- #1
Playdo
- 88
- 0
Is this inequality true ever and when?
[tex]1/2 \phi{(p_{n}^{\sharp})} < \pi{(p_{n}^{\sharp})}-n[/tex]
[tex]1/2 \phi{(p_{n}^{\sharp})} < \pi{(p_{n}^{\sharp})}-n[/tex]
Last edited:
CRGreathouse said:n = 13 was the first example I noticed. What significance does this have?
CRGreathouse said:I don't know that 13 is the least one. I'm not even sure I properly understood the expression -- your ^\sharp is primorial, maybe? Why don't you give some examples of calculations.
Playdo said:Yes it is primorial. Hmm that is a new term. It's good to see that sequence getting special recognition it is very important.
Yes, so if you have Maple or mathematica just calculate that inequality for each sequential k and value of n until it fails. The pi is primes less than the primorial and the phi is eulers totient.
If you don't mind posting that info in a table here. I would really appreciate it.
The equation is simple to understand, all of the primes less than the primorial that are also in the reduced residue system modulo the primorial are accounted for on the right. The size of the reduced residue system is accounted for on the left. All true Goldbach Statements x+y = p#[n] must come from the reduced residue system. Thus if there are more than half as many primes as elements in the reduced residue system Golbach is true for that primorial by the pigeon-hole principle. The k comes in as a way to see how quickly the size of the residue system is decaying with respect to the number of primes in it.
Actually analytically the change between n can be calculated and the limit evaluated.
[itex] 1/2 \phi {(p^{\sharp}_{n+1})} - 1/2 \phi {(p^{\sharp}_n)}= 1/2 \phi{(p^{\sharp}#_{n})}{(p_{n+1}-2)}[/itex]
[itex]\pi{(p^{\sharp}_{n+1})}-n+1-\pi{(p^{\sharp}_{n})} - n =\pi{(p^{\sharp}_{n+1})}-\pi{(p^{\sharp}_n)} +1[/itex]
Is that clearing things up.
Playdo said:I do not have Mathematica on my computer so I can't quickly do what I suspect you did which is just evaluate the inequality using functions in Mathematica or Maple.
Ok I'll do the analysis and post it here eventually. I'll fill in the table too.CRGreathouse said:I have neither Mathematica nor Maple on my computer.
Playdo said:The next thing to do is fill in this table using the follwing inequality
[tex]1/2 \phi{(p_{n}^{\sharp})} < \pi{(p_{n}^{\sharp})}-n+k[/tex]
You already did k=0
Fill in the table
Least n for which the inequality is false | k
shmoe said:I'm sure I explained this to you in that other thread. pi(p#)/phi(p#)->0 as p->infinity, so that inequality will only be true for a finite number of n for any given fixed k.
edit- sorry, I removed a confusing and uneccessary pre-coffee bit and realize I should have quoted your bit in post #3 "If it is not true generally above some finite n then Goldbach being true is quite amazing."
shmoe said:I'm sure I explained this to you in that other thread. pi(p#)/phi(p#)->0 as p->infinity, so that inequality will only be true for a finite number of n for any given fixed k.
edit- sorry, I removed a confusing and uneccessary pre-coffee bit and realize I should have quoted your bit in post #3 "If it is not true generally above some finite n then Goldbach being true is quite amazing."
Playdo said:Yeah, so right within the context of the reduced residue system for the pirmorials (when did we start using that name?)
Playdo said:The thing there I guess I have not been able to put into words more precise that "wow" is that Goldbach being true for the primorials would mean that each new prime is less and less and less free to simply be any number.
Playdo said:For very large primorials the primes are getting less dense (I know the term is not being used properly, but if you can suggest one please do) as we get closer to the base.
Playdo said:We can write that into a more general looking table. But now for very large primorials the number of primes between anytwo numbers near to 1/2 the primorial will be very small and somehow, Goldbach being true says they have to fill up these registers so that at least one prime pair exists summing to the primorial. So if the bins are filled up sequentially left to right across the top and then right to left across the bottom, a probability argument would suggest that we are more likely to find primes around the primorial itself and that there is a sort of self similar thing going on.
