Low coefficients of agree with the finite product
ProvedQseriesFormalization.PartIV.Ch19.coeff_qPochInfPS_eq_coeff_finite_product_of_ltchan-invitation-qseriesformal-power-seriesq-pochhammer
Let be a commutative topological ring (Hausdorff, with continuous operations) and natural numbers. Then
where is qPochInfPS. Factors beyond do not affect coefficients below . Chapter 19.
Preamble
import Definitions.Def_QseriesFormalization_DefCoreQSeries
import Mathlib.Combinatorics.Enumerative.Partition.GenFun
import Mathlib.RingTheory.PowerSeries.Basic
import Mathlib.RingTheory.PowerSeries.Expand
import Mathlib.RingTheory.PowerSeries.Inverse
import Mathlib
import Mathlib.Combinatorics.Enumerative.Partition.Basic
import Mathlib.Tactic.Linarith
open QseriesFormalization
open QseriesFormalization.PartIV
open QseriesFormalization.PartIV.Ch19
open QseriesFormalization.Ch01 (partitionCount)
open PowerSeries
-- Ramanujan's deeper congruences require higher partition values.
-- Here we verify congruences using p(0)..p(11) proved in Ch01.
-- p(25n+24) ≡ 0 (mod 25) requires partitionCount_24 (not yet in Ch01).
-- Additional divisibility and modular facts
-- p(n) mod 3 for n=0..11
-- p(n) mod 4
/-! ### Modular arithmetic for primes 13, 17, 19, 23 -/
-- p(n) mod 13
-- p(n) mod 17
-- p(n) mod 19
-- p(n) mod 23
/-! ### Composite moduli -/
-- p(n) mod 10
-- p(n) mod 12
-- p(n) mod 15
/-! ### Additional divisibility facts -/
/-! ### Ramanujan's congruences verified at available indices
`p(5n+4) ≡ 0 (mod 5)` for n = 0, 1
`p(7n+5) ≡ 0 (mod 7)` for n = 0
`p(11n+6) ≡ 0 (mod 11)` for n = 0
-/
/-! ### Pentagonal recurrence agreement (Chan §1)
`partitionCountRec` is the Euler pentagonal recurrence formulation of p(n).
For n = 0..11, it agrees with the actual partition count `partitionCount n`. -/
/-! ### Specific Ramanujan-congruence identities -/
/-! ### Mod 6 patterns (composite where p(6) is divisible) -/
/-! ### Mod 8 -/
/-! ### Mod 9 -/
/-! ### Cumulative partition sums
`∑_{k=0}^n p(k)` for small n. -/
/-! ### Order properties of p -/
/-! ### Parity (mod 2) -/
/-! ### Bounds: p(n) ≥ n+1 for n=0..4, and various comparisons -/
/-! ### Growth ratios
p(n+1)/p(n) is monotonically increasing for small n. Verified numerically. -/
/-! ### Specific differences -/
/-! ### Number-theoretic properties of specific values
p(n) for n where p(n) has interesting prime factorization. -/
/-! ### Squarefree-ness checks -/
/-! ### Ramanujan family — uniform interface
A Ramanujan congruence is `m ∣ p(m·n + r)`. The classical Ramanujan congruences:
- `m = 5, r = 4`: `5 ∣ p(5n + 4)` (proved at n=0, 1)
- `m = 7, r = 5`: `7 ∣ p(7n + 5)` (proved at n=0)
- `m = 11, r = 6`: `11 ∣ p(11n + 6)` (proved at n=0)
-/
/-! ### Non-Ramanujan: counter-examples
Some `(m, r)` patterns are NOT Ramanujan families. -/
/-! ### Partition generating function and Ramanujan congruences
We use Mathlib's `Nat.Partition.genFun` infrastructure. Setting the character
function `f = (fun _ _ => 1)` gives the standard partition generating function
`∑ p(n) X^n` with the Euler product identity `∏ 1/(1 - X^(k+1))`.
This is the foundation for proving the **general** Ramanujan congruences
(`5 ∣ p(5n+4)`, etc.) by Frobenius-mod-p techniques on power series. -/
open scoped PowerSeries.WithPiTopology
/-! ### Ramanujan congruence at the power series level
Goal: prove `(partitionGenFun (ZMod 5)).coeff (5*n + 4) = 0` for all n.
This is the **algebraic** Ramanujan congruence: it says the coefficients of
the partition generating function over ZMod 5 vanish in arithmetic progression
4 mod 5. Equivalently `5 ∣ partitionCount (5n+4)` for all n.
The classical proof (Ramanujan 1919) uses:
1. Frobenius mod 5: `(1 - X^k)^5 = 1 - X^{5k}` in `(ZMod 5)⟦X⟧`.
2. `(q;q)∞^4 ≡ (q^5;q^5)∞ · J(q) (mod 5)` where `J = 1/(q;q)∞`.
3. Jacobi's identity for `(q;q)∞^3`.
4. Mod-5 coefficient analysis.
We state the result and verified instances. -/
/-! ### Frobenius endomorphism for power series over `ZMod p`
In `(ZMod p)⟦X⟧`, the map `φ ↦ φ^p` is a ring endomorphism (the Frobenius).
