> Full-text translation from Chinese. The original remains authoritative for its research claims; this translation is not independent mathematical verification.

# ACTM Round 8
## Difference-Saturation Graphs, Three-Term Progression Certificates, and the Binary-Tower Survivor
### Difference-Saturation Graphs, Three-Term Progression Certificates, and the Binary-Tower Survivor

**Research programme:** Algebraic Combinatorial Transcendence Methodology (ACTM)  
**Authors:** Neo.K; Aletheia (GPT-5.6 Sol)  
**Institution:** EveMissLab / 一言諾科技有限公司  
**Date:** 2026-10-02  
**Version:** Round 8 / v0.1  
**Document type:** Formal research note / additive-support frontier  
**Status statement:** This round does not claim to prove the Four Exponentials Conjecture. Its objective is to fully abstract the exponent-support restrictions of Round 7, identify the structures that Six/Five Exponentials can and cannot exclude, and pinpoint the next non-combinatorial obstruction that must be introduced.

---

## Abstract

Under hypothetical Type-P + quartic descent, ACTM Round 7 normalises the problem to:

$$
\lambda,
\quad
s\lambda,
\quad
s^2\lambda,
\quad
s^4\lambda
\in
\mathcal L_{\mathrm{alg}},
$$

whereas:

$$
s^3\lambda,
\quad
s^5\lambda,
\quad
s^6\lambda
\notin
\mathcal L_{\mathrm{alg}}.
$$

Define the monomial descent exponent set:

$$
E_{\lambda,s}
=
\left\{
n\in\mathbb Z_{\ge0}:
s^n\lambda
\in
\mathcal L_{\mathrm{alg}}
\right\}.
$$

Round 7 has already obtained from the Six Exponentials Theorem:

$$
r_E(d)
=
\#\left\{
n\ge0:
n,n+d\in E
\right\}
\le2
$$

for all:

$$
d\ge1.
$$

The Five Exponentials Theorem gives:

$$
r_E(d)=2
\Rightarrow
e^{\gamma s^d}
\text{ transcendental}
$$

For all:

$$
0\neq\gamma
\in
\overline{\mathbb Q}
$$

the displayed implication holds.

Round 8 organises these conditions into a "difference-saturation graph".

For finite:

$$
S\subseteq E,
$$

define the saturated difference set:

$$
D_2(S)
=
\left\{
d>0:
r_S(d)=2
\right\}.
$$

If we add just one exponent on the right:

$$
m>\max S,
$$

then the exact condition for preserving Six-Exponential admissibility is:

$$
\boxed{
m-x
\notin
D_2(S)
\qquad
\forall x\in S.
}
$$

Thus saturated differences produce a forbidden halo on the number line:

$$
\boxed{
\mathcal H^+(S)
=
S+D_2(S).
}
$$

A new exponent cannot lie in:

$$
\mathcal H^+(S).
$$

For the quartic base:

$$
S_0
=
\{0,1,2,4\},
$$

we have:

$$
D_2(S_0)
=
\{1,2\}.
$$

Thus:

$$
5,6
$$

are immediately forbidden, while:

$$
7
$$

is the first combinatorially admissible extension on the right.

This round further identifies a key structure: any three-term arithmetic progression:

$$
a,
\quad
a+d,
\quad
a+2d
\in E
$$

forces:

$$
r_E(d)=2.
$$

Thus it has two consequences simultaneously:

1. No third gap-$d$ pair can appear;
2. Five Exponentials forces:
   $$
   e^{\gamma s^d}
   $$
   to be transcendental for every nonzero algebraic $\gamma$.

Thus a three-term AP is an "allowed but saturated" critical motif, while a four-term AP is entirely forbidden.

The quartic support:

$$
\{0,1,2,4\}
$$

consists precisely of two overlapping three-term APs:

$$
\{0,1,2\},
\qquad
\{0,2,4\}.
$$

They respectively saturate the gaps:

$$
1,
\qquad
2.
$$

Most importantly, this round proves that the purely difference-set conditions of Six/Five Exponentials cannot by themselves exclude all infinite descent-support models.

Define the binary tower:

$$
\boxed{
B_\infty
=
\{0\}
\cup
\{2^k:k\ge0\}
=
\{0,1,2,4,8,16,\ldots\}.
}
$$

Then for every:

$$
d>0,
$$

we have:

$$
\boxed{
r_{B_\infty}(d)\le2.
}
$$

More precisely:

$$
\boxed{
r_{B_\infty}(d)=2
\iff
d=2^k
\text{ for some }k\ge0.
}
$$

Thus the binary tower fully satisfies the Six-Exponential difference bound, and each power-of-two gap is exactly Five-Exponentials saturated.

If a hypothetical descent set contains the whole of $B_\infty$, it forms a nested Type-P chain for each $k$:

$$
\boxed{
\lambda,
\quad
s^{2^k}\lambda,
\quad
s^{2^{k+1}}\lambda
\in
\mathcal L_{\mathrm{alg}}.
}
$$

Six Exponentials only excludes:

$$
s^{3\cdot2^k}\lambda.
$$

These exponents:

$$
3\cdot2^k
$$

are not in the binary tower to begin with.

Thus the binary tower is an **infinite survivor model for pure Six/Five Exponentials combinatorics**.

