Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

associatedLegendre.jl #24

Open
wants to merge 7 commits into
base: master
Choose a base branch
from
Open

Conversation

elaineVRC
Copy link

Julia file that calculates some associated Legendre functions

Matlab doesn't calculate associated Legendre type 3 (z>1) function but does do type 2 (-1<=x<=1) approx. .0007 sec compared to average .00014sec  recNM3(n,m,z) which is a recursive implementation.
Julia file that calculates some associated Legendre functions
On line 46 changed 1 to one(x)
On line 84 changed 1.  to one(z)
for arbitrary precision, etc.
@stevengj
Copy link
Member

stevengj commented Apr 4, 2017

To properly evaluate a PR, it is essential to have:

  • Tests
  • Properly formatted code. (Use a programming text editor that knows how to indent Julia code. e.g. I recommend Atom with the Julia plugins, or equivalently Juno or JuliaPro, but tastes vary.)
  • Documentation (docstrings etcetera).
  • Benchmarks against other implementations (e.g. GSL or SciPy.)

Also, I wouldn't submit Jupyter notebooks.

elaineVRC added 2 commits May 22, 2017 17:37
Thanks for editting and making the PR more readable; congratulations on your new position AA.  Thanks to SJ for mentoring and encouragement.  
I can't benchmark against what does not exist: matlab, maxima, scipy.special, gsl do not have associated Legendre type 3 (z > 1).  I have used SLegendreP3 (which I derived and used a long time ago) ; similarly SLegendreP2  gives reliable results compared to scipy.special   for |x| <1 .  function BenchmarkZZLegendreP2 , shows that ZZLegendreP2  is faster than scipy.special.  ZZLegendreP3 comes from a modification of work published by Selezneva, etal. 
If the general direction of the PR is OK; then I can extract all the test... functions to test.jl and do additional work.  All suggestions, comments,etc. are welcome.  Thank you for helping me to learn Julia scientific computing.
<
As requested, @Btime from BenchmarkTools and @time both gave similar results in that ZZLegendreP2 is light years ahead of scipy.special.lpmv.  Accuracy of ZZLegendreP2 ( |z|<=1) against lpmv and ZZLegendreP3 (z > 1) against Fortran table is good.
@elaineVRC
Copy link
Author

elaineVRC commented Aug 18, 2017

# solutions, w, of the differential equation are 
#    associated Legendre functions   
# (1-x*x)d/dx(dw/dx)�-2�x�(dw/dx)+(n(n+1)-m*m/(1-x*x))w=0
#n is the degree and m is the order.
#for m =0 they are a.k.a. Legendre polynomials
# see https://github.com/pjabardo/Jacobi.jl/blob/master/src/legendre.jl
# for m=0 also.
#
#       SSLegendreP2 and 3 : The finite sum I derived is arranged so that 
# each term is a multiple of the preceeding term. 
#
# LegendreP2 uses the recursion equation 8.5.3 Abramowitz & Stegun (AS) 
# https://www.math.hkbu.edu.hk/support/aands/frameindex.htm
#Handbook of Mathematical Functions  is
#  (n-m+1)P(n+1,m,z)=(2n+1)zP(n,m,z)- (n+m)P(n-1,m,z)
# By normalizing so that that P*P integrated is unity, the new P denoted 
# by PP satisfies an altered 3 term recursion. This is discussed in 
# Numerical Recipes 3rd ed. see https://github.com/milthorpe/
#   SphericalHarmonics.jl/blob/master/src/SphericalHarmonics.jl
# T.Fukushima has Fortran codes (https://static-content.springer.
# com/esm/art%3A10.1007%2Fs00190-011-0519-2/MediaObjects
# /190_2011_519_MOESM_ESM.txt);JamesBremerJr(github) and arXiv:1707.03287
#
#  ZZLegendreP2 and 3  uses a different recursion
#  P(n,m+2,x)+2(m+1)x(s(1-x*x))^(-1/2)P(n,m+1,x)+s(n-m)(n+m+1)P(n,m,x)=0
#from http://dlmf.nist.gov/14.10  equation  14.10.1  where s=sign(1-x*x)
#refer to Lebedev, Special Functions, Dover Pub. eqn. 7.12.8 page 194
#             See function BenchmarkZZLegemdreP2() for speed.
#
#These recursions are valid for Legendre functions of the first kind 
#(P(n,m,z))and second kind Q(n,m,z) , 
#(for types 1,2,3 ,definition from Mathematica documentation
# type 1 (?) is m=0, type 2 is -1<=x<=1 and type 3 is z> 1
# http://mpmath.org/doc/0.18/functions/orthogonal.html
#Thus we define functions  LegendreP2,....P3,....Q2,...Q3  
# for n and m positive integers 0<=m<=n
# refer to http://dlmf.nist.gov/14.6
#P(n,m,z) and Q(n,m,�z) (z> 1) are often referred to as the prolate
#spheroidal harmonics of the first and second kinds, respectively
#  We can define spherical harmonics (there are other definitions)
# Y(n,m,theta,phi)= LegendreP2(n,|m|,theta)*exp(im*m*phi)*((-)^(|m|-m)/2))
# *sqrt((2n+1)*factorial(n-|m|)/(4*pi*factorial(n+|m|))) 
# refer to https://en.wikipedia.org/wiki/Spherical_harmonics