In particular `(a + b)^p = a^p + b^p` for all a, b. This gives us:
`(1 + c · X^k)^p = 1 + c^p · X^{kp}`
For `c = -1` (and odd `p`): `(1 - X^k)^p = 1 - X^{kp}`.
This is the algebraic key to Ramanujan's mod-p congruences. -/
/-! ### Finite q-Pochhammer and the Euler product
The finite q-Pochhammer `(q;q)_N = ∏_{k=1}^N (1 - q^k)` is a polynomial-like
power series. Frobenius applied factor-by-factor gives:
`(q;q)_N^p ≡ (q^p; q^p)_N (mod p)` in `(ZMod p)⟦X⟧`. -/
/-! ### The q ↦ q^p substitution on power series
Define a power series operation `subXPow p f` that sends `∑ aₙ Xⁿ` to `∑ aₙ X^{pn}`.
This is the "raise the variable" map, central to Ramanujan's argument:
in `(ZMod p)⟦X⟧`, `(q;q)∞^p = (q^p; q^p)∞`. -/
/-! ### Use Mathlib's `PowerSeries.expand` (the variable substitution X ↦ X^p)
Mathlib provides `PowerSeries.expand p hp : R⟦X⟧ →ₐ[R] R⟦X⟧` which sends
`∑ aₙ Xⁿ` to `∑ aₙ X^{p·n}`. This is the same operation as our `subXPow` but
already comes with the full algebra-homomorphism structure.
The key theorem is `MvPowerSeries.map_frobenius_expand`:
`(f.expand p).map (frobenius R p) = f ^ p`
For `R = ZMod p`, the Frobenius map on ZMod p is the identity (Fermat's little
theorem), so we get the clean form:
**`f.expand p hp = f ^ p` in `(ZMod p)⟦X⟧`** -/
/-! ### The `(q;q)∞` formal power series
We define `qPochInfPS R := (partitionGenFun R)⁻¹` (as a power series inverse,
which exists since the constant term of `partitionGenFun R` is 1, a unit).
This gives us `partitionGenFun R * qPochInfPS R = 1` algebraically. -/
open scoped PowerSeries.WithPiTopology
open Filter Topology
open scoped PowerSeries.WithPiTopology
open PowerSeries
open scoped PowerSeries.WithPiTopology
/-! ### Specific qPochInfPS coefficient values via Euler pentagonal
Derived from `coeff_qPochInfPS_eq_pentagonalSign` + `decide` on `pentagonalSign n` for
specific small n. These extend beyond the n=0..8 values proved via the J·qPoch=1 recurrence
in earlier sections. -/
/-! ### B2 groundwork: Jacobi triple product at z=1 (cube identity)
Goal: prove `(qPochInfPS R)^3 = ∑ (-1)^n (2n+1) X^{n(n+1)/2}` as formal power series.
This is the classical Jacobi identity `∏(1-q^n)^3 = ∑(-1)^n(2n+1)q^{T_n}`.
Ch04_T43 has the analytic version (`jacobiIdentity`) for `|q| < 1`.
The formal-PS version requires either:
(a) Combinatorial proof via Sylvester-style involution on triple partitions
(b) Analytic → formal bridge via Taylor expansion uniqueness
Both are substantial work; we define the target and prove small cases here. -/
-- Sanity: triangular numbers 0, 1, 3, 6, 10, 15, ...
-- Values: 1, -3, 5, -7, 9, -11, ...
-- **B2 target theorem**: `(qPochInfPS R)^3 = jacobiThetaPS R` in `R⟦X⟧`.
-- This is the formal-PS version of Jacobi's identity (Ch04.jacobiIdentity in
-- analytic form). Full proof requires either Sylvester's combinatorial bijection
-- or an analytic ↔ formal Taylor uniqueness bridge.
-- Small case verification: `(qPochInfPS R)^3 . coeff n` for small n is computable
-- via the generic `coeff_n_pow_of_constantCoeff_one` helpers in Ch20 and Euler-pentagonal
-- coefficient values. See Ch20.ramanujanTau_two for the analogous tau computation pattern.Formal statement
theorem QseriesFormalization.PartIV.Ch19.coeff_qPochInfPS_eq_coeff_finite_product_of_lt
(R : Type*) [CommRing R] [TopologicalSpace R] [IsTopologicalRing R] [T2Space R]
(k N : Nat) (hkN : k < N) :
(qPochInfPS R).coeff k =
(∏ i ∈ Finset.range N, ((1 : R⟦X⟧) - PowerSeries.X ^ (i + 1))).coeff k := by sorrySource
Lean formalization (Xiang Huang) of Hei-Chi Chan, An Invitation to q-Series: From Jacobi's Triple Product Identity to Ramanujan's "Most Beautiful Identity" (World Scientific, 2011), Chapter 19. Lean source: https://github.com/xiangyazi24/invitation-to-qseries-lean/blob/5df34b280b6a9c4a68ab09105b4e5d3c7bbe88d1/QseriesFormalization/Chapter19.lean#L1723-L1731