OEIS/Eta products: Difference between revisions

From tehowiki
Jump to navigation Jump to search
imported>Gfis
Created page with "===Dedekind eta η=== * [https:oeis.org/A000700 A000700] ===Jacobi theta_2 === ===Jacobi θ_3, Ramanujan phi φ === [https://oeis.org/A000122 A000122] 1, 2, 0, 0,..."
 
Product g.f.s
 
(18 intermediate revisions by 2 users not shown)
Line 1: Line 1:
===Dedekind eta η===
===Eta product signatures (EPSIG)===
* [https:oeis.org/A000700 A000700]
The coefficient sequences for generating functions that are products of Dedkind's eta function can be computed conveniently by an Euler transform of some periodic integer sequence.
===Jacobi theta_2 ===
Following Michael Somos, we describe an '''eta product''' by a matrix resp. a list of pairs ''(q<sub>i</sub>, e<sub>i</sub>)'' of the form
===Jacobi &theta;_3, Ramanujan phi &phi; ===
[q1,e1;q2,e2;q3,e3;...].
The pairs are separated by ";".
* The ''q<sub>i</sub>'' are the powers of the argument ''q'' inside the eta function , and
* the ''e<sub>i</sub>'' are the powers of the eta functions. Only these can be negative.
We call such a  list the ''signature'' ("EPSIG") of the eta product. The following sections contain a number of examples.
 
The order of the pairs is irrelevant for the computation of the period, but see the section about normalization, below.
Once the signatures are normalized, they signatures can be sorted and stored in a database table.
The original expression of eta functions can be reconstructed from them, while in the generated periods the original structure is lost.
 
