Definition. Let V V V be a vector space over the field F F F and let T T T be a linear operator on V V V. A characteristic value of T T T is a scalar c c c in F F F such that there is a non-zero vector α \alpha α in V V V with T α = c α T\alpha=c\alpha Tα=cα. If c c c is a characteristic value of T T T, then
( a ) any α \alpha α such that T α = c α T\alpha=c\alpha Tα=cα is called a characteristic vector of T T T associated with the characteristic value c c c;
( b ) the collection of all α \alpha α such that T α = c α T\alpha=c\alpha Tα=cα is called the characteristic space associated with c c c.
Theorem 1. Let T T T be a linear operator on a finite-dimensional space V V V and let c c c be a scalar. The following are equivalent.
(i) c c c is a characteristic value of T T T.
(ii) The operator ( T − c I ) (T-cI) (T−cI) is singular (not invertible).
(iii) det ( T − c I ) = 0 \det (T-cI)=0 det(T−cI)=0.
Definition. If A A A is an n × n n\times n n×n matrix over the field F F F, a characteristic value of A A A in F F F is a scalar c c c in F F F such that the matrix ( A − c I ) (A-cI) (A−cI) is singular (not invertible).
The polynomial f = det ( x I − A ) f=\det (xI-A) f=det(xI−A) is called the characteristic polynomial of A A A.
Lemma. Similar matrices have the same characteristic polynomial.
Definition. Let T T T be a linear operator on the finite-dimensional space V V V. We say that T T T is diagonalizable if there is a basis for V V V each vector of which is a characteristic vector of T T T.
Lemma. Suppose that T α = c α T\alpha=c\alpha Tα=cα. If f f f is any polynomial, then f ( T ) α = f ( c ) α f(T)\alpha=f(c)\alpha f(T)α=f(c)α.
Lemma. Let T T T be a linear operator on the finite-dimensional space V V V. Let c 1 , … , c k c_1,\dots,c_k c1,…,ck be the distinct characteristic values of T T T and let W i W_i Wi be the space of characteristic vectors associated with the characteristic value c i c_i ci. If W = W 1 + ⋯ + W k W=W_1+\cdots+W_k W=W1+⋯+Wk, then
dim W = dim W 1 + ⋯ + dim W k \dim W=\dim W_1+\cdots+\dim W_k dimW=dimW1+⋯+dimWk
In fact, if B i \mathfrak B_i Bi is an ordered basis for W i W_i Wi, then B = ( B 1 , … , B k ) \mathfrak B=(\mathfrak B_1,\dots,\mathfrak B_k) B=(B1,…,Bk) is an ordered basis for W W W.
Theorem 2. Let T T T be a linear operator on a finite-dimensional space V V V. Let c 1 , … , c k c_1,\dots,c_k c1,…,ck be the distinct characteristic values of T T T and let W i W_i Wi be the null space of ( T − c i I ) (T-c_iI) (T−ciI). The following are equivalent.
(i) T T T is diagonalizable.
(ii) The characterristic polynomial for T T T is
f = ( x − c 1 ) d 1 ⋯ ( x − c k ) d k f=(x-c_1)^{d_1}\cdots (x-c_k)^{d_k} f=(x−c1)d1⋯(x−ck)dk
and dim W i = d i , i = 1 , … , k \dim W_i=d_i,i=1,\dots,k dimWi=di,i=1,…,k.
(iii) dim W 1 + ⋯ + dim W k = dim V \dim W_1+\cdots+\dim W_k=\dim V dimW1+⋯+dimWk=dimV.
Definition. Let T T T be a linear operator on a finite-dimensional vector space V V V over the field F F F. The minimal polynomial for T T T is the (unique) monic generator of the ideal of polynomials over F F F which annihilate T T T.
If A A A is an n × n n\times n n×n matrix over F F F, we define the minimal polynomial for