"""
    LegendreP2(n::Integer,m::Integer,x::Number)
    associated Legendre function of first kind type 2 (-1.<=x<=1.).
    

"""

function LegendreP2(n::Integer,m::Integer,x::Number) 
  
     
    # 0 <= m  <= n
    # -1<=x <= 1
    M = (2*m -1) # M must be odd
    # (2m-1)!!= (2m)!/( m! 2^m)   double factorial 
    dblfac=one(x)
    for j=1:2:M    
        dblfac=j*dblfac  
    end
   
    pj2= ((-1)^m)*dblfac*(one(x)-x*x)^(m/2)   
    pj1=x*(2.*m+1.)*pj2  
    if n == m
        return pj2
    elseif n == m+1
        return pj1
    end
    

    for j = m+2 :n 
        pjj=(x*(2.*j-1.)*pj1 - pj2*(j +m-1.)) /(j-m)
        pj2=pj1
        pj1=pjj
    end
return pj1 
end




using BenchmarkTools

using PyCall
pyimport_conda("scipy.special","scipy") ####
@pyimport scipy.special as s


function timelpmv()
    for k=1:10^6
        k2=k-1
        x=.1 +k2*.9e-6
         s.lpmv(15,30,x)
    end
end

@btime timelpmv()     

function timeLegendreP2()
    for k=1:10^6
        k2=k-1
        x2= .1 + k2*.9e-6
    LegendreP2(30,15,x2)
    end
end

@btime timeLegendreP2()


"""
    SSLegendreP2(n::Integer,m::Integer,x::Number)
    associated Legendre function 1st kind type 2 , -1.<=x<=1.
    each term in sum is multiple of previous, from SLegendreP2 


"""

function SSLegendreP2(n::Integer,m::Integer,x::Number)#n,m integers 
   
    M = (2*n -1) # M must be odd
    # (2m-1)!!= (2m)!/( m! 2^m)   double factorial 

    DT=one(x) #1.0
    if n > 0
        for j=1:M
            if  isodd(j)
                DT=DT*j
            end
            if  j <= (n-m)
                DT=DT/j
            end
        end  
    end
    MXP=div(m+n,2)
    
    sum= (x^(n+m))*DT/((one(x)-x*x)^(m/2))
    prev=sum
    for p=1:MXP
        term=(-prev)*(n-p+1)*(n+m-2*p+1)*(n+m-2*p+2
        )/(p*(2*n-2*p+1)*(2*n-2*p+2)*x*x)
        sum=sum+term
        prev=term
    end
    return sum
end



