dee...@pmail.ntu.edu.sg writes:
> On Monday, September 10, 2012 7:08:51 PM UTC+5:30, hidy ho wrote:
>> <
dee...@pmail.ntu.edu.sg> wrote in message
>> news:e7b08c8f-acdf-435d...@googlegroups.com...
>> > Let A and B be two fixed reals, where B is not equal to 0.
>> take out "fixed"
>> >
>> > THEOREM: As K tends to positive infinity, the value of (A+(B/K)) is A, if
>> > and only if, A is not equal to 0.
>>
>> wrong. K could go to negative infinity, also A being 0 has no special
>> requirement (if and only if) on (A+(B/K))
>>
>> > PROOF: Consider the expression (A+(B/K)), which is (AK+B)/K. As K is
>> > extremely large, we can neglect B and this expression is approximately
>> > equal to AK/K = A, if and only if, A is not equal to 0.
>>
>> this is not a proof, rather a spoof.
>>
>> > Look forward to your comments on my THEOREM.
>>
>> see above
>
> My Theorem intended to have the "as K tends to positive infinity"
> phrase completely outside the if and only if part (I was already aware
> that this works even if K tends to negative infinity). My Theorem also
> intended to use the phrase "value tends to" instead of "value is".
There was no problem in how you had it before. Moving the phrase "Let K
tend to positive infinity" from where you had it, just makes thing less
clear. Also, what problem is that you see when K tends to negative
infinity? Why insist on +oo?
> Fine, I agree it makes sense to remove the word "fixed".
>
> Let me rephrase my Theorem as follows (and simultaneously I re-word
> the proof):
>
> Definition: Let A and B be two reals, where B is not equal to 0. Let K
> tend to positive infinity.
>
> THEOREM: (A+(B/K)) tends to A, if and only if, A is non-zero.
I'd write:
For all real a, b, lim k->oo (a+b/k) = a.
But then I probably would not bother at all. The more general point is
that the limit of f(k)+g(k) is the sum on the limits of f(k) and of
g(k).
> PROOF: Given two Boolean statements P and Q, then one way of proving
> "P if and only if Q" is to first prove Q implies P, and to then prove
> not-Q implies not-P. So lets first prove that A is non-zero implies
> that (A+(B/K)) tends to A. (A+(B/K)) = (AK+B)/K. Here B can be
> neglected as AK is very much larger in magnitude. So (AK+B)/K tends to
> AK/K = A, if A is non-zero. So now lets focus on proving that A is
> zero implies that (A+(B/K)) cannot tend to 0. Let us assume that B/K
> tends to 0. Let us introduce a large positive L where L is much
> greater than K. Multiplying both sides of B/K = 0 by L, means that
> BL/K can be approximated as 0, which is a contradiction. This means
> that A is zero implies that (A+(B/K)) cannot tend to 0. HENCE PROVED
The second part is wrong. BL/K tends to 0 just as B/K does. In fact, BL
is just a different choice for B.
The first part is not really a proof. "What can be neglected" can only
be used in sketches of proofs and can't be used here at all where the
theorem is so simple. If you still think it's worth proving, do it from
first principles.
> Note what happens if my argument of multiplying both sides by L is
> carried out for the case of A not equal to 0. Assuming that (A+(B/K))
> tends to A. So with this assumption, multiplying both sides by L means
> that L(A+(B/K)) tends to LA, which means that (L(KA+B))/K tends to
> LA. We are allowed to neglect B, so (L(KA+B))/K tends to (L(KA))/K,
> which is equal to LA. So we do not get any contradiction in this case.
--
Ben.