Which is larger, Graham's number or


by MathJakob
Tags: graham, larger, number
MathJakob
MathJakob is offline
#1
Sep15-13, 12:11 PM
P: 153
This isn't a homework problem or anything just simply curious. Which is a larger number, Graham's number or the estimated number of atoms in the universe factorial?

So Graham's number or ##(10^{80})!##
Phys.Org News Partner Mathematics news on Phys.org
Researchers help Boston Marathon organizers plan for 2014 race
'Math detective' analyzes odds for suspicious lottery wins
Pseudo-mathematics and financial charlatanism
Boorglar
Boorglar is offline
#2
Sep15-13, 01:42 PM
PF Gold
P: 161
Well, as an upper bound, [itex]\left(10^{80}\right)! < \left(10^{80}\right)^{\left(10^{80}\right)} = 10^{80 * 10^{80}} = 10^{8*10^{81}} [/itex]

which, if I am not mistaken, is far less than [itex]g_1[/itex].
MathJakob
MathJakob is offline
#3
Sep15-13, 01:46 PM
P: 153
So ##(10^{80})!## isn't even as large as g1... and Graham's number is g64... WOW

arildno
arildno is offline
#4
Sep15-13, 01:48 PM
Sci Advisor
HW Helper
PF Gold
P: 12,016

Which is larger, Graham's number or


The Arildno number is even greater:

Arildno number =Graham's number+1.

Can I also get a WOW! from you?
MathJakob
MathJakob is offline
#5
Sep15-13, 02:23 PM
P: 153
Quote Quote by arildno View Post
The Arildno number is even greater:

Arildno number =Graham's number+1.

Can I also get a WOW! from you?
No. Your number is meaningless, it's never been used in any meaningful way.
arildno
arildno is offline
#6
Sep15-13, 02:27 PM
Sci Advisor
HW Helper
PF Gold
P: 12,016
Quote Quote by MathJakob View Post
No. Your number is meaningless, it's never been used in any meaningful way.
fedaykin
fedaykin is offline
#7
Sep17-13, 12:21 PM
P: 124
Sorry to link to wikipedia, but you may be interested in TREE(3), and I can't find too much info on it.

an extremely weak lower bound for TREE(3), is A(A(...A(1)...)), where the number of As is A(187196)
Graham's number, for example, is approximately A^64(4) which is much smaller than the lower bound A^(A(187196))(1).

A is the Ackermann function.

http://en.wikipedia.org/wiki/Kruskal%27s_tree_theorem
johnqwertyful
johnqwertyful is offline
#8
Sep17-13, 09:48 PM
P: 308
Quote Quote by MathJakob View Post
This isn't a homework problem or anything just simply curious. Which is a larger number, Graham's number or the estimated number of atoms in the universe factorial?

So Graham's number or ##(10^{80})!##
Tip: if you can write out the number in any "normal" way, it's waaaaaaaaaaaaaaaaay smaller than Graham's number.
Deedlit
Deedlit is offline
#9
Sep22-13, 01:43 PM
P: 13
What's amusing is that the above lower bound for TREE(3) is SO much smaller than TREE(3) that it gives people the wrong idea about it's size. Although the lower bound is larger than Graham's number, it's still at about the level ω+1 in the fast-growing hierarchy. TREE(3), on the other hand, is higher than the Small Veblen Ordinal in the fast-growing hierarchy, which is MUCH larger!
arildno
arildno is offline
#10
Sep22-13, 01:57 PM
Sci Advisor
HW Helper
PF Gold
P: 12,016
Quote Quote by Deedlit View Post
What's amusing is that the above lower bound for TREE(3) is SO much smaller than TREE(3) that it gives people the wrong idea about it's size. Although the lower bound is larger than Graham's number, it's still at about the level ω+1 in the fast-growing hierarchy. TREE(3), on the other hand, is higher than the Small Veblen Ordinal in the fast-growing hierarchy, which is MUCH larger!
It seems the Arildno number is a dwarf, after all..
Mentallic
Mentallic is offline
#11
Sep22-13, 07:48 PM
HW Helper
P: 3,436
Quote Quote by arildno View Post
It seems the Arildno number is a dwarf, after all..
Poor thing, the Arildno number is coming under lots of fire in here!

