Let T be the tridiagonal matrix with diagonal entries A(1) ,...,


    A(N) and offdiagonal entries B(1) ,..., B(N-1)).  DSTECH checks to


    see if EIG(1) ,..., EIG(N) are indeed accurate eigenvalues of T.


    It does this by expanding each EIG(I) into an interval


    [SVD(I) - EPS, SVD(I) + EPS], merging overlapping intervals if


    any, and using Sturm sequences to count and verify whether each


    resulting interval has the correct number of eigenvalues (using


    DSTECT).  Here EPS = TOL*MAZHEPS*MAXEIG, where MACHEPS is the


    machine precision and MAXEIG is the absolute value of the largest


    eigenvalue. If each interval contains the correct number of


    eigenvalues, INFO = 0 is returned, otherwise INFO is the index of


    the first eigenvalue in the first bad interval.

N

          N is INTEGER


          The dimension of the tridiagonal matrix T.

A

          A is DOUBLE PRECISION array, dimension (N)


          The diagonal entries of the tridiagonal matrix T.

B

          B is DOUBLE PRECISION array, dimension (N-1)


          The offdiagonal entries of the tridiagonal matrix T.

EIG

          EIG is DOUBLE PRECISION array, dimension (N)


          The purported eigenvalues to be checked.

TOL

          TOL is DOUBLE PRECISION


          Error tolerance for checking, a multiple of the


          machine precision.

WORK

          WORK is DOUBLE PRECISION array, dimension (N)

INFO

          INFO is INTEGER


          0  if the eigenvalues are all correct (to within


             1 +- TOL*MAZHEPS*MAXEIG)


          >0 if the interval containing the INFO-th eigenvalue


             contains the incorrect number of eigenvalues.

Univ. of Tennessee

Univ. of California Berkeley

Univ. of Colorado Denver

NAG Ltd.