"""
    ZZLegendreP2(n::Integer,m::Integer,z::Number)
    associated Legendre function First kind type 2 |x|<=1.
    uses different three term recursion
"""

function ZZLegendreP2(n::Integer,m::Integer,z::Number)    
    nz=1
    pnm = zeros(typeof(z),2*n+1)  
    #fac = prod(2.:n)
    fac= one(z) #1.0
    for k=1:n
        fac=fac*k
    end
    sqz2 = sqrt((one(z)-z.*z))
    hsqz2 = 0.5*sqz2
    ihsqz2 = z./hsqz2
    if(n==0)
        pnm[1]=one(z)
        return ((-one(z))^m)*pnm[n+1+m] #pnm
    end
    if(n==1)
        pnm[1]=-.5*sqz2
        pnm[2]=z 
        pnm[3]=sqz2
        return ((-one(z))^m)*pnm[n+1+m]  #pnm
    end
    pnm[1] = (1-2*abs(n-2*floor(n/2)))*hsqz2.^n/fac
    pnm[2] = -pnm[1]*n.*ihsqz2
    for mr=1:2*n-1
        pnm[mr+2]=(mr-n).*ihsqz2.*pnm[mr+1]-(2*n-mr+1)*mr*pnm[mr]
    end
    return ((-one(z))^m)*pnm[n+1+m]
end
 

#checking LegendreP2 against scipy.special
#   for n< 18 the relative error is less than 2e-14
#     for n < 45 rel error  less than 2e-13
#     for n < 62 the rel.error is less than 5e-12
#      for n < 75  the rel.error is less than 5e-11
# the error depends on the n amd m values and x values to s(ome extent

using PyCall
pyimport_conda("scipy.special","scipy") ####

@pyimport scipy.special as s

"""
    function testLegendreP2()
    relative error LegendreP2 versus scipy special
"""


function testLegendreP2()
    x=.6
    rerr = 5e-11
    for n=0:75 #17
        for m=0:n      #m=rand(0:n)
            a=LegendreP2(n,m,x)
            b=s.lpmv(m,n,x)
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            #if (norm(a-b)/norm(b)) > rerr)
            println(n," ", m," ", a," ",b," ",df)
            #@test_approx_eq_eps a b 1e-14
            end
        end
    
    end

end
 
testLegendreP2()


"""
    function testLegendreP2F()
    compare LegendreP2 against Float64(SSLegendreP2)
"""
function testLegendreP2F()
#=
   checking LegendreP2 against Float64SSLegendreP2
   for n < 18 the relative error is less than 7e-15
    error depends on the n amd m values and 
        x values to some extent
 =#
    

    x=.6
    rerr = 7e-15
    for n=1:75
        if n > 18
            rerr=2e-13
        end
        
            
        if n > 44 
            rerr= 5e-12
        end
        if n > 62 
            rerr= 5e-11
        end
        for  m=0:n   #rand(1:n)
            a=LegendreP2(n,m,x)
            b=Float64(SSLegendreP2(n,m,big(x)))
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            println(n," ", m," ", a," ",b," ",df)
            end
        end
    end        
    
end
testLegendreP2F()


#checking SSLegendreP2 against Float64(SSLegendreP2)
#   for n< 17 the relative error is less than 3e-11
# the error depends on the n amd m values and x values to some extent
"""
    function testSSLegendreP2F()
    compare SSLegendreP2 to float64SSLegendreP2
"""
function testSSLegendreP2F()
    x=.6
    rerr = 3e-11
    for n=0:17
        for m=0:n    #m=rand(0:n)
            a=SSLegendreP2(n,m,x)
            b=Float64(SSLegendreP2(n,m,big(x)))
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            println(n," ", m," ", a," ",b," ",df)
            end
        end
    end
end
testSSLegendreP2F()



#checking SSLegendreP2 against scipy.special
#   for n< 17 the relative error is less than 3e-11
# the error depends on the n amd m values and x values to some extent

using PyCall

@pyimport scipy.special as s

"""
    function testSSLegendreP2()
    compare SSLegendreP2 with scipy special
"""
function testSSLegendreP2()
    x=.6
    rerr = 3e-11
    for n=1:17
        for m=0:n      #m=rand(1:n)
            a=SSLegendreP2(n,m,x)
            b=s.lpmv(m,n,x)
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            #if (norm(a-b)/norm(b)) > rerr)
            println(n," ", m," ", a," ",b," ",df)
            #@test_approx_eq_eps a b 1e-14
            end
        end
    
    end

end
testSSLegendreP2()