Quote Quote by Deedlit View Post
What's amusing is that the above lower bound for TREE(3) is SO much smaller than TREE(3) that it gives people the wrong idea about it's size. Although the lower bound is larger than Graham's number, it's still at about the level ω+1 in the fast-growing hierarchy. TREE(3), on the other hand, is higher than the Small Veblen Ordinal in the fast-growing hierarchy, which is MUCH larger!
I haven't really been able to find any sources that explain some of the larger operators in the fast growing hierarchy. For example,

http://googology.wikia.com/wiki/Fast-growing_hierarchy

gives a list of growth rates, but there's a lot of details being left out. Sadly, I only understand how to calculate up to and including [itex]f_{\epsilon_0}(n)[/itex].

And to do so, I use a calculator of course
Deedlit
Deedlit is offline
#12
Sep23-13, 08:58 PM
P: 13
Good to see you found the Googology Wiki!

Beyond [itex]\varepsilon_0[/itex] we have:

[itex]\varepsilon_1 = \varepsilon_0 ^ {\varepsilon_0 ^ {\varepsilon_0 ^\ddots}}[/itex]

[itex]\varepsilon_2 = \varepsilon_1 ^ {\varepsilon_1 ^ {\varepsilon_1 ^\ddots}}[/itex]

[itex]\varepsilon_\omega = \sup \lbrace \varepsilon_0, \varepsilon_1, \varepsilon_2, \ldots \rbrace [/itex]

and so on. Eventually we get to

[itex]\zeta_0 = \varepsilon_{\varepsilon_{\varepsilon_\cdots}}[/itex]

We can set [itex]\varphi(0, \alpha) = \omega^\alpha, \varphi(1, \alpha) = \varepsilon_\alpha, \varphi(2, \alpha) = \zeta_\alpha[/itex]. More generally,

[itex]\varphi(\alpha+1, \beta) = [/itex] the [itex]\beta[/itex]th fixed point of [itex]f(\gamma) = \varphi(\alpha, \gamma)[/itex].

When [itex]\alpha[/itex] is a limit ordinal, [itex]\varphi(\alpha, \beta)[/itex] is the [itex]\beta[/itex]th ordinal in the intersection in the ranges of [itex]f(\delta) = \varphi(\gamma, \delta)[/itex] for all [itex]\gamma < \alpha[/itex].

So, for example,

[itex]\varphi(3, 0) = \varphi(2, \varphi(2, \varphi(2, \ldots)))[/itex]

[itex]\varphi(\omega, 0) = \sup (\varphi(1, 0), \varphi (2, 0), \varphi (3, 0), \ldots)[/itex]

and so on. This takes us up to

[itex]\Gamma_0 = \varphi(\varphi(\varphi(\ldots, 0),0),0)[/itex]

or alternatively, [itex]\Gamma_0[/itex] is the smallest ordinal [itex]\alpha[/itex] such that [itex]\alpha = \varphi(\alpha, 0)[/itex].

We can continue the notation with [itex]\varphi(1, 0, 0) = \Gamma_0[/itex], and [itex]\varphi(1, 0, \alpha)[/itex] is the [itex]\alpha[/itex]th fixed point of [itex]f(\beta) = \varphi(\beta, 0)[/itex].

The ordinals [itex]\varphi (1, \alpha, \beta)[/itex] are defined analogously to [itex]\varphi(\alpha, \beta)[/itex], i.e. each function [itex]f(\beta) = \varphi (1, \alpha+1, \beta)[/itex] enumerates the fixed points of [itex]g(\beta) = \varphi(1, \alpha, \beta)[/itex], and at limit ordinals you enumerate the intersection of the ranges of previous ordinals. Then [itex]\varphi(2, 0, \alpha)[/itex] is the [itex]\alpha[/itex]th fixed point of [itex]f(\beta) = \varphi(1, \beta, 0)[/itex], and you can construct the hierarchy [itex]\varphi(2, \alpha, \beta)[/itex] similarly. At limit ordinals you take the intersection of the ranges again, and so we define [itex]\varphi (\alpha, \beta, \gamma)[/itex] for all ordinals [itex]\alpha, \beta, \gamma[/itex]. Then we can define

[itex]\varphi (1, 0, 0, \alpha)[/itex] to be the [itex]\alpha[/itex]th fixed point of [itex]f(\beta) = \varphi(\beta, 0, 0)[/itex].

