OEIS/Eta products: Difference between revisions

From tehowiki
Jump to navigation Jump to search
imported>Gfis
Cookbook for qpf
imported>Gfis
Normalization, Exponentiation, Multiplication
Line 1: Line 1:
===Eta product signatures (eps)===
===Eta product signatures (EPS)===
The coefficient sequences for generating functions that are products of the eta function can be computed conveniently by an Euler transform of some periodic integer sequence.  
The coefficient sequences for generating functions that are products of the 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 <code>(q<sub>i</sub>, e<sub>i</sub>)</code> of the form
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
  [q1,e1;q2,e2;q3,e3;...]
  [q1,e1;q2,e2;q3,e3;...]
The q<sub>i</sub> the powers of q and the e<sub>i</sub> are powers of the eta function. The latter can be negative.
The ''q<sub>i</sub>'' are the powers of ''q'' and the ''e<sub>i</sub>'' are the powers of the eta functions. The latter can be negative.
===Partition numbers===
===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
  [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]
  Euler transform of period 1: [1]
===Dedkind eta &eta; (without the q^(1/24) factor) ===
====Dedkind eta &eta; (without the q^(1/24) factor)====
  [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
  [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
  eta(q)
  eta(q)
  Euler transform of period 1: [-1]
  Euler transform of period 1: [-1]
  eps P="[1,1]"
  eps P="[1,1]"
===Jacobi theta_2 &theta;_2===
====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)
  [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; ===
====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
  a(0) = 1, for n >= 1: a(n) = 2 if n is a square, otherwise 0.
  a(0) = 1, for n >= 1: a(n) = 2 if n is a square, otherwise 0.
Line 20: Line 21:
  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]"
  eps P="[2,5;1,-2;4,-2]"
===Jacobi theta_4 &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
  a(0) = 1, for n >= 1: a(n) = 2 * (-1)^sqrt(n) if n is a square, otherwise 0.
  a(0) = 1, for n >= 1: a(n) = 2 * (-1)^sqrt(n) if n is a square, otherwise 0.
Line 26: Line 27:
  Euler transform of period 2: [2,-1]
  Euler transform of period 2: [2,-1]
  eps P="[1,2;2,-1]"
  eps P="[1,2;2,-1]"
===Ramanujan psi &psi;===
====Ramanujan psi &psi;====
  [https://oeis.org/A010054 A010054] 1, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0
  [https://oeis.org/A010054 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.
  a(n) = 1 if n is a triangular number, otherwise 0.
Line 32: Line 33:
  Euler transform of period 2: [1,-1]
  Euler transform of period 2: [1,-1]
  eps P="[2,2;1,-1]"
  eps P="[2,2;1,-1]"
===Ramanujan chi &chi;===
====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
  [https://oeis.org/A000700 A000700] 1, 1, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 4, 5
  q^(1/24) * eta(q^2)^2 /(eta(q) * eta(q^4))
  q^(1/24) * eta(q^2)^2 /(eta(q) * eta(q^4))
  Euler transform of period 4: [1,-1,1,0]
  Euler transform of period 4: [1,-1,1,0]
  eps P="[2,2;1,-1;4,-1]"
  eps P="[2,2;1,-1;4,-1]"
===Ramanujan f===
====Ramanujan f====
  [https://oeis.org/A121373 A121373] 1, 1, -1, 0, 0, -1, 0, -1, 0, 0, 0, 0, -1, 0, 0, 1
  [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)
  a(n) = (-1)^n * [https://oeis.org/A010815 A010815](n)
Line 56: Line 57:
  eps P="[5,1;6,1;7,1;210,1]" gives 18 + 210 = 228 and pqf = q^(-228/24) = -9 -1/2.  
  eps P="[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.
The sequence is shifted right by 9 positions in this case.
===Normalization of an EPS===
For long lists of generating functions it is convenient to require all EPSs 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.
===Exponentiation of an EPS===
An EPS is taken to a power ''m'' by simply multiplying all ''q<sub>i</sub>'' (the first 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>''.
===Multiplication of two EPSs===
Two EPSs could be multiplied by simply concatenating their list of pairs. It is convenient, howver, to combine all occurrences of specific ''q<sub>i</sub>''
by adding the corresponding eta exponents ''e''.

Revision as of 21:56, 23 January 2023

Eta product signatures (EPS)

The coefficient sequences for generating functions that are products of the 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 qi are the powers of q and the ei are the powers of the eta functions. The latter can be negative.

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]

Dedkind eta η (without the q^(1/24) factor)

A010815 1, -1, -1, 0, 0, 1, 0, 1, 0, 0, 0, 0, -1, 0, 0, -1, 0, 0, 0, 0, 0, 0, 1
eta(q)
Euler transform of period 1: [-1]
eps P="[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.
eta(q^2)^5 / (eta(q)*eta(q^4))^2
Euler transform of period 4: [2,-3,2,-1]
eps P="[2,5;1,-2;4,-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.
eta(q)^2 / eta(q^2)
Euler transform of period 2: [2,-1]
eps P="[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.
q^(-1/8) * eta(q^2)^2 / eta(q)
Euler transform of period 2: [1,-1]
eps P="[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
q^(1/24) * eta(q^2)^2 /(eta(q) * eta(q^4))
Euler transform of period 4: [1,-1,1,0]
eps P="[2,2;1,-1;4,-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)
q^(-1/24) * eta(q^2)^3 / (eta(q) * eta(q^4))
Euler transform of period 4: [[1,-2,1,-1]
eps P="[2,3;1,-1;4,-1]"

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.

Expansion of q^(-1/6) * eta(q^4) * eta(q^6)^5 / (eta(q^3) * eta(q^12))^2
eps P="[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).

q^(-9/8) * (eta(q^2) * eta(q^16))^2 / (eta(q) * eta(q^8))
eps P="[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
q^(-1/2) * eta(q^5) * eta(q^6) * eta(q^7) * eta(q^210)
eps P="[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.

Normalization of an EPS

For long lists of generating functions it is convenient to require all EPSs 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 ei comes before the sublist with negative ei.
  • Inside a sublist the first components qi are in descending order.
  • Successive pairs with the same qi are combined into a single pair with the sum of the ei as second component.

Exponentiation of an EPS

An EPS is taken to a power m by simply multiplying all qi (the first 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.

Multiplication of two EPSs

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