#checking SSLegendreP2 against Float64(SSLegendreP2)
#   for n< 17 the relative error is less than 3e-11
# the error depends on the n amd m values and x values to some extent
"""
    function testSSLegendreP2F()
    compare SSLegendreP2 to float64SSLegendreP2
"""
function testSSLegendreP2F()
    x=.6
    rerr = 3e-11
    for n=0:17
        for m=0:n    #m=rand(0:n)
            a=SSLegendreP2(n,m,x)
            b=Float64(SSLegendreP2(n,m,big(x)))
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            println(n," ", m," ", a," ",b," ",df)
            end
        end
    end
end
testSSLegendreP2F()

# derived from work published at 
# i. a. selezneva, yu. l. ratis, e. hernández, j. pérez-quiles 
#and p. fernández de córdoba:
#a code to calculate high order legendre polynomials
#rev. acad. colomb. cienc.: volumen xxxvii, número 145 - diciembre 2013
#www.scielo.org.co/pdf/racefn/v37n145/v37n145a09.pdf   
    #Selezneva.....pdf    
       

"""
ZZLegendreP3(n::Integer,m::Integer,z::Number)
    associated Legendre function First kind type 3 ( z > 1.)
    uses different three term recursion 
"""

function ZZLegendreP3(n::Integer,m::Integer,z::Number)  
   
    nz=1
    pnm = zeros(typeof(z),2*n+1)#was nz,n+1
    #fac = prod(2.:n)
    fac= one(z) #1.0
    for k=1:n
        fac=fac*k
    end
    sqz2 = sqrt((z.*z - one(z)))
    hsqz2 = 0.5*sqz2
    ihsqz2 = z./hsqz2
    if(n==0)
        pnm[1]=one(z)
        return  ((-one(z))^m)*pnm[n+1+m]  
    end
    if(n==1)
        pnm[1]=-.5*sqz2
        pnm[2]=z
        pnm[3]=-sqz2 # sign
        return   ((-one(z))^m)*pnm[n+1+m] 
    end
    pnm[1] = (1-2*abs(n-2*floor(n/2)))*hsqz2.^n/fac
    pnm[2] = -pnm[1]*n.*ihsqz2
    for mr=1:2*n-1
        pnm[mr+2]=(mr-n).*ihsqz2.*pnm[mr+1]+(2*n-mr+1)*mr*pnm[mr] 
        
    end
    return ((-one(z))^m)*pnm[n+1+m]
end
 


   
 """
    SSLegendreP3(n::Integer,m::Integer,x::Number)
    associated Legendre function 1st kind type 3 , x >1.
    each term in sum is multiple of previous. 

"""

function SSLegendreP3(n::Integer,m::Integer,x::Number)#n,m integers 
    
   
    M = (2*n -1) # M must be odd
    # (2m-1)!!= (2m)!/( m! 2^m)   double factorial 
   # dblfac=1
    DT=one(x)   #1.0
    if n > 0
        for j=1:M
            if  isodd(j)
                DT=DT*j
            end    
            if j <= (n-m)
                DT=DT/j
            end
        end
    end
    
    MXP=div(m+n,2)
    sum= (x^(n+m))*DT/((x*x - one(x))^(m/2))
    prev=sum
    for p=1:MXP
        term=(-prev)*(n-p+1)*(n+m-2*p+1)*(n+m-2*p+2
        )/(p*(2*n-2*p+1)*(2*n-2*p+2)*x*x)
        sum=sum+term
        prev=term
    end
    return sum
end


         



 

 


#A representation of associated Legendre function was derived from 8.6.6,
#    8.6.18,8.2.5,3.1.1 in AS (https://www.math.hkbu.edu.hk
#        /support/aands/frameindex.htm). 
#    14.7.8,14.7.10,14.7.11,14.7.14.1.2.2 in DLMF(dlmf.nist.gov),
#similar formula for type 2.
#(8.6.18 AS) Rodrigues' formula for integer n
#  P(n,z)=(1/((2^n)n!)) (d/dz)^n (z^2 - 1)^n
#(8.6.6 AS) P(n,m,z)= ((z^2 -1)^(m/2))(d/dz)^m P(n,z) for z > 1
#   P(n,m,x) = ((-)^m)((1-x^2)^(m/2))(d/dz)^m P(n,x)  for |x| <= 1   
# expand using binomial theorem where a=z^2 and b = -1
#(3.1.1 AS) define binomial coefficient = C(n,k) = n!/((n-k)! k!)
# defining  n!=n*(n-1)*(n-2)*...3*2*1 = factorial(n)
# (a + b)^n = a^n  + C(n,1)*(a^(n-1))*b  + C(n,2)*(a^(n-1))*b^2 
#      +C(n,3)*(a^(n-3))*b^3 +.......+b^n  where n is a positive integer
#differentiate to obtain the finite sum expression



