Next: Acknowledgements
Up: p35
Previous: Convolutions of generalized Fibonacci
Generalized
Fibonacci numbers
are defined by
eq. 1 for
,
and
, together with the
inputs
for
. For
one recovers the generalized Fibonacci numbers
treated above.
Lemma 6: The generating function
for the generalized
Fibonacci numbers is given by eq. 43.
Proof: From the recurrence with inputs given in eq. 1.
Lemma 7 (Riccati eq. for the generalized
Fibonacci case):
If
(non-degenerate
case) then
satisfies the Riccati eq. 45 with the
polynomials
and
defined in
eqs. 46 and 47.
Proof: Use the ansatz
 |
(60) |
together with
 |
(61) |
from eq. 43. This implies
 |
(62) |
Comparing first coefficients of the powers
, for
,
leads to the eqs. (remember that
)
 |
(63) |
With this result the coefficients of the powers
for
become, after iteration,
 |
(64) |
Therefore the eqs. 63 are now
 |
(65) |
Finally, the coefficient of
has to satisfy
which implies, together with the above results,
 |
(66) |
Inserting this into the above found expressions finally leads to
With
one thus finds eqs. 47 and 46. The assertion can now be
proved directly with the found polynomials.
Note 4: i) If
,
generates
powers of
, and one has to put
and
. This means that one puts
for
.
ii) For given non-vanishing
and
is a polynomial in
of
degree
, and
is one of degree
. The sum in
can be evaluated to yield the second of eqs. 47
provided
.
Lemma 8 (Coefficient triangles of numbers for polynomials
and
):
The coefficients of the polynomials defined in eqs. 48 are given
by eqs. 49 and 50.
Proof:
from eq. 47 leads immediately to
eq. 50, remembering that
. Then eq. 49
follows from eq. 46 and
.
Proposition 5 (Uniqueness of Riccati solution; non-degenerate
case):
If
then
is the unique solution of the
Riccati eq. 45 with eqs. 46, 47 and
initial value
.
Proof: As a special Bernoulli eq. the Riccati eq.
is equivalent to the inhomogeneous linear differential eq. for
:
. Because
is continuous in the strip
,
and is there
-Lipschitz,
the existence and uniqueness theorem for linear differential eqs. proves the
assertion (see e.g.[8], §6,I, p.62ff). In order to find
one uses the summed expression for
from eq. 47 and repeated
applications of the triangle inequality.
Lemma 9 (First convolution of
; non-degenerate case):
The first convolution of the sequence
, which is defined
analogously to the first of eqs. 17, is given by eq. 51
with eq. 52.
Proof: One has to compute the coefficients of
of the lhs of
eq. 45, taking into account the
dependence of
and
with the coefficients from eqs. 49 and 50.
The case
has to be considered separately. The recurrence eq. 1
cannot be used in order to simplify the sum in eq. 51.
Proposition 6 (Recurrence for
st power of the generating function
; non-degenerate case):
For
eq. 53 gives
.
Proof:
due to eq. 45.
Proposition 7 (
fold convolution of
; non-degenerate case):
For
the entry
of the
fold convolution of the sequence
is given by eq. 54.
Proof: This follows immediately from Proposition 5 after comparing
coefficients of
using the definition
Lemma 10 (Degenerate case
):
If
then
satisfies the first order linear differential eq. 57.
Proof: One proves
with eqs. 47 and 46 in the version where the sum has been evaluated (the case
is
treated separately). If one factors out
one finds that all
terms cancel provided one replaces
by
.
Note 5: The solution
of this linear differential eq. with input
is unique. The proof is analogous to the one of proposition 5.
Note 6: If
we do not have a formula for
, valid for all
, like in the non-degenerate case. Therefore, we cannot derive results for convolutions
along the line shown above.
Lemma 11 (Recurrence in the degenerate case):
If
(and
) then one can replace the recurrence eq. 1 which
has depth
, by the following one with depth
.
 |
(73) |
where one uses the inputs
for
.
Proof:
This derives from the sum on the rhs of eq. 51 which now
vanishes. If the coefficients
from eq. 52 are used with
the replacement of
by
one arrives at the
desired recurrence, after the common factor
has been dropped. The inputs
are adopted from the original recurrence except that
can now be
computed to be
.
Next: Acknowledgements
Up: p35
Previous: Convolutions of generalized Fibonacci
Wolfdieter Lang
2002-04-04