You are here:

Advanced Math/Monotone sequences and natural logarithm

Advertisement


Question
Hi Paul,  I have had a go at this question but I think my attempt is pretty clumsy and possibly too weak.

Question is:

Show the sequence {a_n} given by a_n = [SUM (r=1 to n)  (1/r) - ln n] is monotone.

My attempt goes along the lines:

using induction, taking n = 1 and n=2, I get a_1 = 1 and a_2 = 0.8069 and thus a_2 - a_1 < 0, therefore it appears to be decreasing.

If i then assume true n = k (k > 0), we have a_k = SUM r=1 to k (1/k) - ln k

And then if you insert n = k+1, you have a_(k+1) = SUM r=1 to k+1 (1/(k+1) - ln (k+1).

a_(k+1) - a_k = 1/(k+1) - ln (k+1) - ln(k), which will be < 0  ##

Thus we have by induction that a_n is strictly decreasing and thus, according to Defn 2.1.8 (given in my unit book), it is monotone.

## I have thought about this since and am not convinced I can jump to the conclusion that this is always < 0

Can you suggest an alternative approach?

Thanks.

Answer

Log
Questioner:   Steve
Country:  Australia
Category:  Advanced Math

Subject:  Monotone sequences
Question:  Hi Paul,  I have had a go at this question but I think my attempt is pretty clumsy and possibly too weak.

Question is:

Show the sequence {a_n} given by a_n = [SUM (r=1 to n)  (1/r) - ln n] is monotone.

My attempt goes along the lines:

using induction, taking n = 1 and n=2, I get a_1 = 1 and a_2 = 0.8069 and thus a_2 - a_1 < 0, therefore it appears to be decreasing.

>> Yes, that appears so.

If i then assume true n = k (k > 0), we have a_k = SUM r=1 to k (1/k) - ln k

And then if you insert n = k+1, you have a_(k+1) = SUM r=1 to k+1 (1/(k+1) - ln (k+1).

a_(k+1) - a_k = 1/(k+1) - ln (k+1) - ln(k), which will be < 0  ##

>> OOPS -- I think you blew a sign here.  I think you want PLUS ln(k)

Thus we have by induction that a_n is strictly decreasing and thus, according to Defn 2.1.8 (given in my unit book), it is monotone.

## I have thought about this since and am not convinced I can jump to the conclusion that this is always < 0

Can you suggest an alternative approach?

Thanks.
...............................

a[k]   =  SUM(j=1 to k)(1/r)  - ln k

a[k+1] =  SUM(j=1 to k+1)(1/r)  - ln (k+1)

a[k+1] - a[k] = 1/(k+1) - ln(k+1) + ln k

a[k+1] - a[k] = 1/(k+1) - (ln(k+1) - ln k)

I think we have the signs right now.

We want to determine whether this is positive or negative.  How about this approach?

ln k = (by definition)

{k
| dt/t
}1

and ln(k+1) = (by definition)

{k+1
| dt/t
}1

So ln (k+1) - ln k is certainly positive and, by rules for adding integrals, is:

{k+1
|   dt/t
}k

Now that is the area under the graph of 1/t from t = k to t = k+1

Now what is 1/(k+1)?
That is the area of the rectangle from t = k to t = k+1 and bounded on top by  y = 1/(k+1), which is a smaller area than the integral.

So

1/(k+1) - (ln(k+1) - ln k)  is smaller minus bigger, so it is negative.  Does that do it?

Paul Klarreich

Expertise

I can answer questions in basic to advanced algebra (theory of equations, complex numbers), precalculus (functions, graphs, exponential, logarithmic, and trigonometric functions and identities), basic probability, and finite mathematics, including mathematical induction. I can also try (but not guarantee) to answer questions on Abstract Algebra -- groups, rings, etc. and Analysis -- sequences, limits, continuity. I won't understand specialized engineering or business jargon.

Experience

I taught at a two-year college for 25 years, including all subjects from algebra to third-semester calculus.

Education/Credentials
-----------

©2012 About.com, a part of The New York Times Company. All rights reserved.