Arithmetic Derivative
@Kainui
\[ k=ab \implies k' = a'b+ab' \]When \(k\) is prime, \(k'=1\).
We established that \[ (k^n)' = k'nk^{n-1} \]
For every prime \(p\) such that \(p|k\): \[ k' = k\sum_p\frac 1p \]
Rules about primes shouldn't matter too much, generalizing to composite is better, in my opnion.
We also established a couple of fun facts: \[(p^p)' = p^p\] and \[(n!)' = n! \sum_{k=1}^n \frac{k'}{k}\]
What would \( (-1)' \) be? It would extend us to integers.
I think we should think more generally than (-1)' and try to look at \[\large (e^{i \theta})'\]
Well, \(-1\) is tough because a negative number by a positive number is negative.
Actually maybe we should consider looking at just simply fractions first, they seem like they might be fairly interesting. For instance, look at this pattern that begins to emerge when I write their derivatives out. \[\large 1' \ 2' \ 3' \ 4' \ 5' \ 6' \ 7' \ 8' \ 9' \ 10' \\ \large 0 \ \ 1 \ \ 4 \ \ \ 1 \ \ 5 \ \ 1 \ 12 \ \ 6 \ \ 7 \ \ 1\] It sort of begins to count 4, 5, 6, 7 with apparently a random 12 and some 1's in there while completely avoiding 2 and 3. Maybe this isn't really a pattern? But it might give us an idea of how to interpolate?
Oh yeah, remember the formula: \[ \frac{a'}{a}+ \frac{b'}{b} = \frac{a'b+ab'}{ab} = \frac{c'}{c} \]
\[ 1 \sim c\\ p \sim x\\ c \sim f(x) \]
\[ \frac{(-2)'}{-2}+\frac{(-2)'}{-2} = \frac{-2(-2)' + -2(-2)'}{4} = \frac{-4(-2)'}{4} \]We should expect that \[ \frac{(-2)'}{-2}+\frac{(-2)'}{-2} = \frac{4'}{4} \]This iimplies \[ -4(-2)' = 4' \]
This implies that \((-2)' = -1\)
\[1 = (-1)(-1)\] implies \[0 = (-1)'(-1)+(-1)(-1)'\]\[0=-2(-1)'\]\[0=(-1)'\]
\[ (-a)' = (-1)'a +(-1)a' = -a' \]
Yeah you beat me to it haha.
Next is:\[ \left(\frac{1}{a}\right)' \]I know the answer, but I wonder how we can get it using our rules.
Try something like 4 = 12/3
it should be consistent.
\[ (a^{-1})' = -a'a^{-2} \]
\[ \left(\frac 1a\right)' = -\frac{a'}{a^2} \]
\[1= a *a^{-1}\]\[0 = (a)'a^{-1}+a(a^{-1})'\]
\[ \left(\frac ab\right)' = (ab^{-1})' = a'b^{-1} + a(b^{-1})' \]
Yeah, it's all consistent with what we had before with (k^n)' like you wrote at the beginning.
\[ = \frac{a'}{b} + \frac{-ab'}{b^2} = \frac{a'b-ab'}{b^2} \]
Next step is rational exponents.
We would expect:\[ a^n = a'na^{n-1} \]to still hold true.
\[ (\sqrt{a})' = a'\frac12a^{-1/2} = \frac{a'}{2\sqrt{a}} \]
What is weird is how similar it is to derivative, yet anti-derivative has no meaning.
\[ (a+b)' = a' (1+b/a) + b'(1+a/b) \]
\[a=a^{k/k}=(a^k)^{1/k}\]\[\large a' = \frac{1}{k}(a^k)^{\frac{1}{k}-1}(a^k)'=\frac{1}{k}(a^k)^{\frac{1}{k}-1}*ka^{k-1}a'\] \[\large 1=(a^k)^{\frac{1}{k}-1}a^{k-1}\]\[\large 1=a^{1-k +k-1}\]
\[ a' + \frac{a'b}{a} + b' + \frac{ab'}{b} = \frac{a'ab + ab^2+abb'+a^2b'}{ab} = (a+b)' \]
Did I mess this last one up?
It looks like it could be useful as hell.
Yeah I'm trying to figure it out, I have been wanting that for a while now haha
What I did was \(a+b = a(1+b/a)=b(1+a/b)\)
If let \(b=1\), then we get \[ (a+1)' = \frac{a'a+a}{a} \]
Oh no, did I crack the code?
I messed up, because it would mean \[ (a+1)' = a' + 1 \]which is not true.
Oh, I know what I did wrong now.