The generation of a period from an EPSIG, the following Euler transform and the generation of the coefficients are implemented in
[https://github.com/archmageirvine/joeis/blob/master/src/irvine/oeis/transform/EtaProductSequence.java transform/EtaProductSequence.java]
as part of the project [https://github.com/archmageirvine/joeis jOEIS] of Sean A. Irvine.
===Common generating functions===
====Partition numbers====
[https://oeis.org/A000041 A000041] 1, 1, 2, 3, 5, 7, 11, 15, 22, 30, 42, 56, 77, 101, 135, 176, 231
Euler transform of period 1: [1]
G.f.: Product_{k>0} 1/(1-x^k)
EPSIG="[1,-1]" Q="1/24"
====Dedekind eta &eta; (without the q^(1/24) factor), Ramanujan f(-q)====
[https://oeis.org/A010815 A010815] 1, -1, -1, 0, 0, 1, 0, 1, 0, 0, 0, 0, -1, 0, 0, -1, 0, 0, 0, 0, 0, 0, 1
G.f.: eta(q)
Euler transform of period 1: [-1]
EPSIG="[1,1]"
====Jacobi theta_2 &theta;_2====
[https://oeis.org/A089800 A089800] 2, 0, 2, 0, 0, 0, 2, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 2 ... = theta_2(q)/q^(1/4)
====Jacobi theta_3 &theta;_3, Ramanujan phi &phi; ====
  [https://oeis.org/A000122 A000122] 1, 2, 0, 0, 2, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 2
  [https://oeis.org/A000122 A000122] 1, 2, 0, 0, 2, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 2
  eta(q^2)^5 / (eta(q)*eta(q^4))^2
  a(0) = 1, for n >= 1: a(n) = 2 if n is a square, otherwise 0.
G.f.: eta(q^2)^5/(eta(q)*eta(q^4))^2
G.f.: Product_{k>0} (1 - x^(2*k))*(1 + x^(2*k-1))^2.
  Euler transform of period 4: [2,-3,2,-1]
  Euler transform of period 4: [2,-3,2,-1]
  eps P="[2,5;1,-2;4,-2]"
  EPSIG="[2,5;4,-2;1,-2]"
===Jacobi &theta;_4===
====Jacobi theta_4 &theta;_4====
  [https://oeis.org/A002448 A002448] 1, -2, 0, 0, 2, 0, 0, 0, 0, -2, 0, 0, 0, 0, 0, 0, 2
  [https://oeis.org/A002448 A002448] 1, -2, 0, 0, 2, 0, 0, 0, 0, -2, 0, 0, 0, 0, 0, 0, 2
  eta(q)^2 / eta(q^2)
  a(0) = 1, for n >= 1: a(n) = 2 * (-1)^sqrt(n) if n is a square, otherwise 0.
G.f.: eta(q)^2/eta(q^2)
  Euler transform of period 2: [2,-1]
  Euler transform of period 2: [2,-1]
  eps P="[1,2;2,-1]"
  EPSIG="[1,2;2,-1]"
===Ramanujan psi &psi;===
====Ramanujan psi &psi;====
* [https:oeis.org/A000700 A000700]
[https://oeis.org/A010054 A010054] 1, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0
===Ramanujan chi &chi;===
a(n) = 1 if n is a triangular number, otherwise 0.
* [https:oeis.org/A000700 A000700]
G.f.: q^(-1/8) * eta(q^2)^2/eta(q)
===Ramanujan f===
G.f.: Product_{k>0} (1-x^(2*k))/(1-x^(2*k-1))
Euler transform of period 2: [1,-1]
EPSIG="[2,2;1,-1]"
====Ramanujan chi &chi;====
[https://oeis.org/A000700 A000700] 1, 1, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 4, 5
G.f.: q^(1/24) * eta(q^2)^2/(eta(q)*eta(q^4))
G.f.: Product_{k>=1} (1 + x^(2*k-1))
Euler transform of period 4: [1,-1,1,0]
EPSIG="[2,2;4,-1;1,-1]"
==== Ramanujan f ====
[https://oeis.org/A121373 A121373] 1, 1, -1, 0, 0, -1, 0, -1, 0, 0, 0, 0, -1, 0, 0, 1
a(n) = (-1)^n * [https://oeis.org/A010815 A010815](n)
G.f.: q^(-1/24) * eta(q^2)^3/(eta(q)*eta(q^4))
Euler transform of period 4: [1,-2,1,-1]
EPSIG="[2,3;4,-1;1,-1]"
====Ramanujan tau &tau;====
[https://oeis.org/A000594 A000594] 1, -24, 252, -1472, 4830, -6048, -16744, 84480, -113643, -115920, 534612
G.f.: eta(q)^24
G.f.: x * Product_{k>=1} (1 - x^k)^24
Euler transform of period 1: [-24]
EPSIG="[1,24]"
====Ramanujan a cubic AGM theta====
[https://oeis.org/A004016 A004016] 1, 6, 0, 6, 6, 0, 0, 12, 0, 6, 0, 0, 6, 12, 0, 0, 6
G.f.: theta_3(q)*theta_3(q^3) + theta_2(q)*theta_2(q^3)
