
โข Using what we have learned so far, the basis for the combined Hilbert space can be
written as all the possible products of the basis states for ๐ป1 and ๐ป2, thus
|0 > โจ |0 > = |00 >, |0 > โจ |1 > = |01 >, |1 > โจ|0 > = |10 >, |1 > โจ|1 > = |11 >;
โข In a more general example, given the state ๐ = ๐1 ๐ฅ > +๐2 ๐ฆ > with basis vectors
|๐ฅ > and |๐ฆ > in ๐ป1 and the state ๐ > = ๐1 ๐ข > +๐2|๐ฃ > with the basis vectors |๐ข >and |๐ฃ > in ๐ป2 one can create the combined Hilbert space ๐ป1โจ๐ป2 where the new state
is described by
|๐ > = ๐โจ๐ = ๐1 ๐ฅ > +๐2 ๐ฆ > โจ ๐1 ๐ข > +๐2 ๐ฃ >
โน ๐ >= ๐1๐1 ๐ฅ > โจ ๐ข > +๐1๐2 ๐ฅ > โจ ๐ฃ > +๐2๐1 ๐ฆ > โจ ๐ข > +๐2๐2 ๐ฆ > โจ๐ฃ >.
24
Tensors and Process-State Duality
โข We learned that ๐ฃโจ๐ข can also be a matrix, which is a mathematical process (operator)
representing a linear transformation, but we also found that ๐ฃโจ๐ข is an abstract vector,
which represents a state of a system;
โข Evidently matrices encode processes and vectors encode states;
โข We can therefore view the tensor product ๐โจ๐ as either a process or a state simply by
reshaping the numbers as a rectangle or a list;
โข By generalizing the idea of processes to higher dimensional arrays, we get tensors that
can be constructed to create tensor networks;
25
Graphical Perspective of a Tensor Productโข We previously created a tensor ๐ขโจ๐ฃ out of the two vectors ๐ข and ๐ฃ as illustrated below;
โข Another way to look at this process is by examining the interactions between the โnodesโ that
perform the multiplication or computation of the product; this perspective leads to the graphical
illustration alongside the vector tensor multiplication;
โข The graphical perspective is reminiscent of artificial neural network interconnections; in fact,
this perspective illuminates a computational scheme known as โtensor flowโ.
26
Operators and Tensors
โข Given and operator ๐ด that acts on |๐ > in ๐ป1 and an operator ๐ต that acts on |๐ > in ๐ป2,
we can create an operator ๐ดโจ๐ต that acts on the vectors in the combined vector space
๐ป1โจ๐ป2, thus
(๐ดโจ๐ต)( ๐ > โจ ๐ >) = ๐ด ๐ > โจ๐ต ๐ >;
โข If the two operators ๐ด and ๐ต are given as the matrices
๐ด =๐ ๐๐ ๐
and ๐ต =๐ ๐๐ โ
, we calculate the tensor product as
๐ดโจ๐ต =๐๐ต ๐๐ต๐๐ต ๐๐ต
=
๐๐ ๐๐ ๐๐ ๐๐๐๐ ๐โ ๐๐ ๐โ๐๐ ๐๐ ๐๐ ๐๐๐๐ ๐โ ๐๐ ๐โ
Eqn. (3.18).
27
Describing Quantum Particle Interactions with Tensors
โข Quantum particles, such as atoms, ions, electrons, and photons, can be manipulated to
perform some computing; the catch is that we must describe these particles interact with one
another;
โข We may need to know what these particles are doing, what their status (state) is, in how
many ways they are interacting, or what the probability of their being in certain states is; the
state of a quantum particle can be described by a complex unit vector in ๐๐;
โข The combined state of two particles, one described by a basis vector ๐ฃ1, in ๐๐and the other
by ๐ฃ2 in ๐๐ is a tensor product of their individual Hilbert spaces; thus
๐๐ โ๐๐;
โข As we discussed previously, the resulting basis vectors of the combined Hilbert space can
be thought of as a matrix, called the density matrix, ๐;
โข For N particles, density matrix ๐ is on ๐ถ๐โจ๐ถ๐โจ๐ถ๐โฆโจ๐ถ๐ = ๐ถ๐ ๐ Eqn. (3.19).
28
SVD of a Complex Matrix : Schmidt Decomposition
โข When matrix ๐ is complex, as in quantum mechanical operators, then ๐ = ๐ฮ๐โ ;
โข If we write the columns of ๐ as ๐ข๐ and those of ๐ as ๐ฃ๐, then
๐ =. | .. ๐ข๐ .. | .
โฑ๐๐
โฑ
โ โ ๐ฃ๐ โ โ = ๐1๐ข1๐ฃ1โ + ๐2๐ข2๐ฃ2
โ +โฏ+ ๐๐๐ข๐๐ฃ๐โ
Eqn. (3.20);
โข Recall that ๐ข๐ฃ๐ = ๐ขโจ๐ฃ = |๐ข >< ๐ฃ|; because of this, (3.20) above can be written as
(3.21) below;
โข Matrix ๐ can be written as state a |๐ > = ๐1๐ข1โจ๐ฃ1 + ๐2๐ข2โจ๐ฃ2 +โฏ+ ๐๐๐ข๐โจ๐ฃ๐โน |๐ >= ฯ๐=1
๐ ๐๐๐ข๐โจ๐ฃ๐ Eqn. (3.21).
29
SVD of a Complex Matrix and Quantum Entanglement
โข When we decompose a complex matrix ๐ we can write it as state vector
|๐ > = ฯ๐=1๐ ๐๐๐ข๐โจ๐ฃ๐ Eqn. (3.21);
โข The singular values ๐๐ are renamed the Schmidt coefficients, and the rank of the matrix
๐ (which is the number of non-zero singular values) ๐, is renamed the Schmidt rank;
โข Quantum state vector |๐ > now represents entanglement when the Schmidt rank ๐ > 1;however, the system is not entangled when the rank is not larger than 1;
โข Quantum entanglement is a resource in quantum computing and communication.
30
โข Reviewed complex linear algebra relevant for quantum mechanics (used in quantum
computing and communication applications);
โ Complex vectors
โ Complex matrices
โข Introduced the notion of tensor product
โ A way to interconvert between vectors and matrices and vice versa
โ Briefly introduced (mathematically) the idea of entanglement
31
Summary