Playdo said:[tex]0=\lim_{n\rightarrow\infty}\frac{{\frac{p_n^{\sharp}}{ln(p_n^{\sharp})}}}{\prod_{i=1}^n{(p_i^{\sharp}-1)}}>\lim_{n\rightarrow\infty}\frac{\pi{(p_n^{\sharp})}}{\phi{(p_n^{\sharp})}}[/tex]
shmoe said:I've seen the name 'primorial' credited to H. Dubner, sometime in the 80's I guess. It's a combination between prime and factorial. I mentioned this term in that other thread.
Goldbach being true wouldn't say the primes are less random. A more precise version of Goldbach's conjecture gives an asymptotic for the number of prime pairs that sum to a given number (due to Hardy and Littlewood). This is essentially based on treating the primes as a random sequence with density 1/log(n) as the prime number theorem suggests (and a few adjustments afterwards).
see the end of http://mathworld.wolfram.com/GoldbachConjecture.html for this asymptotic.
Less dense is fine, the limit I gave you is an asymptotic density.
I don't see what probability argument you are suggesting here, Goldbach being true doesn't mean you'd somehow need more primes near the primorial.
Look at the asymptotic for the number of solutions, and you should take the time to test how good this asymptotic is with some large numbers. I have a feeling you are going to be suprised on just how many prime pairs will add up to a given number.
You don't need any inequality like that (the denominator of the limit on the left isn't quite phi(p#) by the way, you shouldn't have the sharps "#" there). Just use pi(n)~n/log(n), log(p#)~p, and [tex]\prod_{p\leq x}(1-1/p)\sim e^{-\gamma}/\log{x}[/tex].
Playdo said:Ok, I'll check out the link. I would not be too surprised if there were many prime pairs adding to most even numbers. Trying to determine which number might have fewer such pairs would be one way of possibly discovering a class of numbers that had none.
Playdo said:But on your asymptotics. Maybe I am just not getting what tilda does for us, but is it not true that a tilda tells us nothing about how often such functions might be different? For instance phi(p#) is locally small, and if p# is between a prime pair it will be much smaller than the totient of both of those. That pi(x) is asymptotic to something does not reveal the details.
Playdo said:How do we know that there is not a subset of natural numbers for which pi(x) is always greater than x/logx? or less than it?
Playdo said:So that leaves me with what type of function as a guess for pi(x)? It must be one in which the first derivative is decreasing, but the first derivative can never reach zero because Euclid showed there are infinitely many primes.
Playdo said:It is an important point what you are saying. Let me go off and brush up on these asymptotic issues and then I will come back and either approve or give a good argument against your approach in solving (or avoiding the solution of) that inequality.
Playdo said:Yeah here we go http://mathworld.wolfram.com/AsymptoticNotation.html, it is just as I thought. That is not a very strong or exclusive result about the distribution of primes.
Playdo said:Much lauded the PNT, but not the end of research on the matter.
Playdo said:I'm not asking a question about the asymptotic density of primes in the reduced residue system of the primorials.
Playdo said:You do acknowledge that asymptotic arguments do not reveal the precise number of primes in any of those reduced residues systems under consideration, yes? Therefore asymptotics do not answer the question of how often (precisely) the above inequality is true or false.
The inequality involving phi() and pi() is a mathematical expression that compares the values of the golden ratio (phi) and the ratio of a circle's circumference to its diameter (pi). It is often represented as phi > pi.
The inequality involving phi() and pi() is used in mathematics to understand the relationship between the golden ratio and the ratio of a circle's circumference to its diameter. It can also be used to solve problems involving geometry and trigonometry.
Phi and pi are both important mathematical constants that have been studied extensively throughout history. Phi, also known as the golden ratio, appears in many natural and man-made structures and is considered to be aesthetically pleasing. Pi, on the other hand, is a fundamental constant in geometry and is essential for understanding circles and other curved shapes.
Yes, the inequality involving phi() and pi() can be proven using mathematical equations and principles. However, it is important to note that this inequality is based on the accepted values of phi and pi, which are both irrational numbers and can only be approximated to a certain degree of accuracy.
Yes, the inequality involving phi() and pi() can be applied to real-world scenarios such as architecture, art, and design. For example, the golden ratio is often used in the design of buildings, furniture, and other objects to create aesthetically pleasing proportions. Pi is also used in various fields such as engineering, physics, and statistics to solve real-world problems involving circles and curves.