One good way to get an upper bound for the partial sums of a series is with an integral:
PROOF:
By constructing step functions based on the series we can see that
![]()
Now ifconverges then we get
![]()
giving a bound on the partial sums. Thusconverges.
If, on the other hand,
diverges then the numbers
get arbitrarily large, and thus so do the partial sums. This tells us that the series also diverges.
PROOF:
The integral
![]()
For p>1 this converges to. For p<1 it diverges. If p=1 then the integral gives a natural
which diverges as
.
Examples:
Usually we use the limit comparison theorem with p-series when checking endpoints of intervals of convergence:
PROOF:
Suppose thatand
converges. Then for large enough n we know that
![]()
thus an<(L+1)bn. Now ifconverges then so does
so the comparison theorem will tell us that
converges.
If L>0 we can invert to see that
![]()
so ifconverges, so does
.
Examples:
converges by limit comparison with
.
Here
diverges by limit comparison with
.
Here
Exercises: Here are some for you to try:




