Taranovsky’s notation is a simple yet strong ordinal notation, introduced by Dmytro Taranovsky.

# Definition

First, Taranovsky’s notation is actually made up of many systems, called 1st system, 2nd system, 3rd system, and so on.

In the n-th system, we uses a binary function: C(α,β), and two constants: 0 and Ω_{n}. Here the 0 is not an ordinal or a number – it’s just a notation for “something”. And the same, Ω_{n} is not an ordinal – it’s just “something”.

Here’s the definition of *standard form*. First, 0 and Ω_{n} are in standard form. Then C(α,β) is in standard form iff it fits all those shown below:

- α and β are in standard form
- β = 0, or β = Ω
_{n}, or β = C(γ,δ) with α ≤ γ - α is n-built from below from <C(α,β)

But what’s “≤” and what’s “n-built from below from”? To answer this question, we need to define some more things.

First we need to define “n-built from below from” as follows.

- α is 0-built from below from <β iff α < β.
- α is (k+1)-built from below from β iff for all subterm γ of α, γ ≤ α or there is such a subterm δ of α that γ is subterm of δ and δ is k-built from below from β.
- The “subterm” in α can be defined as follows.
- In any part of expression of α, η is subterm of η itself.
- In any part of expression of α, if η = C(γ,δ), and x is a subterm of γ or δ, then x is a subterm of η.

To define binary relation “<” and “≤”, we need another form of the notation – *postfix form*. We can make a postfix form of every normal expression in such a way. First delete all the (′s, )′s and commas, which means we’ll get a string only containing 0, Ω_{n} and C’s. Then reverse the string. Now we get the postfix form. For **expressions in standard form**, we make the postfix form of them, then the comparisons (“≤”, “≥”, “<” and “>”) are done in the lexicographical order where C < 0 < Ω_{n}.

Property: In n-th system, for b < Ω_{n}, Ω_{n} is 1-built from below from b but not 0-built from below from b.

So why is it an ordinal notation? Because the “<” relation of expressions in standard form fits the ordinal axioms.

Now, we know that 0 is truly the ordinal 0. But unlike in θ function, Ω_{n} is not the n-th uncountable cardinal – it’s just some very large ordinal, but they’re also such common fixed point that , etc.

## One variable function

Any ordinal α can be expressed as or in standard form (k can also be 0). The former one is less than Ω_{n} (we call it *low-ordinal* here) while the latter one is greaterequal to Ω_{n} (we call it *high-ordinal* here).

Define the one variable function C_{1} as follows, it’s actually another representation of ordinals: for low-ordinal α, .

For high-ordinal α, , which is the same as ordinal addition and exponentiation.

It turns out that, if , then C_{1}(α) = α, and C_{1}(β+α) = C_{1}(β)+α.

All the 1st system, 2nd system, 3rd system, n-th systems can be combined into one notation as follows: the constants are 0 and Ω_{n} (for every positive integer n), and binary function C. Ω_{i} = C(Ω_{i+1}, 0) and the standard form always uses Ω_{i} instead of C(Ω_{i+1}, 0). And, to check for standard form and compare ordinals, use Ω_{i} = C(Ω_{i+1}, 0) to convert each Ω to Ω_{n} for a single positive integer n (it does not matter which n) and then use the n-th ordinal notation system.

Here‘s a calculator in Python program, which convert ordinals into Cantor’s normal form, binary φ function, and C_{1}-representation. Just note that it uses “C” instead of “C_{1}” for the one variable function, and “C2” instead of “C” for the binary function.

# Analysis

Below , doesn’t appear. For expressions without any Ω_{n}′s, α is 1-built from below from β, so “α is n-built from below from β” is always true.

So C(0,0) = 1, C(0,C(0,0)) = 2, C(0,C(0,C(0,0))) = 3, and C(0,α) = α+1.

Then and .

Then and .

and .

and .

and .

and .

and .

It turns out that, for , if C(α,β) is standard.

Here’re more results in 1st system.

. Note that ordinals above Ω_{1} are high-ordinals and C(0,Ω_{1}) = Ω_{1}+1 in 1st system.

It seems that the Ω_{1} in C function works very similar to the Ω in θ function. And they are also very similar to my EAN: Ω_{1} or Ω expands into “ω layers of” the C or θ where they are, and the results are the supremum of any finite layers of the C or θ where the Ω_{1} or Ω is.

The 1st system only handles ordinals up to the Bachmann-Howard ordinal. So we need 2nd system now. In 2nd system, Ω_{1} = C(Ω_{2},0), and the Bachmann-Howard ordinal is the supremum of all the C(C(Ω_{2},0),0), C(C(C(Ω_{2},0),C(Ω_{2},0)),0), C(C(C(C(Ω_{2},0),C(Ω_{2},0)),C(Ω_{2},0)),0), C(C(C(C(C(Ω_{2},0),C(Ω_{2},0)),C(Ω_{2},0)),C(Ω_{2},0)),0), etc. Thus the BHO equals .

