next up previous
Next: The inverse eigenvalue problem Up: A discrete Gelfand - Previous: Introduction

Transmutation matrices

Let T, tex2html_wrap_inline577 be tridiagonal (n,n)-matrices, i.e.

displaymath581

By a transmutation matrix for T, tex2html_wrap_inline577 one usually means a (n,n)-matrix L with tex2html_wrap_inline591 . We shall use (approximate) transmutation matrices for which this holds only in rows 1 through n-1. More precisely, with E the projection

displaymath597

we define a transmutation L by

  equation79

theorem83

Proof: With tex2html_wrap_inline613 the rows of L, (2.1) reads

displaymath617

Since tex2html_wrap_inline619 this recursion determines the tex2html_wrap_inline621 uniquely once tex2html_wrap_inline623 is given. For tex2html_wrap_inline625 , tex2html_wrap_inline621 is nonzero only in its first i components since T is tridiagonal. tex2html_wrap_inline633

We remark that Theorem 2.1 holds also for lower Hessenberg matrices tex2html_wrap_inline577 , T.



Frank Wuebbeling
Fri Oct 9 14:01:16 MET DST 1998