On a vectorized version of a generalized Richardson extrapolation process

Convergence and divergence of series and sequences (40A05) Approximation to limiting values (summation of series, etc.) (40A25) Extrapolation to the limit, deferred corrections (65B05)

Abstract: Let

x x_{m}

be a vector sequence that satisfies

xx_msim sss+sum^infty_{i=1}alpha_i gg_i(m)quad ext{as

m o i n f t y

},

s s s

being the limit or antilimit of

x x_{m}

and

{g g_{i} (m)}^{i} n f t y_{i = 1}

being an asymptotic scale as

m o i n f t y

, in the sense that

lim_{m oinfty}frac{|gg_{i+1}(m)|}{|gg_{i}(m)|}=0,quad i=1,2,ldots.

The vector sequences

{g g_{i} (m)}^{i} n f t y_{m = 0}

,

i = 1, 2, l d o t s,

are known, as well as

x x_{m}

. In this work, we analyze the convergence and convergence acceleration properties of a vectorized version of the generalized Richardson extrapolation process that is defined via the equations

sum^k_{i=1}�raket{yy,Deltagg_{i}(m)}widetilde{alpha}_i=�raket{yy,Deltaxx_m},quad nleq mleq n+k-1;quad sss_{n,k}=xx_n+sum^k_{i=1}widetilde{alpha}_igg_{i}(n),

s s s_{n, k}

being the approximation to

s s s

. Here

y y

is some nonzero vector,

is an inner product, such that

, and

D e l t a x x_{m} = x x_{m + 1} - x x_{m}

and

D e l t a g g_{i} (m) = g g_{i} (m + 1) - g g_{i} (m)

. By imposing a minimal number of reasonable additional conditions on the

g g_{i} (m)

, we show that the error

s s s_{n, k} - s s s

has a full asymptotic expansion as

n o i n f t y

. We also show that actual convergence acceleration takes place and we provide a complete classification of it.

Review text

Let \(\mathbb{X}\) be a finite- or infinite-dimensional linear inner product space with the inner product \(\langle\cdot,\cdot\rangle\) defined such that \(\langle \alpha\, \mathbf{a}, \beta\, \mathbf{b}\rangle =\bar{\alpha}\beta \langle \mathbf{a}, \mathbf{b}\rangle\), and let \(\|\cdot\|\) be the norm induced by this inner product, namely, \(\|z\|=\sqrt{\langle z,z \rangle}\). Further, let \(\{\mathbf{x}_m\}\) be a vector sequence in \(\mathbb{X}\), and let the \(\mathbf{x}_m\) have an asymptotic expansion of the form \[ \mathbf{x}_m \sim \mathbf{s}+\sum_{i=1}^\infty \alpha_i \mathbf{g}_i(m), \qquad m \to \infty, \] \(\mathbf{s}\) being the limit or antilimit of \(\{\mathbf{x}_m\}\) and \(\mathbf{g}_i(m)\) being an asymptotic scale as \(m \to \infty\), in the sense that \[ \lim_{m \to \infty} \frac{\|\mathbf{g}_{i+1}(m)\|}{\|\mathbf{g}_{i}(m)\|}=0, \quad i=1,2,\dots. \] A vectorized version of the generalized Richardson extrapolation process that is defined via the equations \[ \sum_{i=1}^k \langle \mathbf{y}, \Delta \mathbf{g}_i(m)\rangle \widetilde{\alpha}=\langle \mathbf{y}, \Delta \mathbf{x}_m\rangle, \; n\leq m \leq n+k-1; \] \[ \mathbf{s}_{n,k}=\mathbf{x}_n+\sum_{i=1}^k \widetilde{\alpha}_i \mathbf{g}_i(n), \] \(\mathbf{s}_{n,k}\) being the approximation to \(\mathbf{s}\) is investigated in this work. The properties of convergence and convergence acceleration of generalized Richardson process are analysed. It is shown that the error \(s_{n,k}-s\) has a full asymptotic expansion as \(m \to \infty\). It shows that actual convergence acceleration takes place, too.

Recommendations

Cites work

Cited in

(2)

This page was built for publication: On a vectorized version of a generalized Richardson extrapolation process

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q513676)