. This ordinal (C(C(C(C(Ω_{2},Ω_{1}),Ω_{1}),C(Ω_{2},Ω_{1})),0)) is the supremum of all the C(C(C(Ω_{1},Ω_{1}),C(Ω_{2},Ω_{1})),0), C(C(C(C(Ω_{1},Ω_{1}),Ω_{1}),C(Ω_{2},Ω_{1})),0), C(C(C(C(C(Ω_{1},Ω_{1}),Ω_{1}),Ω_{1}),C(Ω_{2},Ω_{1})),0), etc. It’s not C(C(Ω_{2},Ω_{1})2,0)=C(C(C(Ω_{2},C(Ω_{2},0)),C(Ω_{2},C(Ω_{2},0))),0)! And is standard in 2nd system.

. Note again that C(0,C(Ω_{2},Ω_{1})) = C(Ω_{2},Ω_{1})+1.

. This ordinal (C(C(C(Ω_{2},Ω_{1}),C(Ω_{2},Ω_{1})),0)) is the supremum of all the C(C(Ω_{2},Ω_{1}),0), C(C(C(C(Ω_{2},Ω_{1}),Ω_{1}),C(Ω_{2},Ω_{1})),0), C(C(C(C(C(C(Ω_{2},Ω_{1}),Ω_{1}),C(Ω_{2},Ω_{1})),Ω_{1}),C(Ω_{2},Ω_{1})),0), etc.

.

. This ordinal (C(C(C(C(Ω_{2},Ω_{1}),C(Ω_{2},Ω_{1})),C(Ω_{2},Ω_{1})),0)) is the supremum of all the C(C(C(Ω_{2},Ω_{1}),C(Ω_{2},Ω_{1})),0), C(C(C(C(C(C(Ω_{2},Ω_{1}),C(Ω_{2},Ω_{1})),Ω_{1}),C(Ω_{2},Ω_{1})),C(Ω_{2},Ω_{1})),0), C(C(C(C(C(C(C(C(C(Ω_{2},Ω_{1}),C(Ω_{2},Ω_{1})),Ω_{1}),C(Ω_{2},Ω_{1})),C(Ω_{2},Ω_{1})),Ω_{1}),C(Ω_{2},Ω_{1})),C(Ω_{2},Ω_{1})),0), etc.

The supremum of all the C(C(C(0,Ω_{2}),0),0), C(C(C(C(C(0,Ω_{2}),0),Ω_{2}),0),0), C(C(C(C(C(C(C(0,Ω_{2}),0),Ω_{2}),0),Ω_{2}),0),0), etc. is C(C(C(Ω_{2},Ω_{2}),0),0) = C(C(Ω_{2}2,0),0). And that’s the limit of θ function.

The fundamental sequences in Taranovsky’s notation can be simply defined. First, we define a function: L(α), it’s the amount of C’s in the standard form of ordinal α. Then we define:

That’s it! So we get a very fast-growing function: with n C’s. It marks the strength of Taranovsky’s notation.

The full strength of Taranovsky’s notation is unknown yet. It might be as weak as second-order arithmetic (Z_{2}), and might be as strong as second-order arithmetic with projective determinacy (Z_{2}+PD), according to Taranovsky’s page.

How much is C(C(W_2*2+1,0),0)?

LikeLike

See the pDAN vs TON page. It is approx to psi(W_(psi_I(0)w)).

LikeLike

θ(α↦Ω_α) = θ(Ω_Ω_Ω_Ω_Ω_…) ω-times – is limit of θ() function.

But I can continue:

θ(Ω_Ω_Ω_Ω_Ω_…) ω-times = α_0

θ(Ω_Ω_Ω_Ω_Ω_…) α_0-times = α_1

θ(Ω_Ω_Ω_Ω_Ω_…) α_1-times = α_2

e.t.c.

α_ω

α_θ(Ω)

α_θ(Ω^Ω)

α_θ(Ω_Ω)

α_θ(Ω_Ω_Ω_Ω_Ω_…) = α_α_0

e.t.c.

α_α_0

α_α_α_0

α_α_α_α_… ω-times = β↦α_β = β_0

How to write in Taranovsky’s notation or pDAN values:

α_1

α_2

α_ω

α_α_0

β_0

e.t.c.

?