G.f.: phi(x)*phi(x^3) + 4*x*psi(x^2)*psi(x^6)
==== Ramanujan b cubic AGM theta====
[https://oeis.org/A005928 A005928] 1, -3, 0, 6, -3, 0, 0, -6, 0, 6, 0, 0, 6, -6, 0, 0, -3
G.f.: eta(q)^3/(eta(q^3)
Euler transform of period 3: [-3,-3,-2]
EPSIG="[1,3;3,-1]"
==== Ramanujan c cubic AGM theta====
[https://oeis.org/A005882 A005882] 3, 3, 6, 0, 6, 3, 6, 0, 3, 6, 6, 0, 6, 0, 6, 0, 9, 6, 0
G.f.: 3*eta(q^3)^3/eta(q)
3 * (Euler transform of period 3: [-3,-3,-2])
EPSIG="[3,3;1,-1]" Q="-1/3" F=3
=== Eisenstein series===
==== E_2 = Ramanujan P(q) ====
[https://oeis.org/A006352 A006352] 1, -24, -72, -96, -168, -144, -288, -192, -360
a(n) = -24*sigma(n)
G.f.: 1 + 24 *q*deriv(eta(q))/eta(q)
==== E_4 = Ramanujan Q(q) ====
[https://oeis.org/A004009 A004009] 1, 240, 2160, 6720, 17520, 30240, 60480, 82560, 140400
G.f.: (eta(q)^24 + 256*eta(q^2)^24) / (eta(q)*eta(q^2))^8
EPOLY="[1,24]+256*[2,24;1,-8;2,-8]"
==== E_6 = Ramanujan R(q) ====
[https://oeis.org/A013973 A013973] 1, -504, -16632, -122976, -532728, -1575504, -4058208, -8471232
a(0) = 1, for n >= 1: a(n) = -504*sigma_5(n)
G.f.: (eta(q)^8 + 32*eta(q^4)^8) * (eta(q)^16 - 512*eta(q)^8*eta(q^4)^8 - 8192*eta(q^4)^16) / eta(q^2)^12
"[1,8]+32*[4,8]" convolved with "[1,16;2,-12]-512*[1,8;4,8;2,-12]-8192*[4,16;2,-12]"
==== E_8 ====
[https://oeis.org/A008410 A008410] 1, 480, 61920, 1050240, 7926240, 37500480, 135480960
a(0) = 1, for n >= 1: a(n) = 480*sigma_7(n)
G.f.: (theta2(q)^16 + theta3(q)^16 + theta4(q)^16)/2.
G.f.: ((eta(q)^24 + 256*eta(q^2)^24) / (eta(q)*eta(q^2))^8)^2
"[1,24;1,-8;2,-8]+256*[2,24;1,-8;2,-8]" squared
==== E_2k, k=1..16 ====
E_2:  [https://oeis.org/A006352 A006352] a(0) =            1, for n >= 1: a(n) =    -24*sigma_1(n)
E_4:  [https://oeis.org/A004009 A004009] a(0) =             1, for n >= 1: a(n) =     240*sigma_3(n)
E_6:  [https://oeis.org/A013973 A013973] a(0) =            1, for n >= 1: a(n) =    -504*sigma_5(n)
E_8:  [https://oeis.org/A008410 A008410] a(0) =            1, for n >= 1: a(n) =    480*sigma_7(n)
E_10: [https://oeis.org/A013974 A013974] a(0) =            1, for n >= 1: a(n) =    -264*sigma_9(n)
E_12: [https://oeis.org/A029828 A029828] a(0) =          691, for n >= 1: a(n) =  65520*sigma_11(n)
E_14: [https://oeis.org/A058550 A058550] a(0) =            1, for n >= 1: a(n) =    -24*sigma_13(n)
E_16: [https://oeis.org/A029829 A029829] a(0) =          3617, for n >= 1: a(n) =  16320*sigma_15(n)
E_18: [https://oeis.org/A279892 A279892] a(0) =        43867, for n >= 1: a(n) =  -28728*sigma_17(n)
E_20: [https://oeis.org/A029830 A029830] a(0) =        174611, for n >= 1: a(n) =  13200*sigma_19(n)
E_22: [https://oeis.org/A279893 A279893] a(0) =        77683, for n >= 1: a(n) =    -552*sigma_21(n)
E_24: [https://oeis.org/A029831 A029831] a(0) =    236364091, for n >= 1: a(n) =  131040*sigma_23(n)
E_26: [https://oeis.org/A282356 A282356] a(0) =        657931, for n >= 1: a(n) =    -24*sigma_25(n)
E_28: [https://oeis.org/A282401 A282401] a(0) =    3392780147, for n >= 1: a(n) =    6960*sigma_27(n)
E_30: [https://oeis.org/A282182 A282182] a(0) = 1723168255201, for n >= 1: a(n) = -171864*sigma_29(n)
E_32: [https://oeis.org/A282540 A282540] a(0) = 7709321041217, for n >= 1: a(n) =  32640*sigma_31(n)
The <code>a(0)</code> constants are given by <code>abs([https://oeis.org/A001067 A001067: Numerator of Bernoulli(2*n)/(2*n)])</code>, while the factors of <code>sigma_xx(n)</code> are found in <code>[https://oeis.org/A006863 A006863: Denominator of B_{2*n}/(-4*n</code>]


