I am experiencing an error trying to use the interalg functionality.

System:

• OpenOpt 0.34

• Windows 7

• Enthought Python 7.1-1

I ran exactGlobalNLP.py and got the following traceback:

------------------------- OpenOpt 0.34 -------------------------
solver: interalg_0.21   problem: unnamed    type: GLP
 iter   objFunVal   
    0  1.163e+02 
Traceback (most recent call last):
  File "C:\Users\breisfel\Documents\Programming\python\exactGlobalNLP.py", line 34, in <module>
    r = p.minimize(solver)
  File "C:\Python27\lib\site-packages\openopt-0.34-py2.7.egg\openopt\kernel\baseProblem.py", line 196, in <lambda>
    self.minimize = lambda *args, **kwargs: minimize(self, *args, **kwargs)
  File "C:\Python27\lib\site-packages\openopt-0.34-py2.7.egg\openopt\kernel\baseProblem.py", line 774, in minimize
    return runProbSolver(p, *args, **kwargs)
  File "C:\Python27\lib\site-packages\openopt-0.34-py2.7.egg\openopt\kernel\runProbSolver.py", line 238, in runProbSolver
    solver(p)
  File "C:\Python27\lib\site-packages\openopt-0.34-py2.7.egg\openopt\solvers\UkrOpt\interalg_oo.py", line 225, in __solver__
    C = p._FD.nonBoxConsWithTolShift
AttributeError: EmptyClass instance has no attribute 'nonBoxConsWithTolShift'

My name is Guillaume BENOIT from EDF Research and Development and I'm trying to use openopt for internal studies.
I'm using Python 2.6 and I've installed the latest release available (whole suit).
I'm need to maximise function Psi (4 variables : alpha, a0, a1 and b1). Here is the following code :

###################################################

    j = Taille-1    # actually ThetaSto is an existing array correctly defined (12775x1)

    Psi = lambda alpha, a0, a1, b1: math.log(stats.norm.pdf(ThetaSto[j],ThetaSto[j]*math.exp(-alpha),\
                                             (a0 + a1*math.cos(2*pi*j/365) + b1*math.sin(2*pi*j/365))*\
                                              math.sqrt((1-math.exp(-2*alpha))/(2*alpha)))).sum()

    # parameters estimation
    alpha0 = 0.2
    a00 = 2
    a10 = 0.2
    b10 = 0.1
   
    StartPoint = np.array([alpha0, a00, a10, b10])

    p = NLP(Psi, StartPoint) 

    r = p.solve('ralg')

##############################################################

I always get this error message :

  File "c:\python26\lib\site-packages\openopt-0.34-py2.6.egg\openopt\kernel\ooMisc.py", line 200, in setNonLinFuncsNumber
    setattr(p, 'n'+userFunctionType, asfarray(FV(*(X, ) + args, **kwargs)).size)
TypeError: <lambda>() takes exactly 4 arguments (1 given)

Do you have any idea of the origin of the problem ?

Thank you in advance for your help.

Regards,

Guillaume

Here is simple model, in which I try to fit some parameters to experimental data, essentially using least-square model + additional parameter, making it non-linear. It is here presented as one in 3 variables (original have about Zmiennych=200 + 1 for K), but error is the same.

from openopt import GLP, oosolver
from FuncDesigner import *

from numpy import zeros, ones

#K_ = oovars(1)
#K = K_[0]
K = oovar(size=1)
A0 = 2.0e-6
B0 = 5.0e-6

Zmiennych = 2

Theta_DNA = zeros(Zmiennych) # , int/float)
Theta_YY1 = zeros(Zmiennych)
Theta_Total = zeros(Zmiennych)
Theta_DNA_YY1 = oovar(size=Zmiennych)

for i in range(Zmiennych):
    Theta_YY1[i] = 4.5
    Theta_DNA[i] = 8.7
    Theta_Total[i] = 4.1


S_AB = 0.5*(A0 + B0 + K*1.0e-6 - sqrt( (A0 + B0 + K*1.0e-6)**2 - 4.0*A0*B0 ) )
S_A = (A0 - S_AB)
S_B = (B0 - S_AB)

