Title says all. For the readers more deeply interested: the
authors consider a non-homogeneous linear differential equation of
an $m-$th order containing also terms with a negative shift
$-\tau$ of the coordinate. On a closed interval $(a,b)$ containing
a point $c$ the mixed three-point boundary conditions connecting
solutions at $a,\ b$ and $c$ are imposed. Then, everything
(including the solution) is assumed approximated by the truncated
Taylor-series expansion of order $N$ at the point $c$. The claim
is that via a careful insertion of all the resulting ``matrix"
Taylor ansatzs one can determine the Taylor coefficients of the
solution by the standard linear matrix inversion technique. For
confirmation, four numerical examples are added, with error terms
of the order of percents and with $N$ never larger than 11 and
with $m=3$ (once) and $m=2$.

MR2176354 Gülsu, Mustafa; Sezer, Mehmet
A Taylor polynomial approach for solving differential-difference equations.
J. Comput. Appl. Math. 186 (2006), no. 2, 349--364.
65Q05 (34K28)