To keep things simple, let's start with the case
where and
,
,
, and let's just try to find
. Thus we
want
so that
,
,
.
As a first goal, let's just try to make a polynomial
that is zero at and at
but not at
. That is easy: The function
works. It is nonzero at
because the value there
is a product of two nonzero factors.
However, at the value is
and
not
as desired. Fortunately, all we need to do is
to divide the function by the 10 and we're done:
.
To show where this result came from, we could write
as
.
Now we're ready for . The same
reasoning gives
. You can
check that
at
has values
respectively. Similarly,
.
Now it should be clear how to make ...,
for general
:
is a fraction whose numerator is a product of factors
except for
; the denominator is a
product of the same factors except with
instead of
. To write all this more
compactly, it is handy to use a sign
(capital pi)
for a product, the same way that
is used for
summation.
Theorem 4.1 . The solution to the polynomial interpolation problem can be expressed as
, where for each
,
.
Proof of Theorem . Just substitute in
, and you will find that each
has the right
value, and consequently
. Observe that each
has degree
. Each coordinate function of
is a linear combination of
and
so has degree at most
.
Proof of Theorem . The existence of the required
solution is shown by Theorem
. The uniqueness is shown in
the Exercises.
Note. The polynomials depend only
on
and the times
, and not on the
data points
. In Figure
are shown
graphs of the polynomials
for
and times
.