Skip to main content
MathOverflow

Return to Question

removed capitals from title, formatting
Source Link
YCor
YCor
  • 63.9k
  • 5
  • 187
  • 286

Relationship between Eigenvectorseigenvectors of Projectedprojected and Original Matrixoriginal matrix

Let $A = Diag(\lambda_1, \dots, \lambda_n)$$A = \mathrm{Diag}(\lambda_1, \dots, \lambda_n)$ where $\lambda_1 \le \lambda_2 \dots \le \lambda_n$. Let $P = I - ww^T$ be a projection operator on an arbitrary $n$-dimensional hyperplane. Let $B = PAP$ have eigenvalues $\mu_1 \le \mu_2 \le \dots \mu_{n - 1}$ and (unit norm) eigenvectors $v_1, \dots, v_{n - 1}$. From Cauchy's interlacing theorem it follows that $\lambda_1 \le \mu_1 \le \lambda_2 \le \dots \mu_{n-1} \le \lambda_n$.

Is there anything that can be said about the relationship between inner products $\langle e_i, v_j \rangle$ depending on the eigenvalues of $A$ and $B$?

Relationship between Eigenvectors of Projected and Original Matrix

Let $A = Diag(\lambda_1, \dots, \lambda_n)$ where $\lambda_1 \le \lambda_2 \dots \le \lambda_n$. Let $P = I - ww^T$ be a projection operator on an arbitrary $n$-dimensional hyperplane. Let $B = PAP$ have eigenvalues $\mu_1 \le \mu_2 \le \dots \mu_{n - 1}$ and (unit norm) eigenvectors $v_1, \dots, v_{n - 1}$. From Cauchy's interlacing theorem it follows that $\lambda_1 \le \mu_1 \le \lambda_2 \le \dots \mu_{n-1} \le \lambda_n$.

Is there anything that can be said about the relationship between inner products $\langle e_i, v_j \rangle$ depending on the eigenvalues of $A$ and $B$?

Relationship between eigenvectors of projected and original matrix

Let $A = \mathrm{Diag}(\lambda_1, \dots, \lambda_n)$ where $\lambda_1 \le \lambda_2 \dots \le \lambda_n$. Let $P = I - ww^T$ be a projection operator on an arbitrary $n$-dimensional hyperplane. Let $B = PAP$ have eigenvalues $\mu_1 \le \mu_2 \le \dots \mu_{n - 1}$ and (unit norm) eigenvectors $v_1, \dots, v_{n - 1}$. From Cauchy's interlacing theorem it follows that $\lambda_1 \le \mu_1 \le \lambda_2 \le \dots \mu_{n-1} \le \lambda_n$.

Is there anything that can be said about the relationship between inner products $\langle e_i, v_j \rangle$ depending on the eigenvalues of $A$ and $B$?

Source Link
Dmitry
Dmitry
  • 231
  • 1
  • 8

Relationship between Eigenvectors of Projected and Original Matrix

Let $A = Diag(\lambda_1, \dots, \lambda_n)$ where $\lambda_1 \le \lambda_2 \dots \le \lambda_n$. Let $P = I - ww^T$ be a projection operator on an arbitrary $n$-dimensional hyperplane. Let $B = PAP$ have eigenvalues $\mu_1 \le \mu_2 \le \dots \mu_{n - 1}$ and (unit norm) eigenvectors $v_1, \dots, v_{n - 1}$. From Cauchy's interlacing theorem it follows that $\lambda_1 \le \mu_1 \le \lambda_2 \le \dots \mu_{n-1} \le \lambda_n$.

Is there anything that can be said about the relationship between inner products $\langle e_i, v_j \rangle$ depending on the eigenvalues of $A$ and $B$?