This round thereby reaches an important methodological conclusion:

$$
\boxed{
\text{difference sparsity alone cannot close Type P}.
}
$$

The genuinely missing bridge must use:

$$
\boxed{
\text{common }\lambda
+
\text{nested squaring}
+
\text{projective / period / height structure}
}
$$

rather than merely continuing to accumulate forbidden difference-set regions.

---

# 1. Initial Setup

Fix:

$$
0\neq
\lambda
\in
\mathcal L_{\mathrm{alg}},
$$

and a transcendental:

$$
s.
$$

Define:

$$
\boxed{
E
=
E_{\lambda,s}
=
\left\{
n\ge0:
s^n\lambda
\in
\mathcal L_{\mathrm{alg}}
\right\}.
}
$$

where:

$$
\mathcal L_{\mathrm{alg}}
=
\exp^{-1}
\left(
\overline{\mathbb Q}^{\times}
\right).
$$

---

# 2. Difference representation function

For:

$$
d\ge1,
$$

define:

$$
\boxed{
r_E(d)
=
\#\left\{
n\ge0:
n,n+d\in E
\right\}.
}
$$

This counts how many times the difference:

$$
d
$$

occurs in $E$.

---

# 3. Six Exponentials bound

Round 7 has proved:

$$
\boxed{
r_E(d)\le2
}
$$

for all:

$$
d\ge1.
$$

Its proof only requires supposing that there are three distinct:

$$
n_1,n_2,n_3
$$

such that:

$$
n_j,
\quad
n_j+d
\in E,
$$

and then taking:

$$
x_1=1,
\qquad
x_2=s^d,
$$

and:

$$
y_j=s^{n_j}\lambda.
$$

The six:

$$
e^{x_i y_j}
$$

would all be algebraic, violating Six Exponentials.

---

# 4. Five Exponentials saturation certificate

If:

$$
r_E(d)=2,
$$

take two distinct starting points:

$$
m,n
$$

such that:

$$
m,m+d,n,n+d
\in E.
$$

Five Exponentials gives:

$$
\boxed{
e^{\gamma s^d}
\text{ transcendental}
}
$$

For all:

$$
0\neq\gamma
\in
\overline{\mathbb Q}
$$

the displayed transcendence holds.

Thus:

$$
r_E(d)=2
$$

is not ordinary combinatorial saturation.

It comes with a transcendence certificate.

---

# 5. Saturated difference set

For finite:

$$
S\subseteq E,
$$

define:

$$
\boxed{
D_2(S)
=
\left\{
d>0:
r_S(d)=2
\right\}.
}
$$

If:

$$
d\in D_2(S),
$$

then the gap:

$$
d
$$

has already reached the maximum multiplicity allowed by Six Exponentials.

Thus no added exponent can create a third gap-$d$ pair.

---

# 6. Right-extension criterion

Let:

$$
m>\max S.
$$

After adding:

$$
m
$$

the new differences formed with the old points:

$$
x\in S
$$

are:

$$
m-x.
$$

Since:

$$
m>\max S,
$$

different $x$ yield different differences.

Thus adding $m$ keeps every difference multiplicity at most $2$ if and only if:

$$
m-x
\notin
D_2(S)
$$

For all:

$$
x\in S.
$$

this condition holds, that is:

$$
\boxed{
m
\text{ admissible}
\iff
m\notin
S+D_2(S).
}
$$

---

# 7. Forward Saturation Halo

Define:

$$
\boxed{
\mathcal H^+(S)
=
S+D_2(S)
=
\left\{
x+d:
x\in S,
\ d\in D_2(S)
\right\}.
}
$$

and call it the:

**Forward Saturation Halo**.

For:

$$
m>\max S,
$$

we have:

$$
\boxed{
m
\text{ combinatorially admissible}
\iff
m\notin
\mathcal H^+(S).
}
$$

---

# 8. Quartic base halo

Take:

$$
S_0
=
\{0,1,2,4\}.
$$

Its positive differences are:

$$
1:
\quad
0\to1,
\quad
1\to2;
$$

$$
2:
\quad
0\to2,
\quad
2\to4;
$$

$$
3:
\quad
1\to4;
$$

$$
4:
\quad
0\to4.
$$

Thus:

$$
\boxed{
D_2(S_0)
=
\{1,2\}.
}
$$

---

# 9. First right extensions

Since:

$$
D_2(S_0)=\{1,2\},
$$

for:

$$
m>4,
$$

if:

$$
m-4\in\{1,2\},
$$

that is:

$$
m=5
\quad\text{or}\quad
m=6,
$$

then it is forbidden.

Thus:

$$
\boxed{
5,6
\text{ forbidden}.
}
$$

Whereas:

$$
7
$$

is the first integer outside the forward halo.

Thus:

$$
\boxed{
7
\text{ is the first combinatorially admissible right extension}.
}
$$

---

# 10. But $7$ Is Not the Only Admissible Extension

We must distinguish:

> first admissible

from:

> uniquely admissible.

For the base:

$$
S_0=\{0,1,2,4\},
$$

all:

$$
m\ge7
$$

are admissible under the difference-multiplicity test for "adding just one point".

Thus the:

$$
7
$$

of Round 7 is only the first position not immediately excluded, not the only candidate.

---

# 11. Adding $7$

Let:

$$
S_1
=
\{0,1,2,4,7\}.
$$

The added difference is:

$$
7-4=3.
$$

The gap $3$ already had:

$$
1\to4.
$$

so:

$$
4\to7
$$

makes:

$$
r_{S_1}(3)=2.
$$

Thus:

$$
\boxed{
D_2(S_1)
=
\{1,2,3\}.
}
$$

---

# 12. The Escape Certificate of $7$

Since:

$$
r_{S_1}(3)=2,
$$

Five Exponentials immediately gives:

$$
\boxed{
e^{\gamma s^3}
\text{ transcendental}
}
$$

for all:

$$
0\neq\gamma
\in
\overline{\mathbb Q}.
$$

This is consistent with:

$$
3\notin E
$$

but stronger.

---

# 13. The New Halo After $7$

Since:

$$
D_2(S_1)
=
\{1,2,3\},
$$

any new exponent on the right:

$$
m>7
$$

cannot differ from an existing element by:

$$
1,2,3.
$$

In particular:

$$
\boxed{
8,9,10
\text{ forbidden}.
}
$$

Thus, if $7$ is selected, the next possible position is at least:

$$
11.
$$

---

# 14. Alternative extension $8$

If, starting from:

$$
S_0
$$

we add:

$$
8,
$$

we obtain:

$$
S'_1
=
\{0,1,2,4,8\}.
$$

The gap $4$ has:

$$
0\to4,
$$

and:

$$
4\to8.
$$

so:

$$
\boxed{
D_2(S'_1)
=
\{1,2,4\}.
}
$$

This begins precisely the powers-of-two pattern.

---

# 15. Three-term arithmetic progression

Suppose:

$$
a,
\quad
a+d,
\quad
a+2d
\in E.
$$

Then the gap:

$$
d
$$

is represented by at least:

$$
a\to a+d
$$

and:

$$
a+d\to a+2d
$$

twice.

By the Six bound:

$$
r_E(d)\le2,
$$

hence:

$$
\boxed{
r_E(d)=2.
}
$$

---

# 16. Three-term AP certificate

By Five Exponentials saturation:

$$
r_E(d)=2
$$

we immediately obtain:

$$
\boxed{
e^{\gamma s^d}
\text{ transcendental}
}
$$

for all:

$$
0\neq\gamma
\in
\overline{\mathbb Q}.
$$

Thus:

$$
\boxed{
\text{3-term AP in }E
\Rightarrow
\text{full algebraic-scalar escape at its step}.
}
$$

---

# 17. Four-term arithmetic progression is impossible

If:

$$
a,
\quad
a+d,
\quad
a+2d,
\quad
a+3d
\in E,
$$

then the gap:

$$
d
$$

has at least three representations:

$$
a\to a+d,
$$

$$
a+d\to a+2d,
$$

$$
a+2d\to a+3d.
$$

so:

$$
r_E(d)\ge3,
$$

contradicting Six Exponentials.

Thus:

$$
\boxed{
E
\text{ contains no nontrivial 4-term arithmetic progression}.
}
$$

---

# 18. Critical motif hierarchy

For a fixed:

$$
d>0,
$$

the local additive motifs can be classified as follows:

### One pair

$$
r_E(d)=1.
$$

Allowed, unsaturated.

### Two pairs

$$
r_E(d)=2.
$$

Allowed, but Five Exponentials produces a transcendence certificate.

### Three pairs

$$
r_E(d)\ge3.
$$

Forbidden by Six Exponentials.

Thus:

$$
\boxed{
1
\rightarrow
2
\rightarrow
3
}
$$

corresponds to:

$$
\boxed{
\text{unsaturated}
\rightarrow
\text{critical}
\rightarrow
\text{forbidden}.
}
$$

---

# 19. Quartic support as two nested 3-APs

The quartic base:

$$
S_0
=
\{0,1,2,4\}
$$

contains:

$$
\boxed{
0,1,2
}
$$

the three-term AP of step $1$,

and:

$$
\boxed{
0,2,4
}
$$

the three-term AP of step $2$.

Thus the quartic support is:

$$
\boxed{
\text{two nested critical 3-APs}.
}
$$

---

# 20. Double saturation

The two APs give:

$$
r_E(1)=2,
$$

and:

$$
r_E(2)=2.
$$

Thus:

$$
\boxed{
d=1
\quad\text{and}\quad
d=2
}
$$

are saturated simultaneously.

Five Exponentials consequently implies:

$$
e^{\gamma s}
$$

and:

$$
e^{\gamma s^2}
$$

For all:

$$
0\neq\gamma\in\overline{\mathbb Q}
$$

both expressions are transcendental.

---

# 21. The General Nested Type-P Motif

If:

$$
0,
\quad
d,
\quad
2d
\in E,
$$

then:

$$
\lambda,
\quad
s^d\lambda,
\quad
s^{2d}\lambda
\in
\mathcal L_{\mathrm{alg}}.
$$

Put:

$$
u=s^d.
$$

We obtain:

$$
\boxed{
\lambda,
\quad
u\lambda,
\quad
u^2\lambda
\in
\mathcal L_{\mathrm{alg}}.
}
$$

This is precisely a Type-P chain.

Thus:

$$
\boxed{
\text{3-term AP in exponent support}
\iff
\text{nested Type-P chain}.
}
$$

---

# 22. Nested Type-P transcendence degree

If:

$$
u=s^d,
$$

where:

$$
d\ge1,
$$

then $s$, over:

$$
\mathbb Q(u)
$$

is algebraic, since:

$$
X^d-u=0.
$$

Thus:

$$
\mathbb Q(\lambda,s)
$$

extends:

$$
\mathbb Q(\lambda,u)
$$

algebraically.

Therefore:

$$
\boxed{
\operatorname{trdeg}_{\mathbb Q}
\mathbb Q(\lambda,u)
=
\operatorname{trdeg}_{\mathbb Q}
\mathbb Q(\lambda,s)
=
2.
}
$$

Every nested Type-P chain still lies exactly in the transcendence-degree-$2$ critical layer.

---

# 23. Binary Tower

Now define:

$$
\boxed{
B_\infty
=
\{0\}
\cup
\{2^k:k\ge0\}.
}
$$

That is:

$$
B_\infty
=
\{0,1,2,4,8,16,32,\ldots\}.
$$

This support naturally extends the quartic base:

$$
\{0,1,2,4\}.
$$

---

# 24. Proposition R8.1: The Binary Tower Satisfies the Six Difference Bound

For all:

$$
d>0,
$$

we have:

$$
\boxed{
r_{B_\infty}(d)\le2.
}
$$

---

# 25. Proof of R8.1: Positive-Power Pairs

Consider:

$$
2^j-2^i,
\qquad
j>i\ge0.
$$

It can be written as:

$$
2^i
\left(
2^{j-i}-1
\right).
$$

The factor in parentheses is odd.

Thus:

$$
\boxed{
v_2
\left(
2^j-2^i
\right)
=i.
}
$$

Consequently, given the difference:

$$
d,
$$

if two positive powers represent it as:

$$
2^j-2^i=d
$$

then:

$$
i=v_2(d)
$$

is uniquely determined.

Next:

$$
2^{j-i}-1
=
\frac{d}{2^i}
$$

uniquely determines:

$$
j-i.
$$

Thus positive-power pairs supply at most one representation.

---

# 26. Zero-based representation

If:

$$
d=2^k,
$$

there is also the representation:

$$
0\to2^k.
$$

Now ask whether a positive-power pair can produce the same difference:

$$
2^j-2^i=2^k.
$$

From:

$$
2^i(2^{j-i}-1)=2^k,
$$

the odd factor:

$$
2^{j-i}-1
$$

must equal:

$$
1.
$$

so:

$$
j-i=1,
$$

and:

$$
i=k.
$$

We obtain the unique second representation:

$$
2^k\to2^{k+1}.
$$

Thus:

$$
\boxed{
r_{B_\infty}(2^k)=2.
}
$$

---

# 27. Non-power-of-two differences

If:

$$
d
$$

is not a power of two, then there is no binary-tower pair:

$$
0\to d
$$

of the form displayed above.

There is at most one positive-power pair as well.

Thus:

$$
\boxed{
d\neq2^k
\Rightarrow
r_{B_\infty}(d)\le1.
}
$$

Combining these gives:

$$
\boxed{
r_{B_\infty}(d)
=
2
\iff
d=2^k.
}
$$

---

# 28. The Binary Tower Is an Infinite Combinatorial Survivor

Thus:

$$
B_\infty
$$

fully satisfies:

$$
r(d)\le2.
$$

It contains no 4-term AP and produces no difference multiplicity forbidden by Six Exponentials.

Thus:

$$
\boxed{
\text{Six difference bound alone permits an infinite extension of quartic support}.
}
$$

This is the core negative result of Round 8.

---

# 29. The Five Certificates of the Binary Tower

Every:

$$
2^k
$$

is a saturated difference:

$$
r_{B_\infty}(2^k)=2.
$$

Thus if:

$$
B_\infty
\subseteq
E_{\lambda,s},
$$

Five Exponentials gives, for all:

$$
k\ge0
$$

the certificate:

$$
\boxed{
e^{\gamma s^{2^k}}
\text{ transcendental}
}
$$

for all:

$$
0\neq\gamma
\in
\overline{\mathbb Q}.
$$

---

# 30. But the Five Certificates Do Not Conflict with Weighted Descent

Binary-tower descent assumes:

$$
\boxed{
s^{2^k}\lambda
\in
\mathcal L_{\mathrm{alg}}.
}
$$

That is:

$$
e^{s^{2^k}\lambda}
$$

is algebraic.

The Five certificate instead says:

$$
e^{\gamma s^{2^k}}
$$

For algebraic:

$$
\gamma\neq0
$$

the exponential is transcendental.

The two do not directly conflict, because:

$$
\lambda
$$

is not itself an algebraic scalar; it is an algebraic logarithm.

Thus:

$$
\boxed{
\text{scalar escape of }s^{2^k}
\neq
\text{weighted descent of }s^{2^k}\lambda.
}
$$

This precisely exposes the missing bridge in the current method.

---

# 31. Infinite nested Type-P tower

If:

$$
B_\infty
\subseteq E,
$$

then for every:

$$
k\ge0,
$$

we have:

$$
0,
\quad
2^k,
\quad
2^{k+1}
\in E.
$$

Thus:

$$
\boxed{
\lambda,
\quad
s^{2^k}\lambda,
\quad
s^{2^{k+1}}\lambda
\in
\mathcal L_{\mathrm{alg}}.
}
$$

Put:

$$
u_k
=
s^{2^k}.
$$

Then:

$$
\boxed{
u_{k+1}=u_k^2.
}
$$

At every layer:

$$
\boxed{
\lambda,
\quad
u_k\lambda,
\quad
u_k^2\lambda
\in
\mathcal L_{\mathrm{alg}}.
}
$$

and together these form an infinite nested Type-P critical tower.

---

# 32. At Each Layer, Six Exponentials Excludes Only the Cubic Child

For every:

$$
u_k,
$$

Six Exponentials forces:

$$
u_k^3\lambda
\notin
\mathcal L_{\mathrm{alg}}.
$$

That is:

$$
\boxed{
s^{3\cdot2^k}\lambda
\notin
\mathcal L_{\mathrm{alg}}.
}
$$

But:

$$
3\cdot2^k
$$

is never a power of two.

Thus:

$$
3\cdot2^k
\notin
B_\infty.
$$

So the binary tower avoids all cubic children exactly.

---

# 33. The Binary Tower as an Obstruction Witness

Thus:

$$
B_\infty
$$

is not a known actual descent set.

It is a:

$$
\boxed{
\text{combinatorial consistency witness}.
}
$$

Its significance is:

> Any proof strategy using only
>
> $$
> r_E(d)\le2
> $$
>
> and
>
> $$
> r_E(d)=2
> \Rightarrow
> \text{Five certificate}
> $$
>
> cannot exclude an infinite nested Type-P pattern solely through exponent-support combinatorics.

---

# 34. Methodological stop condition

Round 8 therefore establishes a stop condition:

$$
\boxed{
\text{Do not keep iterating difference halos as the sole proof strategy.}
}
$$

because such iteration ultimately permits at least one explicit infinite survivor skeleton:

$$
B_\infty.
$$

If the objective is to attack Four Exponentials / Type P, additional information must be introduced.

---

# 35. The Missing Bridge

The binary tower tells us that the next theorem must connect two kinds of information that are currently separate.

The first:

$$
\boxed{
e^{\gamma u}
\text{ transcendental for all }
0\neq\gamma\in\overline{\mathbb Q}
}
$$

where:

$$
u=s^{2^k}.
$$

The second:

$$
\boxed{
u\lambda
\in
\mathcal L_{\mathrm{alg}}.
}
$$

That is:

$$
e^{u\lambda}
\in
\overline{\mathbb Q}.
$$

What is genuinely missing is some kind of:

$$
\boxed{
\text{scalar-escape}
+
\text{weighted-descent}
\Rightarrow
\text{joint obstruction}.
}
$$

---

# 36. Weighted Descent Coupling Problem

This round formally proposes the next research interface:

Given:

$$
0\neq
\lambda
\in
\mathcal L_{\mathrm{alg}},
$$

and a transcendental:

$$
u
$$

satisfying:

$$
u\lambda
\in
\mathcal L_{\mathrm{alg}},
$$

and:

$$
e^{\gamma u}
$$

For all:

$$
0\neq\gamma\in\overline{\mathbb Q}
$$

the latter exponential is transcendental.

Ask:

> What additional restrictions do these two conditions together impose on the descent of
>
> $$
> u^2\lambda,
> \quad
> u^4\lambda,
> \quad
> \ldots
> $$
>
> the terms displayed above?

This round calls this the:

**Weighted Descent Coupling Problem (WDCP)**.

---

# 37. Why WDCP Is Stronger than Five Exponentials

Five Exponentials considers only:

$$
u
$$

as a ratio:

$$
x_2/x_1.
$$

WDCP instead requires using simultaneously:

$$
u
$$

and the fixed algebraic logarithm:

$$
\lambda.
$$

That is, it studies:

$$
\boxed{
(u,\lambda,u\lambda)
}
$$

the joint structure of these three quantities.

This goes beyond pure exponent-index combinatorics.

---

# 38. Common-$\lambda$ coupling

All nested Type-P chains of the binary tower share the same:

$$
\lambda.
$$

This is information completely lost in the difference-set abstraction.

If each layer allows a different:

$$
\lambda_k,
$$

the purely combinatorial model is even freer.

But an actual Type-P tower requires:

$$
\boxed{
\lambda_k=\lambda
\qquad
\forall k.
}
$$

Thus:

$$
\boxed{
\text{common-anchor rigidity}
}
$$

should become the first research direction of Round 9.

---

# 39. Squaring coupling

The binary tower also has:

$$
u_{k+1}=u_k^2.
$$

Thus different Type-P layers are not independent.

They satisfy:

$$
\boxed{
u_k
\longmapsto
u_k^2
}
$$

the deterministic recursion displayed above.

Each layer also requires:

$$
\lambda,
\quad
u_k\lambda,
\quad
u_k^2\lambda
\in
\mathcal L_{\mathrm{alg}}.
$$

Thus the middle term of the next layer:

$$
u_{k+1}\lambda
$$

is exactly the third term of the preceding layer.

This is:

$$
\boxed{
\text{overlapping critical chains}.
}
$$

---

# 40. Overlap graph

The binary tower can be drawn as:

$$
\lambda
\rightarrow
u_0\lambda
\rightarrow
u_1\lambda
\rightarrow
u_2\lambda
\rightarrow
\cdots
$$

where:

$$
u_{k+1}=u_k^2.
$$

and each segment:

$$
\lambda,
\quad
u_k\lambda,
\quad
u_{k+1}\lambda
$$

is simultaneously a Type-P triple.

Note that the first node:

$$
\lambda
$$

is shared by all triples, while:

$$
u_{k+1}\lambda
$$

continuously supplies the next layer's data.

Thus this is not a collection of independent Four-Exponential counterexamples.

It is a coupled tower.

---

# 41. Transcendence degree across the tower

For every:

$$
k,
$$

$$
u_k=s^{2^k}.
$$

Since $s$, over:

$$
\mathbb Q(u_k)
$$

is algebraic, we have:

$$
\boxed{
\operatorname{trdeg}_{\mathbb Q}
\mathbb Q(\lambda,u_k)=2.
}
$$

Thus, however deep the tower becomes, each finite stage still does not automatically increase the transcendence degree.

This explains why simply having "many more transcendental numbers" does not directly produce a dimensional contradiction.

---

# 42. Projective avoidance across the tower

The PLA of Round 4 holds for every Type-P ratio.

Thus, if the binary tower truly descends, then for every:

$$
k\ge0
$$

and:

$$
g\in
\operatorname{PGL}_2(\mathbb Q),
$$

we have:

$$
\boxed{
e^{g(u_k)}
\text{ transcendental}.
}
$$

Thus a hypothetical infinite tower must simultaneously avoid infinitely many rational projective logarithmic loci.

This is a larger constraint family than the Five certificate alone.

---

# 43. But PLA Still Does Not Automatically Produce a Contradiction

Since:

$$
u_{k+1}=u_k^2
$$

is not a:

$$
\operatorname{PGL}_2(\mathbb Q)
$$

transformation.

Thus:

$$
u_k
\mapsto
u_k^2
$$

can continue generating the next layer while avoiding all rational Möbius transforms.

This again indicates:

$$
\boxed{
\text{squaring is the missing nonlinear coupling}.
}
$$

---

# 44. Arithmetic progression viewpoint

For every:

$$
k
$$

in the binary tower, there is an exponent-index AP:

$$
0,
\quad
2^k,
\quad
2^{k+1}.
$$

But one cannot add:

$$
3\cdot2^k,
$$

because that would form:

$$
0,
\quad
2^k,
\quad
2^{k+1},
\quad
3\cdot2^k
$$

a four-term AP.

Thus:

$$
\boxed{
\text{each nested Type-P triple is maximal in its own additive direction}.
}
$$

This is a multiscale maximality structure.

---

# 45. Scale separation

Different $k$ correspond to different steps:

$$
d_k=2^k.
$$

Every:

$$
d_k
$$

is saturated:

$$
r_E(d_k)=2.
$$

Thus the binary tower can be viewed as:

$$
\boxed{
\text{infinitely many saturated scales}
}
$$

but these scales are arranged by the doubling law:

$$
d_{k+1}=2d_k
$$

displayed above.

This may have more arithmetic rigidity than a general $B_2[2]$-type set.

---

# 46. Relation to Generalized Sidon Sets

In classical additive combinatorics, $B_2[g]$ sets are usually defined by "bounded representation multiplicity for sums of two terms".

ACTM's condition:

$$
r_E(d)\le2
$$

is "bounded representation multiplicity for positive differences".

The two are closely related through:

$$
a-b=c-d
\iff
a+d=b+c
$$

the displayed equivalence, but they are not identical definitions.

Thus this series adopts the more conservative terminology:

$$
\boxed{
\text{generalized-Sidon-type difference condition}.
}
$$

and does not directly identify $E$ with the standard $B_2[2]$ definition.

---

# 47. Density bound revisited

From:

$$
r_E(d)\le2
$$

we obtain the crude counting bound:

$$
\binom{|E\cap[0,N]|}{2}
\le
2N.
$$

Thus:

$$
\boxed{
|E\cap[0,N]|
\le
\frac{
1+\sqrt{1+16N}
}{2}.
}
$$

The binary tower, however, has only:

$$
\boxed{
|B_\infty\cap[0,N]|
=
O(\log N).
}
$$

Thus the binary tower is far sparser than the limiting density required by the Six bound.

This is also one reason it can readily survive.

---

# 48. The Actual Power and Limits of the Difference Bound

Six/Five Exponentials can:

1. Forbid multiplicity $3$;
2. Produce a scalar-escape certificate at multiplicity $2$;
3. Imply a global square-root-type density bound;
4. Forbid 4-term APs;
5. Pin down the step of every 3-term AP.

But alone they cannot:

1. Forbid every infinite $E$;
2. Forbid the binary tower;
3. Turn the transcendence of $e^{\gamma u}$ and the algebraicity of $e^{u\lambda}$ into a conflict;
4. Use the information that all layers share the same $\lambda$;
5. Use the nonlinear recursion $u_{k+1}=u_k^2$.

This is the methodological boundary of Round 8.

---

# 49. The Core Negative Result of Round 8

The most important conclusion of this round is not the exclusion of a few more exponents.

Rather, it is:

$$
\boxed{
\text{Difference-set combinatorics is insufficient as a complete proof engine.}
}
$$

because we have an explicit infinite combinatorial survivor:

$$
B_\infty.
$$

This lets subsequent research stop repeatedly stacking Six/Five Exponentials in the wrong direction.

---

# 50. The Next Structure That Is Genuinely Needed

Round 9 should reconnect:

$$
E
$$

to the continuous / transcendental data abstracted away from it:

$$
\boxed{
(\lambda,s,E).
}
$$

rather than studying only:

$$
E.
$$

At least the following must be retained together:

$$
\boxed{
\lambda
}
$$

as the common anchor,

$$
\boxed{
s
}
$$

as the multiplicative generator,

and:

$$
\boxed{
n\mapsto s^n\lambda
}
$$

as the embedding.

---

# 51. Anchored Descent System

This round proposes the next research object:

$$
\boxed{
\mathfrak A(\lambda,s)
=
\left(
\lambda,
s,
E_{\lambda,s},
\mathcal L_{\mathrm{alg}}
\right).
}
$$

called the:

**Anchored Descent System (ADS)**.

Unlike a mere exponent set, ADS retains:

1. the common anchor $\lambda$;
2. the multiplicative generator $s$;
3. the exponent support $E$;
4. the embedding $n\mapsto s^n\lambda$;
5. the algebraic-exponential descent condition.

---

# 52. Binary ADS

If:

$$
B_\infty
\subseteq
E_{\lambda,s},
$$

we say:

$$
\mathfrak A(\lambda,s)
$$

contains a:

**Binary Anchored Descent Tower (BADT)**.

Its core condition is:

$$
\boxed{
s^{2^k}\lambda
\in
\mathcal L_{\mathrm{alg}}
\qquad
\forall k\ge0.
}
$$

That is:

$$
\boxed{
e^{s^{2^k}\lambda}
\in
\overline{\mathbb Q}^{\times}
\qquad
\forall k\ge0.
}
$$

---

# 53. The Genuine Problem of BADT

Six/Five Exponentials alone cannot exclude BADT.

Thus the question of Round 9 is:

> Is there an additional transcendence / height / period theorem that can exclude a fixed algebraic logarithm $\lambda$ for which:
>
> $$
> e^{s\lambda},
> \quad
> e^{s^2\lambda},
> \quad
> e^{s^4\lambda},
> \quad
> e^{s^8\lambda},
> \ldots
> $$
>
> are all algebraic?

This target is narrower than the original statement of Four Exponentials, but more structured than a single quartic descent.

---

# 54. The Finite BADT Version

Since the infinite version may be too strong, first define finite depth:

$$
\boxed{
\operatorname{BADT}_m:
\quad
s^{2^k}\lambda
\in
\mathcal L_{\mathrm{alg}}
\quad
0\le k\le m.
}
$$

Type P already gives the:

$$
m=1
$$

layer:

$$
\lambda,
\quad
s\lambda,
\quad
s^2\lambda.
$$

Quartic descent is equivalent to adding the:

$$
m=2
$$

layer:

$$
s^4\lambda.
$$

Thus:

$$
\boxed{
\text{quartic descent}
=
\operatorname{BADT}_2.
}
$$

---

# 55. BADT depth as a new frontier parameter

Define:

$$
\boxed{
\operatorname{BD}(\lambda,s)
=
\sup
\left\{
m:
s^{2^k}\lambda
\in
\mathcal L_{\mathrm{alg}}
\text{ for }
0\le k\le m
\right\}.
}
$$

Type P guarantees:

$$
\operatorname{BD}\ge1.
$$

Quartic descent assumes:

$$
\operatorname{BD}\ge2.
$$

Ultimately, Type-P elimination for Four Exponentials only requires proving:

$$
\boxed{
\operatorname{BD}=1
}
$$

for every Type-P state.

---

# 56. Why BD Is More Suitable than Ordinary Power-Chain Depth

The ordinary depth of Round 4:

$$
D_\lambda(s)
$$

proceeds along:

$$
1,s,s^2,s^3,\ldots
$$

and Six Exponentials has already given exactly:

$$
D_\lambda(s)=2.
$$

Thus the ordinary chain is already dead at the cubic stage.

But quartic descent shows that the genuinely possible way around the cubic wall is:

$$
s
\mapsto
s^2
\mapsto
s^4
\mapsto
s^8
\mapsto
\cdots.
$$

Thus:

$$
\boxed{
\text{doubling depth}
}
$$

is closer to the higher-support survivor geometry than ordinary consecutive-power depth.

---

# 57. Round 8 Assessment

**PASS — Exponent support reduced to a difference-saturation system, and pure Six/Five combinatorics proven insufficient by an infinite binary survivor model**

This round completes:

1. Definition of the saturated difference set:
   $$
   D_2(S).
   $$

2. Construction of the exact right-extension criterion:
   $$
   m\notin S+D_2(S).
   $$

3. Definition of the Forward Saturation Halo.

4. A proof for the quartic base that:
   $$
   D_2(\{0,1,2,4\})=\{1,2\}.
   $$

5. An exact explanation of:
   $$
   5,6
   $$
   why these are forbidden while $7$ is the first admissible right extension.

6. A proof that every 3-term AP automatically saturates its common difference.

7. A proof that every 4-term AP is excluded by Six Exponentials.

8. A proof that every 3-term AP produces a Five-Exponentials scalar-escape certificate.

9. Identification of the quartic base as two nested critical 3-APs:
   $$
   \{0,1,2\},
   \qquad
   \{0,2,4\}.
   $$

10. Definition of the binary tower:
    $$
    B_\infty
    =
    \{0,1,2,4,8,\ldots\}.
    $$

11. A proof that:
    $$
    r_{B_\infty}(d)\le2
    $$
    for all $d$.

12. An exact proof that:
    $$
    r_{B_\infty}(d)=2
    \iff
    d=2^k.
    $$

13. A proof that the binary tower is an infinite combinatorial survivor of the Six difference bound.

14. A demonstration that the Five certificates do not directly conflict with weighted algebraic descent.

15. Reformulation of the binary tower as an infinite nested Type-P tower.

16. A proof that every nested layer still has:
    $$
    \operatorname{trdeg}=2.
    $$

17. A demonstration that Six Exponentials excludes only:
    $$
    3\cdot2^k
    $$
    cubic children of the displayed type, which the binary tower automatically avoids.

18. The methodological stop condition:
    **Difference-halo iteration must no longer be treated as the sole proof strategy.**

19. Definition of the Weighted Descent Coupling Problem (WDCP).

20. Definition of the Anchored Descent System (ADS).

21. Definition of the Binary Anchored Descent Tower (BADT).

22. Definition of binary depth:
    $$
    \operatorname{BD}(\lambda,s).
    $$

23. Repositioning quartic descent as:
    $$
    \operatorname{BADT}_2.
    $$

24. Reformulation of the narrow target for Type-P elimination as:
    $$
    \operatorname{BD}(\lambda,s)=1.
    $$

---

# 58. The Formal Entry Point for Round 9

Round 9 no longer prioritises enumeration of:

$$
7,8,9,\ldots
$$

but directly studies:

$$
\boxed{
\operatorname{BADT}_2
}
$$

itself:

$$
\lambda,
\quad
s\lambda,
\quad
s^2\lambda,
\quad
s^4\lambda
\in
\mathcal L_{\mathrm{alg}}.
$$

Priority questions:

## R9-Q1: The Common Anchor

How can one use the fact that the four algebraic logarithms all share the same:

$$
\lambda
$$

as displayed above?

The difference-set model cannot see this information.

---

## R9-Q2: Squaring Coupling

How can one use:

$$
u_1=s,
\qquad
u_2=s^2,
\qquad
u_3=s^4
$$

and the:

$$
u_{k+1}=u_k^2
$$

nonlinear relation among them?

---

## R9-Q3: Joint Period Structure

Can the four algebraic exponential values:

$$
e^\lambda,
\quad
e^{s\lambda},
\quad
e^{s^2\lambda},
\quad
e^{s^4\lambda}
$$

be organised into some period / logarithmic matrix, allowing existing Baker, Waldschmidt–Masser, Matrix Coefficient, or period-conjecture tools to intervene?

---

## R9-Q4: Height Growth

If the four values above are all algebraic, put:

$$
\alpha_k
=
e^{s^{2^k}\lambda}.
$$

Can one establish an incompatible growth law between the algebraic heights:

$$
h(\alpha_k)
$$

and the squaring recursion of the logarithmic lifts?

---

## R9-Q5: A Finite BADT Obstruction

Leave the infinite tower aside for now.

Ask only:

> Can one directly exclude:
>
> $$
> \operatorname{BADT}_2?
> $$

If so, quartic descent is already excluded.

This is the most focused and valuable target of the next round.

---

# 59. One-Sentence Conclusion

> **Round 8 shows that Six/Five Exponentials compresses Type-P exponent support into a generalized-Sidon-type difference system, but this restriction alone still permits the infinite binary tower $\{0,1,2,4,8,\ldots\}$; thus what is genuinely missing is not more forbidden difference-set regions, but a new obstruction coupling "weighted algebraic descent on the same $\lambda$" with "nonlinear nesting via $s\mapsto s^2$".**

The most important survivor:

$$
\boxed{
B_\infty
=
\{0,1,2,4,8,16,\ldots\}
}
$$

satisfies:

$$
\boxed{
r_{B_\infty}(d)\le2,
}
$$

and:

$$
\boxed{
r_{B_\infty}(d)=2
\iff
d=2^k.
}
$$

Thus the genuine target of Round 9 is no longer difference combinatorics, but:

$$
\boxed{
\text{Common Anchor}
+
\text{Weighted Descent}
+
\text{Squaring Coupling}.
}
$$

---

# References

1. Michel Waldschmidt, expository accounts of the **Five Exponentials Theorem** and **Six Exponentials Theorem**. The Five Exponentials form used here states that if $x_1,x_2$ and $y_1,y_2$ are respectively $\mathbb Q$-linearly independent and $0\neq\gamma\in\overline{\mathbb Q}$, then at least one of:
   $$
   e^{x_1y_1},
   e^{x_1y_2},
   e^{x_2y_1},
   e^{x_2y_2},
   e^{\gamma x_2/x_1}
   $$
   is transcendental.

2. J. Pila, **Geometric and arithmetic postulation of the exponential function**, *Journal of the Australian Mathematical Society* 86 (2009), 111–117. Provides real-variable proofs of classical transcendence results including the Six Exponentials theorem in the real case.

3. Samit Dasgupta and Mahesh Kakde, **Ranks of Matrices of Logarithms of Algebraic Numbers II: The Matrix Coefficient Conjecture**, *Advances in Mathematics* 487 (2026), 110753.  
   DOI: 10.1016/j.aim.2025.110753.

4. Javier Cilleruelo and Carlos Vinuesa, **$B_2[g]$ Sets and a Conjecture of Schinzel and Schmidt**, *Combinatorics, Probability and Computing* 17 (2008), 741–747. Standard generalized Sidon / $B_2[g]$ context. The ACTM difference condition is related but not identical to the standard sum-representation definition.

5. Greg Martin and Kevin O'Bryant, **Constructions of Generalized Sidon Sets**, *Journal of Combinatorial Theory, Series A* 113 (2006), 591–607. Generalized bounded-representation additive sets.

6. Michel Waldschmidt, survey and lecture material on algebraic values among sequences such as:
   $$
   2^{\pi^k},
   $$
   including square-root-type sparsity bounds originating in work of Srinivasan, Ramachandra, Balasubramanian and others.

---

**End of ACTM Round 8**