Error after sympy lambdify function update using vector() dot.class

Niklas Troedsson niklastroedsson at yahoo.com
Fri Jul 11 19:33:14 CEST 2014


I am new to the forum and programming Python.

Recently I updated both Canopy and its packages, I am now using Canopy 1.4.1.1975 and sympy 0.7.5-1.

In an earlier version my code to solve algebra and substitute a lot of constants and transfer functions with more than 1001 frequency points worked. But after the package update to the versions above, lambdify function stopped working and shows an invalid syntax error, pointing to the "dot" in alg.const1.

Can anybody help to get around this without changing the original code of using alg.const1, instead of the cumbersome fix of renaming all variables and make them single variable names such as alg_const1.

%run testofdotclasslambdify.py
  File "<string>", line 1
    lambda alg.npnt1,alg.npnt2,alg.ntf1,alg.const1: (alg.const1*(alg.npnt2 + alg.ntf1)/(alg.const1 + alg.npnt2 + alg.ntf1))
              ^
SyntaxError: invalid syntax
 
I have attached the *.py code and copied in below.

Thank you!

testofdotclasslambdify.py
.

import numpy as np
import pylab as pl
import sympy as sy

# empty class for variables
class vector:
pass
#

# frequency, radians and Laplace representation
fqaxst = 1.0e2
fqaxsp = 100.0e6
fqnop  = 1001
fq     = np.logspace(np.log10(fqaxst),np.log10(fqaxsp),fqnop)
wq     = 2*np.pi*fq
s      = 1.0j*wq

ntf1   = np.exp(-s*1.0/10e6)

npnt1  = 1
npnt2  = 1
const1 = 1

### solve #1 with lambdify 
#   set equation, variables and constants to solve
alg_phiout, alg_npnt1, alg_npnt2 = sy.symbols('alg_phiout, alg_npnt1, alg_npnt2')
alg_ntf1, alg_const1             = sy.symbols('alg_ntf1, alg_const1', complex = True)

subst1       = (alg_npnt1, alg_npnt2, alg_ntf1, alg_const1)

sol          = vector() 
sol.algnpnt1 = (alg_npnt1-alg_phiout/alg_const1)*(alg_ntf1+alg_npnt2)-alg_phiout
sol.solnpnt1 = sy.simplify(sy.diff(sy.solve(sol.algnpnt1, alg_phiout)[0], alg_npnt1,1))
sol.refnum1  = sy.lambdify(subst1, sol.solnpnt1)
tf1          = sol.refnum1(npnt1, npnt2, ntf1, const1)

# plot
pl.figure(1)
pl.semilogx(fq,10*np.log10(np.abs(tf1)**2),label='tf1')
pl.legend(loc='lower right')
pl.title('TF')
pl.xlabel('Frequency, Hz')
pl.ylabel('Spectrum, dB')
pl.grid(b='on')
pl.show()

### solve #2 with lambdify using vector as dot-class variables
#   problem with code below, which wasn't a probelm before updating Canopy package manager 2014-July-01
#   I do not want all variables as signle variables but bundled in a vector dot.class

# set equation, variables and constants to solve
alg = vector()

alg.phiout, alg.npnt1, alg.npnt2 = sy.symbols('alg.phiout, alg.npnt1, alg.npnt2')
alg.ntf1, alg.const1             = sy.symbols('alg.ntf1, alg.const1', complex = True)

subst2       = (alg.npnt1, alg.npnt2, alg.ntf1, alg.const1)
sol.algnpnt2 = (alg.npnt1-alg.phiout/alg.const1)*(alg.ntf1+alg.npnt2)-alg.phiout
sol.solnpnt2 = sy.simplify(sy.diff(sy.solve(sol.algnpnt2, alg.phiout)[0], alg.npnt1,1))
sol.refnum2  = sy.lambdify(subst2, sol.solnpnt2)

# problem with new version of lambdify in new sympy, old version of lambdify/sympy worked
# Gives the following error code:
tf2          = sol.refnum2(npnt1, npnt2, ntf1, const1)
# tf2        = sol.refnum2(npnt1, npnt2, ntf1, const1)
#   In [10]: %run testofdotclasslambdify.py
#    File "<string>", line 1
#      lambda alg.npnt1,alg.npnt2,alg.ntf1,alg.const1: (alg.const1*(alg.npnt2 + alg.ntf1)/(alg.const1 + alg.npnt2 + alg.ntf1))
#               ^
#   SyntaxError: invalid syntax
# In [11]:

# plot
pl.figure(2)
pl.semilogx(fq,10*np.log10(np.abs(tf1)**2),label='tf1')
pl.legend(loc='lower right')
pl.title('TF')
pl.xlabel('Frequency, Hz')
pl.ylabel('Spectrum, dB')
pl.grid(b='on')
pl.show()
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-list/attachments/20140711/b893a10f/attachment.html>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: testofdotclasslambdify.py
URL: <http://mail.python.org/pipermail/python-list/attachments/20140711/b893a10f/attachment.ksh>


More information about the Python-list mailing list