Roznice = 1.0e-3*(S_A*Theta_DNA + S_B*Theta_YY1 + S_AB*Theta_DNA_YY1) - Theta_Total

f = Roznice[0]**2 + Roznice[1]**2
#f = sum(Roznice**2)

constraints = [0 < K, K < 1000]     # Was trying to use <=, but it doesn't help
for i in xrange(Zmiennych):
    constraints.extend([-1.0e6 <= Theta_DNA_YY1[i], Theta_DNA_YY1[i] <= 1.0e6])

startPoint = {K: 0.32, Theta_DNA_YY1: ones(Zmiennych)}

p = GLP(f, startPoint, constraints=constraints)

#solver = oosolver('de')
solver = oosolver('galileo')

r = p.solve(solver, plot=1)

Using de, I got

FuncDesigner warning: Probably scipy installation could speed up running the code involved

------------------------- OpenOpt 0.34 -------------------------
solver: de   problem: unnamed    type: GLP
 iter   objFunVal   log10(maxResidual)   
    0  3.362e+01            -100.00 
OpenOpt Error: this solver requires finite lb, ub: lb <= x <= ub
Traceback (most recent call last):
  File "./bug2.py", line 47, in <module>
    r = p.solve(solver, plot=1)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/baseProblem.py", line 235, in solve
    return runProbSolver(self, *args, **kwargs)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/runProbSolver.py", line 238, in runProbSolver
    solver(p)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/solvers/UkrOpt/de_oo.py", line 74, in __solver__
    if not p.__isFiniteBoxBounded__(): p.err('this solver requires finite lb, ub: lb <= x <= ub')
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/oologfcn.py", line 16, in ooerr
    raise OpenOptException(msg)
oologfcn.OpenOptException: this solver requires finite lb, ub: lb <= x <= ub

and one using galileo

FuncDesigner warning: Probably scipy installation could speed up running the code involved

------------------------- OpenOpt 0.34 -------------------------
solver: galileo   problem: unnamed    type: GLP
OpenOpt Error: the solver galileo cannot handle 'A' data
Traceback (most recent call last):
  File "./bug2.py", line 47, in <module>
    r = p.solve(solver, plot=1)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/baseProblem.py", line 235, in solve
    return runProbSolver(self, *args, **kwargs)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/runProbSolver.py", line 219, in runProbSolver
    nErr = check(p)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/check.py", line 19, in check
    p.err('the solver ' + p.solver.__name__ + ' cannot handle ' + "'" + fn + "' data")
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/oologfcn.py", line 16, in ooerr
    raise OpenOptException(msg)
oologfcn.OpenOptException: the solver galileo cannot handle 'A' data

With interalg,

FuncDesigner warning: Probably scipy installation could speed up running the code involved

------------------------- OpenOpt 0.34 -------------------------
solver: interalg_0.21   problem: unnamed    type: GLP
 iter   objFunVal   log10(maxResidual)   
    0  3.362e+01            -100.00 
OpenOpt Error: 
            solver interalg_0.21 requires finite lb, ub: 
            lb <= x <= ub 
            (you can use "implicitBoounds")
            
Traceback (most recent call last):
  File "./bug2.py", line 48, in <module>
    r = p.solve(solver, plot=1)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/baseProblem.py", line 235, in solve
    return runProbSolver(self, *args, **kwargs)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/runProbSolver.py", line 238, in runProbSolver
    solver(p)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/solvers/UkrOpt/interalg_oo.py", line 46, in __solver__
    ''' % self.__name__)
  File "/usr/local/lib/python2.6/dist-packages/openopt-0.34-py2.6.egg/openopt/kernel/oologfcn.py", line 16, in ooerr
    raise OpenOptException(msg)
oologfcn.OpenOptException: 
            solver interalg_0.21 requires finite lb, ub: 
            lb <= x <= ub 
            (you can use "implicitBoounds")

I really do not understand what is going on in both cases. Name (even automatically created) for which data, variable is a problem would be helpful. In first case, I given all constrains, being finite. In second case, I have no idea what solver is asking me about 'A' data.