LikeLike

Using the 2nd system:

Your α_0 = C(C(Ω2,0),0)

α_1 = C(C(Ω2+C(Ω+C(Ω2,0)+α_0,0),0),0)

α_2 = C(C(Ω2+C(Ω+C(Ω2,0)+α_1,0),0),0)

α_ω = C(C(Ω2+C(Ω+C(Ω2,0)+C(Ω,0),0),0),0)

α_α_0 = C(C(Ω2+C(Ω+C(Ω2,0)+C(Ω,0)×α_0,0),0),0)

β_0 = C(C(Ω2+C(Ω+C(Ω2,0)+C(Ω,0)^2,0),0),0)

The results seem erratic, because things like C(C(Ω+C(Ω2,0),0),C(C(Ω2,0),0)) aren’t standard.

LikeLike

I’m confused as compared…

C(C(Ω2+C(Ω+C(Ω2,0)+C(Ω,0),0),0),0) > C(C(Ω2,C(Ω2*2,0)),0) ~ {1,,2{1,,1,,2}2}

or

C(C(Ω2+C(Ω+C(Ω2,0)+C(Ω,0),0),0),0) < C(C(Ω2,C(Ω2*2,0)),0) ~ {1,,2{1,,1,,2}2}

?

I just want to compare α_1, α_ω, α_α_0, β_0, etc with pDAN

LikeLike

OK. Compared with pDAN, we have:

{1,,1{1,,1,,2}2} has recursion level α_0

{1,,1{1,,1,,2}1{1{1,,1{1,,1,,2}2}2}2} has recursion level α_1

{1,,1{1,,1,,2}1{1{1,,1{1,,1,,2}1{1{1,,1{1,,1,,2}2}2}2}2}2} has recursion level α_2

{1,,1{1,,1,,2}1`2} has recursion level α_ω

{1,,1{1,,1,,2}1`1{1{1,,1{1,,1,,2}2}2}2} has recursion level α_α_0

{1,,1{1,,1,,2}1`1`2} has recursion level β_0

LikeLike

Do I understand correctly that if I compare

{1,,1{1,,1,,2}1`2} – ε_0 level of 1`2

{1,,1{1,,1,,2}1`1`2} – ζ_0 level of 1`1`2

{1,,1{1,,1,,2}1,,1{1,,1,,2}2} – θ(α↦Ω_α) level of 1,,1{1,,1,,2}2

сan it be said that the transition from {1,,1{1,,1,,2}2} to {1,,1{2,,1,,2}2} will be:

( use some function Ф() )

α_0 = Ф(ω)

α_1 = Ф(ω^ω)

α_2 = Ф(ω^ω^ω)

α_ω = Ф(φ(1, 0))

β_0 = Ф(φ(2, 0))

β_ω = Ф(φ(3, 0))

e t.c.

{1,,1{1,,1,,2}1,,1{1,,1,,2}2} – Ф(θ(α↦Ω_α)) = Ф(α) ?

after:

Ф(α)

Ф(Ф(α))

Ф(Ф(Ф(α)))

Ф(Ф(Ф(Ф(…)))) – {1,,1{2,,1,,2}2} ?

LikeLike

You are quite wrong.

LikeLike

So lexicographical order of C represent ordinals? How do you evaluate the value of C’s?

LikeLike

Just see C, 0 and Ω as simply symbols (don’t have “value”), and let C < 0 < Ω.

The values of "C-represented ordinals" come from comparisons.

LikeLike

And I recalled Reverse Polish Notation.

LikeLike

I don’t get how to compare in postfix form.

LikeLike

It’s in lexicographical order. For example, “” < "C" < "CC" < "CCC" < "CCCCCCCCCC" < "C0" < "C0C" < "C0CCCCC" < "C0CC0" < "C0C0" < "C0CΩ" < "C00" < "C0Ω" < "CΩ" < "CΩC" < "CΩ0" < "CΩΩ" < "CΩΩΩ" < "0" < "0C" < "0CC" < "00" < "00C" < "0Ω" < "Ω".

Generally, the empty string is the least one, "C" is larger, then "0" is larger, then "Ω" is larger. And, if string a < string b, then ca < cb for any string c.

LikeLike

Could someone explain to me Aarex Tiaokhiao’s number Otopersuperoogol?

LikeLike

My Old Googology Is Closed!!!!

LikeLike

C(1,1) is stronger, which is the 2nd fixed point of C(0,C(0,…))

LikeLike

1 is C(0,0), so your C(1,1) is C(C(0,0),C(0,0)). But it’s not in standard form, because C(0,0) > 0, and it’s against the 2nd rule of standard form.

In Taranovsky’s notation, we only see expressions in standard form as ordinals.

LikeLike

Oh. I got that. By the way. What C() have level {1,,1{1,,1,,2}3}?

LikeLike