===Cookbook===
===Leading power of q factor (pqf) ===
* For all tuples [m,k] add m*k.
Some of the generating functions have an additional power of q factor, in order to normalize the leading coefficient to <code>q^0</code>.
* The resulting integer part gives the q-shift, while the fractional part is noted before the eta product.
* Example [https://oeis.org/A226862 A226862]:
G.f.: q^(-1/6) * eta(q^4) * eta(q^6)^5 / (eta(q^3) * eta(q^12))^2
EPSIG="[4,1;6,5;3,-2;12,-2]"
For all pairs we multiply the q power with the eta power and add the results, giving 4*1 + 6*5 - 3*2 - 12*2 = 4 + 30 - 6 - 24 = 4. This is the negative numerator of the pqf (expanded for denominator 24).
* Example [https://oeis.org/A286813 A286813]:
G.f.: q^(-9/8) * (eta(q^2) * eta(q^16))^2 / (eta(q) * eta(q^8))
EPSIG="[2,2;16,2;1,-1;8,-1]" gives 2*2 * 16*2  - 1*1 - 8*1 = 36 - 9 = 27 and pqf = q^(-27/24)
* Example [https://oeis.org/A286134 A286134] 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, -1, -1, -1
G.f.: q^(-1/2) * eta(q^5) * eta(q^6) * eta(q^7) * eta(q^210)
EPSIG="[5,1;6,1;7,1;210,1]" gives 18 + 210 = 228 and pqf = q^(-228/24) = -9 -1/2.
The sequence is shifted right by 9 positions in this case.
===Computation of the period from an EPSIG===
This algorithm is rather straightforward. It uses an array of integers that are usually small. The length of the array is the least common multiple (LCM) of all ''q<sub>i</sub>''. The array with indexes [0..lcm-1] is initialized with zeros.
# For all pairs ''(q<sub>i</sub>, e<sub>i</sub>)''
## Build a block of '' q<sub>i</sub> - 1'' zeros, followed by ''-1 * e<sub>i</sub>''.
## Add a sufficient concatenation of such blocks onto the whole array.
# The resulting array is the period for the Euler transform.
As an example, we derive the period of the theta_3 function above from its EPSIG <code>[2,5;4,-2;1,-2]</code>. The period length is LCM(2,1,4) = 4.
  pair 2,5  ->  0 -5  0 -5
  pair 4,-2  ->  0  0  0  2
  pair 1,-2  ->  2  2  2  2
  --------------------------
  sum = period = [2,-3, 2,-1]
This algorithm for expanding a list of powers of theta_3 functions is realized in <code>[https://github.com/gfis/joeis-lite/blob/master/internal/fischer/theta3_epsig.pl theta3_epsig.pl]</code>.
===Normalization of an EPSIG===
For long lists of generating functions it is convenient to require all EPSIGs to be normalized, such that they can be tested for equality and can be sorted. We use the following rules:
# The sublist of pairs with positive ''e<sub>i</sub>'' comes before the sublist with negative ''e<sub>i</sub>''.
# Inside a sublist the first components ''q<sub>i</sub>'' are in descending order.
# Successive pairs with the same ''q<sub>i</sub>'' are combined into a single pair with the sum of the ''e<sub>i</sub>'' as second component.
The normalization rules are implemented in <code>[https://github.com/gfis/OEIS-mat/blob/master/eta/norm_epsig.pl norm_epsig.pl]</code>.
===Multiplication of two EPSIGs===
Two EPSIGs could be multiplied by concatenating their list of pairs. It is convenient, however, to combine all occurrences of specific ''q<sub>i</sub>'' by adding the corresponding eta exponents ''e''.
===Exponentiation of an EPSIG===
An EPSIG is taken to a power ''m'' by multiplying all ''e<sub>i</sub>'' (the second components in the pairs) by ''m''. The exponent may be ''1/2'' for the square root, or any other fraction ''m<sub>n</sub>/m<sub>d</sub>'' as long as the ''q<sub>i</sub>'' are all divisible by ''m<sub>d</sub>''.
===Eta polynomials===
In a number of OEIS sequences the eta products occur as components of a polynomial (cf. for example E_4, E_6, E_8 above).

