PSLQ - Maple Help
For the best experience, we recommend viewing online help using Google Chrome or Microsoft Edge.

Online Help

All Products    Maple    MapleSim


IntegerRelations

  

PSLQ

  

find an integer dependence (relation)

 

Calling Sequence

Parameters

Description

Examples

Calling Sequence

PSLQ(v)

Parameters

v

-

list or Vector of (complex) floating-point numbers

Description

• 

Given a list (or a Vector) v of n real numbers, the PSLQ(v) command outputs a list (or a Vector) u of n integers such that i=1nuivi is minimized.  Thus the PSLQ function finds an integer relation between a vector of linearly dependent real numbers if the input has enough precision.

• 

Given a list (or a Vector) v of n complex numbers, the PSLQ(v) command outputs a list (or a vector) u of n complex integers (Gaussian integers) such that the norm of i=1nuivi is minimized.

• 

This is an implementation of Bailey and Ferguson's PSLQ algorithm. You can also use the Lenstra-Lenstra-Lovasz (LLL) lattice reduction algorithm to find a linear relation.  For more information, see IntegerRelations[LLL]. Generally speaking, PSLQ is faster.

• 

One application of PSLQ is to find the minimal polynomial of an algebraic number given a decimal approximation of the algebraic number.  The examples below illustrate this.  Generally speaking, if the height of the minimal polynomial is m and its degree is n, then you need more than nlog10m correct decimal digits for the algebraic number.  If the input has d digits and this is insufficient, the output of PSLQ will typically be a list of n integers each with approximately d/n digits long.

• 

The internal working precision of the PSLQ command corresponds to the value of Digits. For best results, the same value of Digits should be used with which the input approximation was obtained.

Examples

withIntegerRelations:

v1.570796326,1.414213562,1.101048032

v1.570796326,1.414213562,−1.101048032

(1)

uPSLQv

u−2,3,1

(2)

Check that the following linear combination is small.

u1v1+u2v2+u3v3

2.×10−9

(3)

Finding a integer relation between non-algebraic constants.

LogsVectorlog2,log3,log5,log6,log10

Logsln2ln3ln5ln6ln10

(4)

uPSLQLogs

u1010−1

(5)

RelationaddLogsiui,i=1..5=0

Relationln2+ln5ln10=0

(6)

Using PSLQ to find the minimal polynomial for 2+3.

rsqrt2+sqrt3

r2+3

(7)

vexpandseqri,i=0..4

v1,2+3,5+223,112+93,49+2023

(8)

Approximate with 12 digits and round to 10 digits.

vevalfv,12

v1.,3.14626436994,9.89897948556,31.1448064542,97.9897948556

(9)

vevalfv

v1.,3.146264370,9.898979486,31.14480645,97.98979486

(10)

uPSLQv

u1,0,−10,0,1

(11)

adduivi,i=1..5

0.

(12)

The minimal polynomial for r

madduizi1,i=1..5

mz410z2+1

(13)

Check that r is a root of mz

simplifyevalm,z=r

0

(14)

The next example involves complex numbers. First define a tenth root of unity.

rcosπ5+Isinπ5

rcosπ5+Isinπ5

(15)

aevalfr

a0.8090169943+0.5877852524I

(16)

v1,a,a2,a3,a4:

uPSLQv

u1,−1,1,−1,1

(17)

adduivi,i=1..5

−1.×10−103.×10−10I

(18)

maddui+1zi,i=0..4

mz4z3+z2z+1

(19)

simplifyevalm,z=r

0

(20)

In the next example, a Gaussian integer relation is found. We subsequently find an integer relation from the Gaussian integer relation by eliminating I.

rsqrt1+2I

r1+2I

(21)

vevalf1,r,r2

v1.,1.272019650+0.7861513778I,1.+2.I

(22)

uPSLQv

u1+2I,0,−1

(23)

mu1+u2x+u3x2

mx2+1+2I

(24)

evalm,x=r

0

(25)

msubsI=z,m

mx2+2z+1

(26)

mresultantm,z2+1,z

mx42x2+5

(27)

evalm,x=r

0

(28)

The last example is of much larger degree requiring more than the default 10 digits of precision.  In the example, we are using PSLQ to test if the algebraic number r is of degree 10 or less.

r1+213313

r1+213313

(29)

Digits50

Digits50

(30)

interfacertablesize=11:

vVectorseqri,i=0..10:

Compute to 55 digits and round to 50 digits.

vevalfv,55:

vevalfv

v1.0.817671479587464782445572296498118762178382211202160.668586648530753836390483543971910166229642164394990.546684234136565776411464312109300969376621272021810.447008106593585761052607258369896222261607897925050.365505779905968441244634024080702057868267278201940.298863651853483469754883483854486311766559962747220.244372284405950790233523271912297540143492649866490.199816247360382529945733129575962875449437226729790.163384046624778837551147180404856597118847153613490.13359447514467024355385019361427542093983423064155

(31)

uPSLQv

u−1623240−2971082727−4527−81

(32)

adduivi,i=1..11

−1.75×10−48

(33)

madduizi1,i=1..11

mz108z9+27z845z7+27z6+27z5+108z4297z3+324z162

(34)

Check.

simplifyevalm,z=r

0

(35)

factorm

z+1z99z8+36z781z6+108z581z4+189z3486z2+486z162

(36)

Thus, the minimal polynomial for r must be the degree 9 factor.

Here is what happens if we mistakenly assume that algebraic number r is of degree 6 or less. The output of PSLQ looks like random 7 digit integers, which indicates that it has not found anything interesting.

uPSLQv1..7

u6336266−44918556130884−120731163103150−60126782169170

(37)

adduivi,i=1..7

1.99×10−42

(38)

See Also

identify

IntegerRelations

IntegerRelations[LinearDependency]

IntegerRelations[LLL]