#Benchmarking ZZLegendreP2 against scipy.special

using PyCall
using BenchmarkTools
@pyimport scipy.special as s


function timePY()
    for i=1:100
        s.lpmv(15,30,.6) 
    end
end 

function timeZ2()
    for i=1:100
        ZZLegendreP2(30,15,.6)
    end
end 
"""
    function BenchmarkZZLegendreP2()
    timing for ZZLegendreP2 LegendreP2 
    more than five times as fast 
    compared to s.lpmv from scipy.special
    speed may depend on n,m,x values
"""

function BenchmarkZZLegendreP2()

TPY= @elapsed(timePY())
println("time s.lpmv(15,30,.6)   $TPY")


TZ2= @elapsed(timeZ2())
println("time ZZLegendreP2(30,15,.6)  $TZ2")

#time s.lpmv(15,30,.6)   0.000674941
#time ZZLegendreP2(30,15,.6)  0.000112901
    
end
BenchmarkZZLegendreP2()

function timeZZ2()
    for k=1:10^6
        k2=k-1
        x2=.1 +k2*.9e-6
        ZZLegendreP2(30,15,x2)
    end
end

function LPMV2()
    for k=1:10^6
        k2=k-1
        x2=.1 +k2*.9e-6
        s.lpmv(15,30,x2)
    end
end

println("lpmv time")
@btime LPMV2()
println("ZZLegendreP2 time")
@btime timeZZ2()

#=
    for k=1:10^6
        k2=k-1
        x2=.1 +k2*.9e-6
        s.lpmv(15,30,x2)
    end


println("lpmv time")
@btime LPMV2()
println("ZZLegendreP2 time")
@btime timeZZ2()

=#

        



        



#checking ZZLegendreP2 against scipy special
#   for n< 41 the relative error is less than 2.4e-9
#    for n < 61 the relative error is less than 6e-5
# the error depends on the n amd m values and x values to some extent
#
using PyCall
@pyimport scipy.special as s

"""
    function testZZLegendreP2(
    compare ZZLegendreP2 with scipy special
"""
function testZZLegendreP2()
    x=.6
    rerr = 2.4e-9# 5e-7
    #for n =51: 80
    for n=1:40
        for m= 0:n  #rand(1:n)
            a=ZZLegendreP2(n,m,x)
            b=s.lpmv(m,n,x)
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            println(n," ", m," ", a," ",b," ",df)
            end
        end
    end
end
testZZLegendreP2()


#checking ZZLegendreP2 against Float64(SSLegendreP2)
#   for n< 41 the relative error is less than 2.4e-9
#    for n <61 the relative error is less than 6e-5
# the error depends on the n amd m values and x values to some extent
#  largest errors for n=m or n= m+1 (could be fixed?)
"""
    function testZZLegendreP2F()
    compare ZZLegendreP2 with Float64 SSLegendreP2
"""
function testZZLegendreP2F()
    x=.6
    rerr = 2.4e-9 
    for n=1:40
        for m= 0:n  #rand(1:n)
            a=ZZLegendreP2(n,m,x)
            b=Float64(SSLegendreP2(n,m,big(x)))
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            println(n," ", m," ", a," ",b," ",df)
            end
        end
    end
end
testZZLegendreP2F()


# derived from work published at 
# i. a. selezneva, yu. l. ratis, e. hernández, j. pérez-quiles 
#and p. fernández de córdoba:
#a code to calculate high order legendre polynomials
#rev. acad. colomb. cienc.: volumen xxxvii, número 145 - diciembre 2013
#www.scielo.org.co/pdf/racefn/v37n145/v37n145a09.pdf   
    #Selezneva.....pdf    
       

