Prime-Power Coefficient Equations and OEIS A393867
Abstract
The nth prime divides the nth logarithmic-derivative term of OEIS A393866 for n greater than one.
The literature note hanna2026a393867 records Hanna’s defining equations for A393866 and the divisibility conjecture in A393867. Here F denotes generatingSeries and L denotes logDerivative, both formal power series over the integers. All indices and exponents are natural numbers; subtraction in an index is natural subtraction. Coefficients and divisibility are over the integers, with natural primes cast to integers.
The notation C embeds an integer as a constant series. The operator derivative is the formal derivative over the integers, and invOfUnit(F,1) is the formal inverse with constant coefficient one. The operator div is integer division; every division used by the coefficient construction is exact.
Definition 1.1 (The one-based prime index).
Formalization. D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.prime (✓ std3).
Source. Repository-derived.
Commentary.
Mathlib’s nth operator uses a zero-based index, so prime(n) is Nat.nth Nat.Prime (n-1). The subtraction convention also defines prime(0).
Theorem 1.2 (The prime exceeds its positive index).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.lt_prime (✓ std3). ∎
Source. Repository-derived.
Commentary.
Mathlib’s bound k+2 <= Nat.nth Nat.Prime k, with k=n-1, gives the strict inequality.
Definition 1.3 (The integral coefficient construction).
Formalization. D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.a (✓ std3).
Source. Repository-derived.
Commentary.
Write P(k) for the private approximation(k). Starting from P(0)=1, the displayed correction changes only degree k+1. In the binomial expansion of B^p=(1+(B-1))^p, the interior binomial coefficients are divisible by p and the pth power of B-1 has no coefficient below p. This proves exact divisibility of the correction numerator. Later approximations preserve every earlier coefficient.
Definition 1.4 (The generating series).
Formalization. D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.generatingSeries (✓ std3).
Source. Repository-derived.
Commentary.
The series F is mk(a), the formal integer series with coefficient a(n) at degree n.
Theorem 1.5 (The complete defining equations).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.generating_equation (✓ std3). ∎
Source. Repository-derived.
Commentary.
If two unit-constant series agree below degree n, the difference of their nth coefficients after taking the pth power is p times their original nth-coefficient difference. Factoring the difference of powers proves this identity. The exact correction therefore enforces the nth equation; coefficient stability transfers it to F.
Theorem 1.6 (Uniqueness among integer series).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.generating_unique (✓ std3). ∎
Source. Repository-derived.
Commentary.
Strong induction compares B and F below each degree n. Their defining equations and the difference-of-powers coefficient identity imply prime(n) times the coefficient difference is zero. Cancellation of this nonzero integer proves equality at degree n.
Theorem 1.7 (Low coefficients of a prime power).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.prime_dvd_coeff_pow (✓ std3). ∎
Source. Repository-derived.
Commentary.
Apply the binomial divisibility argument to F itself. The bound 1 <= j < prime(n) excludes both exceptional binomial terms.
Definition 1.8 (The logarithmic derivative).
Formalization. D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.logDerivative (✓ std3).
Source. Repository-derived.
Commentary.
Since F has constant coefficient one, multiplication by its formal unit inverse defines L=F’/F over the integers.
Definition 1.9 (The sequence indexing).
Formalization. D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.a393867 (✓ std3).
Source. Repository-derived.
Commentary.
A393867 starts at index one: its nth term is the coefficient of degree n-1 in L. Natural subtraction also specifies the value at zero.
Theorem 1.10 (Differentiating every power).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.log_derivative_identity (✓ std3). ∎
Source. Repository-derived.
Commentary.
The formal power rule and L times F equals F’ give the displayed identity, including exponent zero.
Theorem 1.11 (Hanna’s divisibility conjecture).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.hanna_conjecture (✓ std3). ∎
Resolves. Problems/oeis-a393867-prime-power-shift-log-derivative (proved) by D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.hanna_conjecture.
Citation. Paul D. Hanna (2026). OEIS A393867, logarithmic derivative of A393866 (g.f. with [x^n] A^prime(n) = prime(n) [x^(n-1)] A^prime(n)). URL: https://oeis.org/A393867.
Commentary.
Let p=prime(n) and h(j) be the coefficient of degree j in F^p. Coefficient extraction from the power rule, followed by h(n)=p h(n-1) and cancellation of p, gives n h(n-1)=[x^(n-1)](L F^p). For n>1, the left side is divisible by p. In the convolution on the right, every term except [x^(n-1)]L times h(0) has a positive coefficient index below p in F^p and is divisible by p. Since h(0)=1, the remaining term proves the conjecture.
Theorem 1.12 (The shifted printed formula is false).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.printed_formula_false (✓ std3). ∎
Source. Repository-derived.
Commentary.
This companion concerns formula (2) of A393866, whose printed coefficient index is n instead of n-1. The first three defining equations give coefficients 1, 2, and 10 for F. Coefficient extraction from L F=F’ then gives [x^2]L=25, which is not divisible by prime(2)=3. The separate oddness conjecture is not addressed.
References
- Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.a - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.a393867 - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.generatingSeries - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.generating_equation - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.generating_unique - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.hanna_conjecture - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.logDerivative - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.log_derivative_identity - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.lt_prime - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.prime - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.prime_dvd_coeff_pow - Truth anchor:
D5/S1/Recurrence/Parity/PrimePowerShiftLogDerivative.printed_formula_false