Latest revision as of 18:43, 17 October 2024

Eta product signatures (EPSIG)

The coefficient sequences for generating functions that are products of Dedkind's eta function can be computed conveniently by an Euler transform of some periodic integer sequence. Following Michael Somos, we describe an eta product by a matrix resp. a list of pairs (qi, ei) of the form

[q1,e1;q2,e2;q3,e3;...]. 

The pairs are separated by ";".

  • The qi are the powers of the argument q inside the eta function , and
  • the ei are the powers of the eta functions. Only these can be negative.

We call such a list the signature ("EPSIG") of the eta product. The following sections contain a number of examples.

The order of the pairs is irrelevant for the computation of the period, but see the section about normalization, below. Once the signatures are normalized, they signatures can be sorted and stored in a database table. The original expression of eta functions can be reconstructed from them, while in the generated periods the original structure is lost.

The generation of a period from an EPSIG, the following Euler transform and the generation of the coefficients are implemented in transform/EtaProductSequence.java as part of the project jOEIS of Sean A. Irvine.

Common generating functions

Partition numbers

A000041 1, 1, 2, 3, 5, 7, 11, 15, 22, 30, 42, 56, 77, 101, 135, 176, 231
Euler transform of period 1: [1]
G.f.: Product_{k>0} 1/(1-x^k)
EPSIG="[1,-1]" Q="1/24"

Dedekind eta η (without the q^(1/24) factor), Ramanujan f(-q)

A010815 1, -1, -1, 0, 0, 1, 0, 1, 0, 0, 0, 0, -1, 0, 0, -1, 0, 0, 0, 0, 0, 0, 1
G.f.: eta(q)
Euler transform of period 1: [-1]
EPSIG="[1,1]"

Jacobi theta_2 θ_2

A089800 2, 0, 2, 0, 0, 0, 2, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 2 ... = theta_2(q)/q^(1/4)

Jacobi theta_3 θ_3, Ramanujan phi φ

A000122 1, 2, 0, 0, 2, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 2
a(0) = 1, for n >= 1: a(n) = 2 if n is a square, otherwise 0.
G.f.: eta(q^2)^5/(eta(q)*eta(q^4))^2
G.f.: Product_{k>0} (1 - x^(2*k))*(1 + x^(2*k-1))^2.
Euler transform of period 4: [2,-3,2,-1]
EPSIG="[2,5;4,-2;1,-2]"

Jacobi theta_4 θ_4

A002448 1, -2, 0, 0, 2, 0, 0, 0, 0, -2, 0, 0, 0, 0, 0, 0, 2
a(0) = 1, for n >= 1: a(n) = 2 * (-1)^sqrt(n) if n is a square, otherwise 0.
G.f.: eta(q)^2/eta(q^2)
Euler transform of period 2: [2,-1]
EPSIG="[1,2;2,-1]"

Ramanujan psi ψ

A010054 1, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0
a(n) = 1 if n is a triangular number, otherwise 0.
G.f.: q^(-1/8) * eta(q^2)^2/eta(q)
G.f.: Product_{k>0} (1-x^(2*k))/(1-x^(2*k-1))
Euler transform of period 2: [1,-1]
EPSIG="[2,2;1,-1]"

Ramanujan chi χ

A000700 1, 1, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 4, 5
G.f.: q^(1/24) * eta(q^2)^2/(eta(q)*eta(q^4))
G.f.: Product_{k>=1} (1 + x^(2*k-1))
Euler transform of period 4: [1,-1,1,0]
EPSIG="[2,2;4,-1;1,-1]"

Ramanujan f

A121373 1, 1, -1, 0, 0, -1, 0, -1, 0, 0, 0, 0, -1, 0, 0, 1
a(n) = (-1)^n * A010815(n)
G.f.: q^(-1/24) * eta(q^2)^3/(eta(q)*eta(q^4))
Euler transform of period 4: [1,-2,1,-1]
EPSIG="[2,3;4,-1;1,-1]"