"""
    ZZLegendreP3(n::Integer,m::Integer,z::Number)
    associated Legendre function First kind type 3 ( z > 1.)
    uses different three term recursion 
"""

function ZZLegendreP3(n::Integer,m::Integer,z::Number)  
   
    nz=1
    pnm = zeros(typeof(z),2*n+1)#was nz,n+1
    #fac = prod(2.:n)
    fac= one(z) #1.0
    for k=1:n
        fac=fac*k
    end
    sqz2 = sqrt((z.*z - one(z)))
    hsqz2 = 0.5*sqz2
    ihsqz2 = z./hsqz2
    if(n==0)
        pnm[1]=one(z)
        return  ((-one(z))^m)*pnm[n+1+m]  
    end
    if(n==1)
        pnm[1]=-.5*sqz2
        pnm[2]=z
        pnm[3]=-sqz2 # sign
        return   ((-one(z))^m)*pnm[n+1+m] 
    end
    pnm[1] = (1-2*abs(n-2*floor(n/2)))*hsqz2.^n/fac
    pnm[2] = -pnm[1]*n.*ihsqz2
    for mr=1:2*n-1
        pnm[mr+2]=(mr-n).*ihsqz2.*pnm[mr+1]+(2*n-mr+1)*mr*pnm[mr] 
        
    end
    return ((-one(z))^m)*pnm[n+1+m]
end
 




#checking ZZLegendreP3 against Float64(SSLegendreP3)
#   for n< 17 the relative error is less than 6.4e-7 for z=2.
#    for n<24 the relative error is less than .0054
# for z=3. n<17 the relative eror is less than 2e-7
# the error depends on the n amd m values and x values to some extent
#n=1 may have error
"""
    function testZZLegendreP3F()
    compare ZZLegendreP3 with float64 SSLegendreP3
"""
function testZZLegendreP3F()
    x=3.#2.
    rerr = 2e-7 #6.4e-7
    for n=1:16
        for m=0:n  #rand(1:n)
            a=ZZLegendreP3(n,m,x)
            b=Float64(SSLegendreP3(n,m,big(x)))
            df=norm(a-b)/norm(b) 
            if ( df > rerr)
            println(n," ", m," ", a," ",b," ",df)
            end
        end
    end
end
testZZLegendreP3F()






using PyCall
@pyimport scipy.special as s

"""
    function ZZSS2(n,m,x,ESTOP)
    Values near .99999 < x <= 1.000000 and -1.000000 <x<-.99999
    causing problems
    Identify n,m, x ,when zz is inaccurate

"""
function ZZSS2(n,m,x,ESTOP)
    valueZZ=Float64(ZZLegendreP2(n,m,big(x)))
    #valueS=Float64(SLegendreP2(n,m,big(x)))
    valueSS=Float64(SSLegendreP2(n,m,big(x)))
    valuezz=ZZLegendreP2(n,m,x)
    #valuess=SSLegendreP2(n,m,x)
    valuepy=s.lpmv(m,n,x)
    prerr=abs((valuezz-valuepy)/valuepy) 
    zzrerr=abs((valuezz-valueZZ)/valueZZ)
    if zzrerr > ESTOP
        println()
        println("n=$n, m=$m , x=$x , zz=$valuezz")
        println("SS=$valueSS,ZZ=$valueZZ, py=$valuepy ")
        println(" zzrerr= $zzrerr, prerr= $prerr")
    end 
    
end


ESTOP=.000000000001

n=10
for ix=1:10
    x=1. - (.1)^ix
        for m=0:n
            ZZSS2(n,m,x,ESTOP)
        end
end     
 
n=10
for ix=1:10
    x=-1. + (.1)^ix
        for m=0:n
            ZZSS2(n,m,x,ESTOP)
        end
end










 

