한국어 잘못해서 영어로 씀. Since A is symmetric, by the spectral theorem there exists an orthonormal basis of eigenvectors. Let v_1, ..., v_n be this basis, then for any x, x=a_1v_1+...+a_n v_n for some constants a_i. Now, since A is a projection (idempotent) matrix, we get that the only eigenvalues of A are 0 or 1, hence each basis vector v_i is either an eigenvector of 0 or 1.
토론토수학(74.15)2019-02-18 06:50
Now, let E_1 be the eigenspace corresponding to the eigenvalue 1, and let dim(E_1)=k<n. WLOG, reorder your basis such that v_1,...,v_k are in E_1. Then we see that Ax=a_1Av_1+...+a_kAv_k+...+a_nAv_n=a_1v_1+...+a_kv_k. So we see that the image of A has dimension k, so rank(A)=dim(E_1)
아 그리고 어제 (a)번 답변 달아주신 수갤형님 감사합니다. (__)
Symmetric이면 대각화가능함
한국어 잘못해서 영어로 씀. Since A is symmetric, by the spectral theorem there exists an orthonormal basis of eigenvectors. Let v_1, ..., v_n be this basis, then for any x, x=a_1v_1+...+a_n v_n for some constants a_i. Now, since A is a projection (idempotent) matrix, we get that the only eigenvalues of A are 0 or 1, hence each basis vector v_i is either an eigenvector of 0 or 1.
Now, let E_1 be the eigenspace corresponding to the eigenvalue 1, and let dim(E_1)=k<n. WLOG, reorder your basis such that v_1,...,v_k are in E_1. Then we see that Ax=a_1Av_1+...+a_kAv_k+...+a_nAv_n=a_1v_1+...+a_kv_k. So we see that the image of A has dimension k, so rank(A)=dim(E_1)