7 Exact first differences, least periods, and dyadic images
On this page · Permalink · Source line 959
The factorial estimates are stronger than needed for the optimal first-difference law. A general Newton criterion gives the exact law from the threshold-free bound and finitely many rational coefficients. This route also yields all addresses at ranks two and eight, and an infinite family at every larger dyadic rank.
Lemma 7.1: A first-difference criterion
Let \(C:\mathbb Z_{\geq0}\to\mathbb Z_2\) have Newton coefficients \(a_k=\Delta_1^kC(0)\). Suppose for an integer \(b\geq0\) that
Then for every \(n\geq0\) and \(H>0\),
The Newton interpolation extends continuously to \(\mathbb Z_2\) and satisfies the same identity for any two distinct dyadic arguments.
Read proof · Read statement and proof side by side · Permalink · Source line 968 · Claim record · Lean coverage · Code · Worked examples
Proof of Lemma 7.1
At a natural argument the exact Newton expansion is finite. Vandermonde gives
For \(1\leq i\leq H\), \(v_2(\binom Hi)\geq v_2(H)-v_2(i) \geq v_2(H)-\lfloor\log_2k\rfloor\); terms with \(i>H\) vanish. Thus each Newton summand with \(k\geq2\) has valuation at least \(b+v_2(H)+1\). The \(k=1\) summand is \(Ha_1\), of exact valuation \(b+v_2(H)\), proving (7.2).
The coefficient hypothesis implies \(a_k\to0\) dyadically, so the Newton series converges uniformly on \(\mathbb Z_2\), where \(\binom xk\) is integral-valued. For distinct \(x,y\in\mathbb Z_2\), approximate the pair by natural integer pairs with their difference valuation eventually equal to \(v_2(x-y)\). The corresponding function differences have the fixed exact valuation \(b+v_2(x-y)\). A limit of elements with that fixed valuation has the same valuation: the normalized residues are units, and \(\mathbb Z_2^\times\) is closed.
Back to statement · Read statement and proof side by side · Permalink · Source line 986 · Claim record · Lean coverage · Code · Worked examples
Theorem 7.2: Exact four-address first differences
For every \(n\geq0\), \(H>0\), and \(j\in\{0,1,2,3\}\),
Consequently, the least positive row period modulo \(2^r\) is exactly
The corresponding displacement in the original Euler subscript is \(4P_j(r)\).
Read proof · Visual guide: First differences determine an exact metric · Visual guide: Least periods, rather than sufficient periods · Read statement and proof side by side · Permalink · Source line 1010 · Claim record · Lean coverage · Code · Worked examples
Proof of Theorem 7.2
The actual reciprocal recurrence is
For \(N=4\), the four kernels are
The threshold-free theorem gives \(v_2(a_k)\geq k\) in every address. The following exact values cover all orders where that bound does not already imply (7.1):
| \(j\) | \(k\) | \(a_k\) | \(v_2(a_k)\) |
|---|---|---|---|
| 0 | 1 | \(-2\) | 1 |
| 0 | 2 | \(72\) | 3 |
| 1 | 1 | \(-6/5\) | 1 |
| 1 | 2 | \(184/45\) | 3 |
| 2 | 1 | \(-16/15\) | 4 |
| 2 | 2 | \(64/45\) | 6 |
| 2 | 3 | \(-14848/1365\) | 9 |
| 2 | 4 | \(71426048/69615\) | 13 |
| 2 | 5 | \(-25509691392/85085\) | 16 |
| 2 | 6 | \(149113198870528/765765\) | 22 |
| 3 | 1 | \(-36/35\) | 2 |
| 3 | 2 | \(256/231\) | 8 |
| 3 | 3 | \(-65792/35035\) | 8 |
| 3 | 4 | \(161644544/4849845\) | 15 |
They are finite rational evaluations of (7.6), followed by \(a_k=\sum_{h=0}^k(-1)^{k-h}\binom kh C_j(h)\). For addresses zero and one, \(k\geq3\) is covered by the generic bound; for address two, \(k\geq7\) is covered; for address three, \(k\geq5\) is covered. In each range the elementary inequality is \(k-\lfloor\log_2k\rfloor\geq b_j+1\). Lemma 7.1 proves the normalized identity. Multiplication by \(j!\) proves the original identity.
A row displacement \(H\) preserves the original coefficient modulo \(2^r\) at even one row exactly when \(v_2(H)+B_j\geq r\). The least positive such integer is \(2^{\max(0,r-B_j)}\), and the same displacement works at every row. This proves both optimality and the absence of a starting threshold.
Back to statement · Read statement and proof side by side · Permalink · Source line 1029 · Claim record · Lean coverage · Code · Worked examples
Corollary 7.3: Exact image and residue enumeration
Let \(\widetilde C_j\) denote the dyadic Newton interpolation. The map
is a bijective isometry \(\mathbb Z_2\to\mathbb Z_2\), and
For \(r>B_j\), one least period modulo \(2^r\) visits every residue in \(j!+2^{B_j}\mathbb Z/2^r\mathbb Z\) exactly once. For \(r\leq B_j\), there is one residue and the least row period is one.
Read proof · Visual guide: A full period visits every allowed residue · Visual guide: Seeing the scaled isometry · Read statement and proof side by side · Permalink · Source line 1081 · Claim record · Lean coverage · Code · Worked examples
Proof of Corollary 7.3
The exact difference identity gives \(v_2(T_j(x)-T_j(y))=v_2(x-y)\) and \(T_j(0)=0\). For every \(s\geq1\), the map induces an injection of \(\mathbb Z_2/2^s\mathbb Z_2\) into itself. It is a bijection because both sets have \(2^s\) elements. These quotient bijections are compatible with reduction. Inverting them gives a compatible preimage system for every dyadic integer, proving surjectivity. Multiplication by \(j!\) changes the image ideal by exactly \(v_2(j!)\). The finite quotient bijection is also precisely the residue enumeration in a least row period.
Back to statement · Read statement and proof side by side · Permalink · Source line 1099 · Claim record · Lean coverage · Code · Worked examples
7.1 An infinite exact family at all dyadic ranks
On this page · Permalink · Source line 1112
Theorem 7.4: Lower-half exact laws
Let \(N=2^a\geq4\) and \(0\leq j<N/2\). Then for all \(n\geq0\) and \(H>0\),
Its normalized interpolation is a bijection onto \(1+2\mathbb Z_2\). For the original coefficients, the least row period modulo \(2^r\) is \(2^{\max(0,r-1-v_2(j!))}\).
Read proof · Visual guide: How far the simple metric law extends · Read statement and proof side by side · Permalink · Source line 1114 · Claim record · Lean coverage · Code · Worked examples
Proof of Theorem 7.4
Every slope \(N/i\) in the source has valuation at least two, since \(i\leq j<N/2\). Hence \(g(1)\equiv1\pmod4\) and \(g(2)\equiv1\pmod8\). The first two Newton coefficients from the reciprocal recurrence are
The first has valuation exactly one. Also \(\binom{2N}{N}\equiv6\pmod8\) for every \(N=2^a\geq2\). To see this, write
The even-\(h\) factors give \(B_{a-1}\); the odd-\(h\) factors are one modulo four for \(a\geq2\). Since \(B_{a-1}\) has valuation one, \(B_a\equiv B_{a-1}\pmod8\), and \(B_1=6\). Therefore \(a_2\equiv6-1+2+1=0\pmod8\). The global Mahler bound covers all \(k\geq3\), completing the criterion with \(b=1\). The period and image deductions are the same as above. The proof also covers \(N=2,j=0\), whose kernel is identically one.
Back to statement · Read statement and proof side by side · Permalink · Source line 1126 · Claim record · Lean coverage · Code · Worked examples
7.2 Complete exact laws at ranks two and eight
On this page · Permalink · Source line 1152
Proposition 7.5: Small-rank complete isometry table
Every address at \(N\in\{2,4,8\}\) satisfies
The following entries are ordered by \(j=0,\ldots,N-1\):
| \(N\) | normalized shifts \(b_{N,j}\) | original shifts \(B_{N,j}=b_{N,j}+v_2(j!)\) |
|---|---|---|
| 2 | \(1,2\) | \(1,2\) |
| 4 | \(1,1,4,2\) | \(1,1,5,3\) |
| 8 | \(1,1,1,1,4,3,2,2\) | \(1,1,2,2,7,6,6,6\) |
For each entry, the least original row period is \(2^{\max(0,r-B_{N,j})}\), its Euler subscript displacement is \(N\) times that period, and the exact-image and residue-enumeration conclusions hold with the corresponding shifts.
Read proof · Visual guide: How far the simple metric law extends · Read statement and proof side by side · Permalink · Source line 1154 · Claim record · Lean coverage · Code · Worked examples
Proof of Proposition 7.5
Rank four was proved above. At ranks two and eight the finite criterion is verified by the actual rational recurrence (7.6). The following valuation lists record every initial Newton coefficient needed in addition to the generic tail \(v_2(a_k)\geq k\):
| \(N\) | \(j\) | \(b_{N,j}\) | \(v_2(a_1),v_2(a_2),\ldots\) through the finite cutoff |
|---|---|---|---|
| 2 | 0 | 1 | \(1,3\) |
| 2 | 1 | 2 | \(2,5,9,11\) |
| 8 | 0 | 1 | \(1,3\) |
| 8 | 1 | 1 | \(1,3\) |
| 8 | 2 | 1 | \(1,3\) |
| 8 | 3 | 1 | \(1,3\) |
| 8 | 4 | 4 | \(4,6,9,13,16,20\) |
| 8 | 5 | 3 | \(3,6,10,13,17\) |
| 8 | 6 | 2 | \(2,6,8,13\) |
| 8 | 7 | 2 | \(2,11,8,15\) |
The first entries can also be read directly from \(a_1=-1-\binom{N+j}{j}^{-1}\). At rank eight they are respectively
All later entries in the table are finite rational arithmetic, with full numerator and denominator records retained in the accompanying exact certificate all_rank_extensions_verified.json. Every entry with \(k\geq2\) meets \(b_{N,j}+1+\lfloor\log_2k\rfloor\). Beyond the displayed cutoff, the inequality \(k\geq b_{N,j}+1+\lfloor\log_2k\rfloor\) holds at every order. Thus the finite table and the proved infinite tail establish the full hypothesis of Lemma 7.1. This is a finite coefficient proof followed by a universal tail argument, rather than extrapolation from sampled rows or steps.
Back to statement · Read statement and proof side by side · Permalink · Source line 1175 · Claim record · Lean coverage · Code · Worked examples
Remark 7.6: Failure of a universal upper-half exact law
The all-rank lower bound remains valid in every address. The exact first-difference law does not automatically extend to all upper-half addresses. Exact rational recurrence calculations give
| \((N,j)\) | \(v_2(C(1)-C(0))\) | \(v_2(C(2)-C(0))\) | naive \(b+v_2(2)\) |
|---|---|---|---|
| \((16,9)\) | 5 | 7 | 6 |
| \((32,18)\) | 6 | 6 | 7 |
| \((32,19)\) | 5 | 7 | 6 |
For example, at \((N,j)=(16,9)\) the first difference is \(-2042976/2042975\), whereas the two-row difference is \(-2215603929728/2215923048625\). These exact counterexamples refute the additional universal isometry proposal. They preserve the threshold-free lower bound and the original conjecture.