"""
    function SSZZ1(n,m,x,ESTOP)
    Values near 1.< x < 1.00001 causing problems
    Identify n,m, x ,when zz goes negative, its underflow
    The cutoff does change with n values
    The user has the option to accept reduced accuracy
    for certain values of n,m,x or to increase the accuracy
    by using higher precision 
"""
function SSZZ1(n,m,x,ESTOP)
    valueZZ=Float64(ZZLegendreP3(n,m,big(x)))#high precision
    #valueS=Float64(SLegendreP3(n,m,big(x)))
    valueSS=Float64(SSLegendreP3(n,m,big(x)))
    valuezz=ZZLegendreP3(n,m,x)
    #valuess=SSLegendreP3(n,m,x)
    zzrerr=abs((valuezz-valueZZ)/valueZZ)
    #ssrerr=abs((valuess-valueSS)/valueSS)
    
    if zzrerr > ESTOP
    #if sign(valuezz) < 1 
        println()
        println("n= $n, m= $m, x= $x")
        println("SS=$valueSS,ZZ=$valueZZ, zz= $valuezz, zzrerr= $zzrerr")
        
        
        #println("S=$valueS,SS=$valueSS,ZZ=$valueZZ, n= $n, m= $m, x= $x")
        #println("  ssrerr= $ssrerr,ss= $valuess,zz= $valuezz, zzrerr= $zzrerr")
        
    end
end 





ESTOP = .000000000001 #e-12

n=10

for ix =1:7   
    for m=0:n
        x=1.+.1^(ix)
       
        SSZZ1(n,m,x,ESTOP)
    end
end

println("  ")

 n=80
ESTOP=.000000000001 # e-12
for ix=1:7
    for m=0:n 
        x= 1.+  (.1)^(ix)   
    
        SSZZ1(n,m,x,ESTOP)
    end
end





"""
    function TableSSZZ(n,m,x,T) 
    compare  SSLegendreP3,ZZLegendreP3 and Table1 value is T
    with and without arbitrary precision to 
    Table 1 from Computer Physics Comm.  181 (2010)2091-7 
    User has the option to accept reduced accuracy for certain
    values of n,m,x or to increase the accuracy by using
    higher values of precision in arbitrary precision arithmetic
"""
function TableSSZZ(n,m,x,T)
    valueZZ=Float64(ZZLegendreP3(n,m,big(x))) #higher precision
    #valueS=Float64(SLegendreP3(n,m,big(x)))
    valueSS=Float64(SSLegendreP3(n,m,big(x)))
    valuezz=ZZLegendreP3(n,m,x)
    #valuess=SSLegendreP3(n,m,x)
    zzrerr=abs((valuezz-T)/T)
    ZZrerr=abs((valueZZ-T)/T)
    println()
    println("n=$n, m=$m, x=$x")
    println("SS=$valueSS,ZZ=$valueZZ, T=$T ,zz=$valuezz " )
   # println("SS=$valueSS,ZZ=$valueZZ, T=$T , n= $n, m= $m, x= $x")
   # println(" ZZrerr=$ZZrerr, ss=$valuess,zz=$valuezz, zzrerr=$zzrerr")
     println(" ZZrerr=$ZZrerr, zzrerr=$zzrerr")
  