Ramanujan tau τ

A000594 1, -24, 252, -1472, 4830, -6048, -16744, 84480, -113643, -115920, 534612
G.f.: eta(q)^24
G.f.: x * Product_{k>=1} (1 - x^k)^24
Euler transform of period 1: [-24]
EPSIG="[1,24]"

Ramanujan a cubic AGM theta

A004016 1, 6, 0, 6, 6, 0, 0, 12, 0, 6, 0, 0, 6, 12, 0, 0, 6
G.f.: theta_3(q)*theta_3(q^3) + theta_2(q)*theta_2(q^3)
G.f.: phi(x)*phi(x^3) + 4*x*psi(x^2)*psi(x^6)

Ramanujan b cubic AGM theta

A005928 1, -3, 0, 6, -3, 0, 0, -6, 0, 6, 0, 0, 6, -6, 0, 0, -3
G.f.: eta(q)^3/(eta(q^3)
Euler transform of period 3: [-3,-3,-2]
EPSIG="[1,3;3,-1]"

Ramanujan c cubic AGM theta

A005882 3, 3, 6, 0, 6, 3, 6, 0, 3, 6, 6, 0, 6, 0, 6, 0, 9, 6, 0
G.f.: 3*eta(q^3)^3/eta(q)
3 * (Euler transform of period 3: [-3,-3,-2])
EPSIG="[3,3;1,-1]" Q="-1/3" F=3

Eisenstein series

E_2 = Ramanujan P(q)

A006352 1, -24, -72, -96, -168, -144, -288, -192, -360
a(n) = -24*sigma(n) 
G.f.: 1 + 24 *q*deriv(eta(q))/eta(q)

E_4 = Ramanujan Q(q)

A004009 1, 240, 2160, 6720, 17520, 30240, 60480, 82560, 140400
G.f.: (eta(q)^24 + 256*eta(q^2)^24) / (eta(q)*eta(q^2))^8 
EPOLY="[1,24]+256*[2,24;1,-8;2,-8]"

E_6 = Ramanujan R(q)

A013973 1, -504, -16632, -122976, -532728, -1575504, -4058208, -8471232
a(0) = 1, for n >= 1: a(n) = -504*sigma_5(n)
G.f.: (eta(q)^8 + 32*eta(q^4)^8) * (eta(q)^16 - 512*eta(q)^8*eta(q^4)^8 - 8192*eta(q^4)^16) / eta(q^2)^12 
"[1,8]+32*[4,8]" convolved with "[1,16;2,-12]-512*[1,8;4,8;2,-12]-8192*[4,16;2,-12]"

E_8

A008410 1, 480, 61920, 1050240, 7926240, 37500480, 135480960
a(0) = 1, for n >= 1: a(n) = 480*sigma_7(n)
G.f.: (theta2(q)^16 + theta3(q)^16 + theta4(q)^16)/2.
G.f.: ((eta(q)^24 + 256*eta(q^2)^24) / (eta(q)*eta(q^2))^8)^2
"[1,24;1,-8;2,-8]+256*[2,24;1,-8;2,-8]" squared

E_2k, k=1..16

E_2:  A006352 a(0) =             1, for n >= 1: a(n) =     -24*sigma_1(n) 
E_4:  A004009 a(0) =             1, for n >= 1: a(n) =     240*sigma_3(n) 
E_6:  A013973 a(0) =             1, for n >= 1: a(n) =    -504*sigma_5(n) 
E_8:  A008410 a(0) =             1, for n >= 1: a(n) =     480*sigma_7(n) 
E_10: A013974 a(0) =             1, for n >= 1: a(n) =    -264*sigma_9(n) 
E_12: A029828 a(0) =           691, for n >= 1: a(n) =   65520*sigma_11(n) 
E_14: A058550 a(0) =             1, for n >= 1: a(n) =     -24*sigma_13(n) 
E_16: A029829 a(0) =          3617, for n >= 1: a(n) =   16320*sigma_15(n) 
E_18: A279892 a(0) =         43867, for n >= 1: a(n) =  -28728*sigma_17(n) 
E_20: A029830 a(0) =        174611, for n >= 1: a(n) =   13200*sigma_19(n) 
E_22: A279893 a(0) =         77683, for n >= 1: a(n) =    -552*sigma_21(n) 
E_24: A029831 a(0) =     236364091, for n >= 1: a(n) =  131040*sigma_23(n)
E_26: A282356 a(0) =        657931, for n >= 1: a(n) =     -24*sigma_25(n)
E_28: A282401 a(0) =    3392780147, for n >= 1: a(n) =    6960*sigma_27(n)
E_30: A282182 a(0) = 1723168255201, for n >= 1: a(n) = -171864*sigma_29(n)
E_32: A282540 a(0) = 7709321041217, for n >= 1: a(n) =   32640*sigma_31(n)