I wasn't consistent with my factoring.
\[ (a+b)' = a'(1+b/a) + a(1+b/a)' \]
I am playing around with the integral and it's sort of bizarre how it doesn't work quite right. \[6'=5=(\frac{1}{2}5^2)'\]
\[ (a+1)' = a'(1+1/a) + a(1+1/a)' \]
Well I guess it does work, I was thinking as though the 1/2 is a constant... But it's not even a thing here @_@
\[ a'\left(\frac{1+a}{a}\right) + a\left(\frac{1+a}{a}\right) ' \]
I think if we can do \((a+1)'\) we win.
but the \(1+a\) inside the derivative means we'll probably end up at a circular equality.
\[(\frac{1}{2}5^2)' = -(\frac{5}{2})^2+5=-\frac{5}{4} \ne 5\] for the record. Hmm I think I might have a way to make the derivative work. It's just so odd that it's not a linear operator but so many other properties seem to hold pretty well.
Where did you get \(\frac 12 5^2\)?
In my misguided attempt to integrate 5. I knew 6'=5 and thought I could just increase the exponent by 1 and divide by it.
Another interesting thing, 21'=25'=10
\[ (a^n)' = a'na^{n-1} \]
\[ (p^n)' = np^{n-1} \]So you can do the anti derivative if you think it comes from a prime.
But only \(1\) can come from a prime.
Are there any cases where a'=1 when a is not prime?
Well now that we have negative numbers floating around, maybe it's possible?
Well, suppose there was a composite \(a\) such that \(a'=1\).
It would mean that \((a^a)' = a^a\)
Well I'm not just saying composite, but any rational or irrational number we have access to now.
Oh
We would need to search for:\[ 1 = \frac{(a^{n})'}{na^{n-1}} \]
\[ 1=\frac{a'a^{n-1}+a(a^{n-1})'}{na^{n-1}} \]
\[(-1)' = (e^{i \pi})'\]\[0=e' i \pi e^{i \pi -1}\]\[0=e'\]
If only we figured out: \[ (f(n))' \]
\[i' = (e^{i \pi /2})'\]\[i' = e' i \frac{\pi}{2}e^{i \pi/2 -1}\]\[i'=0\]
\[ i = (-1)^{1/2} = (-1)'(\ldots) = 0 \]
\[e' = [\sum_{n=0}^\infty \frac{1}{n!} ]'=0\]\[e' = [\lim_{n \rightarrow \infty} (1+\frac{1}{n})^n]'=0\]
What about \(\pi\) as a series?
I don't know, I think it has several kinds of infinite representations. But I'm more interested in working out some kind of sum rule. I just sorta wanted to throw those out there since they sorta came to mind.
There is also vectors.
Ohhhh speaking of which: \[\LARGE a+bi = re^{i \theta}\]
Perhaps gcd thing somehow involves \(aa'\).
For primes \(aa' = a\).
Interesting thought.
\[ \langle a',b' \rangle \cdot \langle b,a \rangle = a'b+ab' \]
We could say \(\mathbf v' \) is just \(\langle v_1',v_2',\ldots\rangle\).
Unless you think there is a better way of defining it.
\[ \mathbf v\cdot \mathbf v' = v_1v_1'+v_2v_2'+\ldots \]
For a vector of primes, it would just be the sum of the elements.
Hmmm well if it is consistent with something that seems natural. I don't know if I'm willing to say anything yet. I'm currently looking at; \[(a+bi)' = (re^{i \theta})'=r'e^{i \theta}+r (e^{i \theta})'=r'e^{i \theta}\] So the derivative seems for complex numbers to be only dependent on the length not the direction.
What I'm saying is the derivative should be the derivative of the length of the vector in the same direction as the old vector.
And \[ r = \sqrt{a^2+b^2} \]?
Yes, exactly. Or also written as \[r= [(a+bi)(a-bi)]^{1/2}\]
But it follows from this fairly well: \[(re^{i \theta})' = r'e^{i \theta}\] since e'=0
So you are saying: \[ \mathbf v' =|\mathbf v|' \frac{\mathbf v}{|\mathbf v|} \]
Yeah perfect.
So when \(|v|\) is prime, you end up with a unit vector.
Yeah exactly so v=<3, 4> has a derivative v' as a unit vector in the same direction
And the anti derivative of any unit vector is simply to multiply it by any prime number.
And the derivative of any unit vector is 0 vector.
Oddly enough. So instead of adding +C it's like multiplying by p^1
I have to wonder now...
For all \(a\), does there exist an \(n\), such that:\[ a^{(n)} = 1 \]Or do loops exist?
Well, loops exist for \(p^p\)
Yeah, and some actually increase for instance 12'=16, 16'=32, etc?
\[ 12=2^23 \\ 16 = 2^4\\ 32 = 2^5 \]
Join our real-time social learning platform and learn together with your friends!