Testwiki:Reference desk/Archives/Mathematics/2016 January 18

{| width = "100%"

|- ! colspan="3" align="center" | Mathematics desk |- ! width="20%" align="left" | < January 17 ! width="25%" align="center"|<< Dec | January | Feb >> ! width="20%" align="right" |Current desk > |}

Welcome to the Wikipedia Mathematics Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

January 18

Increase and decrease

There is an initial number a and every step increases or decreases by $a_{n}b$, and there is n steps.

An example for three steps, where $a_n$ is the number at step n

$a_1=a\pm ab$

$a_2=a_1\pm a_{1}b$

$a_3=a_2\pm a_{2}b$

I figured out the formula for any $a_n$.

(1) $a(\pm b+1{)}^n$

But it works only with b>0 for an increase of a>0 and a decrease of a<0, and -1>b<0 for a decrease of a>0 down to 0 (which is impossible to reach, in $a(-b+1{)}^n=0$ the number n must be infinite).

But for a negative (reaching 0 and then going further from 0) decrease of a>0 with b>|1|, it must be thus

(2) $-a(b-1)(b+1{)}^{n-1}$

where b must be positive, not negative as in (1).

A small alteration of (1) also will work

(3) $-a(b+1{)}^{n-1}$

An example for a=1 decreasing by 200% (b=2) every step:

$a_1=1-1\times 2=-1$

$a_2=-1-1\times 2=-3$

$a_3=-3-3\times 2=-9$

With formula (1) it won't work:

$1\times (-2+1{)}^3=-1$

but formulas (2) and (3) work fine:

$-1\times (2-1)\times (2+1{)}^2=-9$

$-1\times (2+1{)}^2=-9$

Is there an explanation for this?

P.S. By the way, what branch of mathematics am I doing?--Lüboslóv Yęzýkin (talk) 21:45, 18 January 2016 (UTC)

Recurrence relation or difference equation. Loraof (talk) 23:45, 18 January 2016 (UTC)

I think your calculation for a2 is wrong. It should be $a_2=a_1-a_1(b)=-1-(-1)(2)=1.$ So formula (1) gives the right answer. Loraof (talk) 00:01, 19 January 2016 (UTC)

Yes, theoretically, I tried that and came to the same. But if we imagine the number line, we start from 1, then go to the left by the double of that 1*2 and end at -1. Then we go to the left by the double of what we got that is |1|*2 and we end at -3. Then we go to the left again by that pattern and end at -9, -27, -81, -243... Probably, it is not decreasing in a strict sense but what is then, a "negative stepping back"? With (1) it will always end either at -1 for an odd step or at 1 for an even step. Note also that for the opposite case, with a negative start (e.g. a=-1) and b<-1 (e.g. b=-2) formula (1) will always end at 1.
P.S. Sorry, if my reasoning sounds amateurish, I'm not a mathematician at all and I might have used wrong terminology and understood it all wrong, but this case really intrigues me. Did I hack maths? :) --Lüboslóv Yęzýkin (talk) 01:30, 19 January 2016 (UTC)

Imagine a problem (I'd call it "A Mafia Loan Problem"). A person borrowed $1, the debt grows by the above-mentioned pattern every month. In the 1st month his ballance is -1, he owes $2. In the 2nd the balance is -3 and he owes $4. In the 3rd the balance is -9 and he owes $10. And so on. What would the balance be in 12 months? With (3) it is easy, $-1\times (2+1{)}^{11}=-177147$. --Lüboslóv Yęzýkin (talk) 01:39, 19 January 2016 (UTC)

So if I'm understanding this right, you want the value to always decrease, whether a is positive or negative (so for a=1, b=2, you get: a0=1, a1= 1 - 2*1= -1, a2= -1 - |2*-1| = -3, etc.). This means your formula is actually $a_n=a_{n-1}+|a_{n-1}|b$ where |x| is the absolute value of x (i.e. |-a| = a). I think, given the absolute value expression in here, you're not going to be able to get a simple expression for the result, but perhaps someone who knows more maths than I do can come along and provide it. MChesterMC (talk) 11:09, 19 January 2016 (UTC)

Template:Reply to Thanks, you seem to be right. Now it is more straightforward: with b>0 it is always goes to the right on the number line, with b<0 goes to the left. We can also avoid the absolute value and write $a_n=a_{n-1}+b\sqrt{a_{n-1}^2}$.

But I have no idea how to generalize a formula for any $a_n$. E.g.:

$a_0=a$

$a_1=a+b\sqrt{a^2}$

$a_2=a_1+b\sqrt{a_1^2}=(a+b\sqrt{a^2})+b\sqrt{(a+b\sqrt{a^2}{)}^2}$

$a_3=a_2+b\sqrt{a_2^2}=((a+b\sqrt{a^2})+b\sqrt{(a+b\sqrt{a^2}{)}^2})+b\sqrt{((a+b\sqrt{a^2})+b\sqrt{(a+b\sqrt{a^2}{)}^2}{)}^2}$

And so on, more steps the longer and more esoteric the formula becomes. Wolfram Alpha suggests the simplifications: $a_1=a(b+1)$, $a_2=a(b+1{)}^2$, $a_3=a(b+1{)}^3$, but only if a and b are positive, that is the same formula (1) I've suggested at the beginning. But how to find a single formula that will work with any rational number? We returned to the starting question. I suggest we have two different sequences (I have no idea how they are named): one tends to or from zero, the other goes in any direction on the number line, and I've tried to figure out a single formula for the both, but maybe it is impossible?--Lüboslóv Yęzýkin (talk) 09:50, 23 January 2016 (UTC)