The a(0) constants are given by abs(A001067: Numerator of Bernoulli(2*n)/(2*n)), while the factors of sigma_xx(n) are found in A006863: Denominator of B_{2*n}/(-4*n

Leading power of q factor (pqf)

Some of the generating functions have an additional power of q factor, in order to normalize the leading coefficient to q^0.

G.f.: q^(-1/6) * eta(q^4) * eta(q^6)^5 / (eta(q^3) * eta(q^12))^2
EPSIG="[4,1;6,5;3,-2;12,-2]"

For all pairs we multiply the q power with the eta power and add the results, giving 4*1 + 6*5 - 3*2 - 12*2 = 4 + 30 - 6 - 24 = 4. This is the negative numerator of the pqf (expanded for denominator 24).

G.f.: q^(-9/8) * (eta(q^2) * eta(q^16))^2 / (eta(q) * eta(q^8))
EPSIG="[2,2;16,2;1,-1;8,-1]" gives 2*2 * 16*2  - 1*1 - 8*1 = 36 - 9 = 27 and pqf = q^(-27/24)
  • Example A286134 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, -1, -1, -1
G.f.: q^(-1/2) * eta(q^5) * eta(q^6) * eta(q^7) * eta(q^210)
EPSIG="[5,1;6,1;7,1;210,1]" gives 18 + 210 = 228 and pqf = q^(-228/24) = -9 -1/2. 

The sequence is shifted right by 9 positions in this case.

Computation of the period from an EPSIG

This algorithm is rather straightforward. It uses an array of integers that are usually small. The length of the array is the least common multiple (LCM) of all qi. The array with indexes [0..lcm-1] is initialized with zeros.

  1. For all pairs (qi, ei)
    1. Build a block of qi - 1 zeros, followed by -1 * ei.
    2. Add a sufficient concatenation of such blocks onto the whole array.
  2. The resulting array is the period for the Euler transform.

As an example, we derive the period of the theta_3 function above from its EPSIG [2,5;4,-2;1,-2]. The period length is LCM(2,1,4) = 4.

 pair 2,5   ->   0 -5  0 -5 
 pair 4,-2  ->   0  0  0  2
 pair 1,-2  ->   2  2  2  2
 --------------------------
 sum = period = [2,-3, 2,-1] 

This algorithm for expanding a list of powers of theta_3 functions is realized in theta3_epsig.pl.

Normalization of an EPSIG

For long lists of generating functions it is convenient to require all EPSIGs to be normalized, such that they can be tested for equality and can be sorted. We use the following rules:

  1. The sublist of pairs with positive ei comes before the sublist with negative ei.
  2. Inside a sublist the first components qi are in descending order.
  3. Successive pairs with the same qi are combined into a single pair with the sum of the ei as second component.

The normalization rules are implemented in norm_epsig.pl.

Multiplication of two EPSIGs

Two EPSIGs could be multiplied by concatenating their list of pairs. It is convenient, however, to combine all occurrences of specific qi by adding the corresponding eta exponents e.

Exponentiation of an EPSIG

An EPSIG is taken to a power m by multiplying all ei (the second components in the pairs) by m. The exponent may be 1/2 for the square root, or any other fraction mn/md as long as the qi are all divisible by md.

Eta polynomials

In a number of OEIS sequences the eta products occur as components of a polynomial (cf. for example E_4, E_6, E_8 above).