We can then define [itex]\varphi(\alpha, \beta, \gamma, \delta), \varphi(\alpha, \beta, \gamma, \delta, \epsilon)[/itex], and so on. The general definition for the n-ary Veblen funciton is:

[itex]\varphi (\alpha) = \omega^{\alpha}[/itex]

[itex]\varphi (\alpha_1, \alpha_2, \ldots, \alpha_n + 1, 0, \ldots, 0, \beta)[/itex] is the [itex]\beta[/itex]th fixed point of the function [itex]f(\gamma) = \varphi(\alpha_1, \alpha_2, \ldots, \alpha_n, \gamma, 0, \ldots, 0)[/itex]

When [itex]\alpha_n[/itex] is a limit ordinal, [itex]\varphi (\alpha_1, \alpha_2, \ldots, \alpha_n, 0, \ldots, 0, \beta)[/itex] is the [itex]\beta[/itex]th ordinal in the intersection of the ranges of [itex]f(\delta) = \varphi(\alpha_1, \alpha_2, \ldots, \alpha_{n-1}, \gamma, \delta, 0, \ldots, 0)[/itex] for all [itex]\gamma < \alpha_n[/itex].

We define the Small Veblen Ordinal as

[itex]\sup (\varphi (1, 0), \varphi(1, 0, 0), \varphi(1, 0, 0), \ldots)[/itex].

TREE(n) is larger than the fast-growing hierarchy at the level of the Small Veblen Ordinal. (how much further is not known.).

I'll go one step further: we can extend the n-ary Veblen function to transfinitely many places. Obviously, we can't write out transfinitely many variables, so we need to modify our notation: instead of

[itex]\varphi(\alpha, \beta, \gamma, \delta, \epsilon)[/itex]

we write

[itex]\varphi(\alpha @ 4, \beta @ 3, \gamma @ 2, \delta @ 1, \epsilon @ 0)[/itex].

So we append "@ n" to every variable, where n represents the index of the variable. This allows us to skip variables that are 0, and so we can notate things like [itex]\varphi (1 @ \omega, \alpha @ 0)[/itex]. [itex]\varphi(1 @ \omega, \alpha @ 0)[/itex] is defined as the [itex]\alpha[/itex]th ordinal that is a fixed point of [itex]f(\beta) = \varphi (\beta @ n)[/itex] for all [itex]n < \omega[/itex].

More generally, we define

[itex]\varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \alpha_n + 1 @ \beta_n + 1, \gamma @ 0)[/itex] is the [itex]\gamma[/itex]th fixed point of [itex]f(\delta) = \varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \alpha_n @ \beta_n + 1, \delta @ \beta_n)[/itex]

When [itex]\alpha_n[/itex] is a limit ordinal, [itex]\varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \alpha_n @ \beta_n + 1, \gamma @ 0)[/itex] is the [itex]\gamma[/itex]th ordinal in the intersection of the ranges of [itex]f(\epsilon) = \varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \delta @ \beta_n + 1, \epsilon @ \beta_n)[/itex] for all [itex]\delta < \alpha_n[/itex]

When [itex]\beta_n[/itex] is a limit ordinal, [itex]\varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \alpha_n + 1 @ \beta_n, \gamma @ 0)[/itex] is the [itex]\gamma[/itex]th ordinal in the intersection of the ranges of [itex]f(\epsilon) = \varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \alpha_n @ \beta_n, \epsilon @ \delta)[/itex] for all [itex]\delta < \beta_n[/itex]

When [itex]\alpha_n[/itex] and [itex]\beta_n[/itex] are limit ordinals, [itex]\varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \alpha_n @ \beta_n, \gamma @ 0[/itex] is the [itex]\gamma[/itex]th ordinal in the intersection of the ranges of [itex]f(\zeta) = \varphi(\alpha_1 @ \beta_1, \alpha_2 @ \beta_2, \ldots, \delta @ \beta_n, \zeta @ \epsilon)[/itex] for all [itex]\delta < \alpha_n, \epsilon < \beta_n[/itex]

This defines [itex]\varphi(\alpha_1 @ \beta_1, \ldots, \alpha_n @ \beta_n)[/itex] for all [itex]\alpha_i[/itex] and [itex]\beta_i[/itex]. This notation is known as Schutte's Klammersymbolen.

The smallest ordinal [itex]\alpha[/itex] such that [itex]\alpha = \varphi (1 @ \alpha)[/itex] is known as the Large Veblen Ordinal. I would think that TREE(n) would not reach the Large Veblen Ordinal in the fast-growing hierarchy, but I don't think this is known.

Phew! I hope this was at least somewhat comprehensible.


Register to reply

Related Discussions
determine if the next number will be larger or smaller General Math 2
How to represent a small change in a larger number General Math 1
Graham's number Linear & Abstract Algebra 5
Why is it when I divide by a number greater than 1, I get a larger number? General Math 2
Graham's Number compared to... anything? General Math 16