end
x=5.0 
m=0
n=0
T=1.0
x=5.0 
m=0
n=0
T=1.0
TableSSZZ(n,m,x,T)
n=1
T=5.0
TableSSZZ(n,m,x,T)
n=2
T=37.0
TableSSZZ(n,m,x,T)
n=3
T=305.
TableSSZZ(n,m,x,T)
n=4
T=2641.
TableSSZZ(n,m,x,T)
m=1
n=1
T=4.898979485566
TableSSZZ(n,m,x,T)
n=2
T=73.4846922835
TableSSZZ(n,m,x,T)
n=3
T=9.112101843153e2
TableSSZZ(n,m,x,T)
n=4
T=1.053280589397e4
TableSSZZ(n,m,x,T)
m=2
n=2
T=72.
TableSSZZ(n,m,x,T)
n=3
T=1800.
TableSSZZ(n,m,x,T)
n=4
T=3.132e4
TableSSZZ(n,m,x,T)
m=3
n=3
T=1.763632614804e3
TableSSZZ(n,m,x,T)
n=4
T=6.172714151814e4
TableSSZZ(n,m,x,T)
m=4
n=4
T=6.048e4
TableSSZZ(n,m,x,T)
x=2.0
m=0
n=0
T=1.0
TableSSZZ(n,m,x,T)
n=1
T=2.0
TableSSZZ(n,m,x,T)
n=2
T=5.5
TableSSZZ(n,m,x,T)
n=3
T=17.
TableSSZZ(n,m,x,T)
n=4
T=55.375
TableSSZZ(n,m,x,T)
m=1
n=1
T=1.732050807569
TableSSZZ(n,m,x,T)
n=2
T=1.039230484541e1
TableSSZZ(n,m,x,T)
n=3
T=4.936344801571e1
TableSSZZ(n,m,x,T)
n=4
T=2.165063509461e2
TableSSZZ(n,m,x,T)
m=2
n=2
T=9.0
TableSSZZ(n,m,x,T)
n=3
T=90.
TableSSZZ(n,m,x,T)
n=4
T=607.5
TableSSZZ(n,m,x,T)
m=3
n=3
T=7.794228634060e1
TableSSZZ(n,m,x,T)
n=4
T=1.091192008768e3
TableSSZZ(n,m,x,T)
m=4
n=4
T=945.0
TableSSZZ(n,m,x,T)
x=1.000001
m=0
n=0
T=1.0
TableSSZZ(n,m,x,T)
n=1
T=1.0000010
TableSSZZ(n,m,x,T)
n=2
T=1.000003000001
TableSSZZ(n,m,x,T)
n=3
T=1.000006000007
TableSSZZ(n,m,x,T)
n=4
T=1.000010000022
TableSSZZ(n,m,x,T)
m=1
n=1
T=1.414213915900e-3
TableSSZZ(n,m,x,T)
n=2
T=4.242645990341e-3
TableSSZZ(n,m,x,T)
n=3
T=8.485304708618e-3
TableSSZZ(n,m,x,T)
n=4
T=1.414220279870e-2
TableSSZZ(n,m,x,T)
m=2
n=2
T=6.000002999773e-6
TableSSZZ(n,m,x,T)
n=3
T=3.000004499888e-5
TableSSZZ(n,m,x,T)
n=4
T=9.000025499681e-5
TableSSZZ(n,m,x,T)
m=3
n=3
T=4.242643868860e-8
TableSSZZ(n,m,x,T)
n=4  #underflow
T=2.969853678052e-7
TableSSZZ(n,m,x,T)
m=4 #underflow
n=4
T=4.200004199683e-10
TableSSZZ(n,m,x,T)







@mzaffalon
Copy link

Isn't pj1=x*(2.*m+1.)*pj2 converting pj1 to Float64 independently on the type of x?

BTW, you should be careful with using 1. to make it a floating point number and you should instead use 1.0, since 1.+2 is now parsed as 1 .+ 2 and not 1.0 + 2 as you probably intend.

@elaineVRC
Copy link
Author

pj1=x*(2.*m+1.)*pj2 If x is BigFloat, then so is pj1

Thanks for reminding me to use 1.0 to avoid mistakes.

@mzaffalon
Copy link

What about if x is Float32?

@elaineVRC
Copy link
Author

Float32 multiplied by Float64 gives Float64. Float32 multiplied by integer gives Float32. So if I change the stuff in parentheses to integers this might be OK. There are about 6 more places that might be affected by Float32 inputs , so I will change and resubmit. Thanks much.

This PR is the file entitled associatedLegendre.jl.  The first part with all the comments is like a readme.  The rest is the source code and numerous tests(requested) and exploration of area near |x| ~ 1.  As requested @Btime from BenchmarkTools as well as @time gave similar results in that ZZLegendreP2 is light years ahead of scipy.special.lpmv.  Accuracy of ZZLegendreP2 (|z|=<1.) against lpmv and ZZLegendreP3 (z>1.) against Fortran is  good.
@mmikhasenko
Copy link

@elaineVRC
what is the status of this PR? From the first comment of @stevengj .
Are there tests, comparison to other implementations already?

@stevengj
Copy link
Member

As I commented in #267, I feel like there should be a separate package for special polynomials, as opposed to transcendental special functions.

@elaineVRC
Copy link
Author

elaineVRC commented Nov 16, 2020 via email

@elaineVRC
Copy link
Author

elaineVRC commented Nov 16, 2020 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants