Expansion of series and possible induction

CAF123
Gold Member
Messages
2,918
Reaction score
87

Homework Statement


Consider the infinite series $$\frac{x}{e^x - 1} = A_o + A_1 x + \frac{A_2}{2!}x^2 + ... + \frac{A_n}{n!}x^n + ...$$ Determine that ##1 = A_o,\,\,\,\,\,0 = A_o/2! + A_1,\,\,\,\,\,0 = A_o/3! + A_1/2! + A_2/2!##.
Show that for ##n > 1##, one can write the relations as $$(A+1)^n - A^n,$$ where ##A^n \rightarrow A_n##.

2. Homework Equations

Taylor expansions and induction

The Attempt at a Solution


First part is fine. I thought about using induction on the second part since ##n## is confined to an integer. I understand ##A^n \rightarrow A_n## as meaning the powers of the series converge to the actual terms in the series, which is why I thought ##n## was an integer, labeling the terms in the series. However, I can't seem to make sense of the base case (n=2). It is $$(A+1)^2 - A^2 = 2A + 1$$ What is ##A##?

thanks
 
Physics news on Phys.org
Anyone any comments or help? thanks.
 
CAF123 said:

Homework Statement


Consider the infinite series $$\frac{x}{e^x - 1} = A_o + A_1 x + \frac{A_2}{2!}x^2 + ... + \frac{A_n}{n!}x^n + ...$$ Determine that ##1 = A_o,\,\,\,\,\,0 = A_o/2! + A_1,\,\,\,\,\,0 = A_o/3! + A_1/2! + A_2/2!##.
Show that for ##n > 1##, one can write the relations as $$(A+1)^n - A^n,$$ where ##A^n \rightarrow A_n##.

2. Homework Equations

Taylor expansions and induction

The Attempt at a Solution


First part is fine. I thought about using induction on the second part since ##n## is confined to an integer. I understand ##A^n \rightarrow A_n## as meaning the powers of the series converge to the actual terms in the series, which is why I thought ##n## was an integer, labeling the terms in the series. However, I can't seem to make sense of the base case (n=2). It is $$(A+1)^2 - A^2 = 2A + 1$$ What is ##A##?

thanks

What this is saying (I think) is that you can expand the left hand side of (A + 1)^n - A^n = 0, and if you then replace A^k with A_k where 0 \leq k < n, you can find A_{n-1} in terms of A_0, \dots, A_{n-2}.

Thus: (A + 1)^2 - A^2 = 2A + 1 and indeed 2A_1 + A_0 = 0.
 
pasmith said:
What this is saying (I think) is that you can expand the left hand side of (A + 1)^n - A^n = 0, and if you then replace A^k with A_k where 0 \leq k < n, you can find A_{n-1} in terms of A_0, \dots, A_{n-2}.
I see, is there a reason why you neglected the case where ##k=n##?
Thus: (A + 1)^2 - A^2 = 2A + 1 and indeed 2A_1 + A_0 = 0.
I nearly see what you did there except how did you get that ##A=A_1##? Is this just a notation ##A = A^1 = A_1##?

Would you say induction is the way to go? I can try further given I now understand what the base case means, thanks.
 
CAF123 said:
I see, is there a reason why you neglected the case where ##k=n##?

The A^n terms cancel.

I nearly see what you did there except how did you get that ##A=A_1##? Is this just a notation ##A = A^1 = A_1##?

x^1 = x for every x.

Would you say induction is the way to go?

It is the obvious method.
 
CAF123 said:
I nearly see what you did there except how did you get that ##A=A_1##? Is this just a notation ##A = A^1 = A_1##?
It's not that ##A^1 = A_1##, but that it represents it in the analogy between the two sets of equations.
 
I tried the induction argument, but seemed to run into a mistake. Assume ##P(n) = (A + 1)^n - A^n = 0 ## is true for ##n##. Then $$(A+1)^{n+1} - A^{n+1} = (A+1)^n (A+1) - A^n A $$$$= ((A+1)^n - A^n)A + (A+1)^n = A^n \neq 0$$ The ##A^n## are just numbers so I am not sure where the mistake is.
 
I think you want to be using induction to establish that <br /> \sum_{k=0}^{n-1} \binom{n}{k} A_k = 0<br /> since <br /> (A+ 1)^n - A^n = \sum_{k=0}^{n-1} \binom{n}{k} A^k.<br />
 
Back
Top