I'm fighting with this for more than a week, without success. Was trying to use oovar or oovars, in multiple combinations, but then other problems arises. Help will be appreciated. Thanks.

I was experimenting with OpenOpt, and discovered multiple problems.

PROBLEM 1.

K = oovars(1)
Z = (2*K)**2

I know oovars(1) is an array, but it should work, however Z expression (second line) loops infinitely.

PROBLEM 2.

Another, problem, is that in constrains for example to interalg, I cannot use  -1 < x < -1, but need to use two separate conditions -1 < x && x < 1. It would be nice to support such syntax (it is already in Python available). Similarly it should be supported for the arrays -1 < array < 1, which will mean constrain for each coordinate.

PROBLEM 3.

In galileo, there is bug

galileo_oo.py line 75:
   for itn in range(p.maxIter+1):

it should use xrange instead, because for very big maxIter, it fails with MemoryError

PROBLEM 4.

In de and interalg, if I have inequality of form lower <= x, x < upper (one strict and one non-strict), it ends with

Error in de_oo.py line 196:
    bool_v = bool_constr_v + bool_v
ValueError: shape mismatch: objects cannot be broadcast to single shape

It would be nice to support them however.

PROBLEM 5.

in de, I had this

de_oo.py line 102
  best, vals, constr_vals = _eval_pop(pop, p)
de_oo.py line 239
  best = (best_i, vals[best_i], 0, pop[best_i])
IndexError: index out of bounds

PROBLEM 6.

in interalg

interalg_oo.py line 178
   if tmp < frc:
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

PROBLEM 7.

Lack of function sum_squared() for arrays. I can use dot(T,T), however is is suboptimal when doing interval arithmetic. I currently use just

f = reduce(lambda sum, e: sum + e**2, Residuals)

PROBLEM 8:

Theta = oovars(10)
K = oovar()
R = 1.0e-3*K*Theta

ooFun.py line 387, in __mul__
  return other.__mul__(self)
ooFun line 1495, in __mul__
  if not hasSize: raise Func...
FuncDesigner.FDmisc.FuncDesignerException: to perform the operation oofun size should be known

COMMENTS:

This errors appears in various conditions. I want to globally solve simple quadratic problem, with one scalar variable (K), and one array variable (Theta), of size about 10-200.  Problem 6 arise if I wan use constrains of form

0 < Theta

, instead of adding all constrains manually for all i components

0 < Theta[i]

.

I had workarounds for most of the problems here, however most of them are bugs, and should be fixed.

Thanks.

The method should not store all the data, as there will eventually be too much.

Are there such methods, and if so, any example or documentation on usage?

The bubble detection algorithm looks for log-periodic super-exponential patterns in the price history of a security. Such a pattern can have a finite time discontinuity, which often indicates a crash. See [http://www.er.ethz.ch/fco/index] for references and more information.

The formal problem is to fit the function f(t) = A+B*(Tc-t)**M+C*(Tc-t)**M*cos(W*log(Tc-t)+Phi)) to a time series of prices, with constraints  Tc < t and 0 < M < 1. I do this by generating an error function sum((f(ti)-pi)**2) and minimizing it. I've lately switched to using OpenOpt ralg.

Now for my question. For fixed Tc,M,W, and Phi, the problem of finding A,B,C has a unique solution that can be computed directly. In my code, I "slave" the variable A,B,C to the other variables. Specifically, at each error function evaluation, I use the scipy least squares solver to compute the best current A,B,C values. This speeds up the calculation by roughly 2-3x. I would like to do this using FuncDesigner, but I can't figure out how. What I need is the ability to acquire the current values of Tc,M,W,Phi each time the solver evaluates the error function, so I can run the linear solver for A,B,C. This is fairly easy with ordinary Python functions, but then I don't have the automatic differentiation capability.

The following code snippet illustrates what I'm trying to do:

    def solve_ABC(M,Tc,Phi,W):
        # use least squares to find A,B,C
        am = np.column_stack([[f_A(t,M,Tc) for t in rtimes],
                              [f_B(t,M,Tc) for t in rtimes],
                              [f_C(t,M,Tc,Phi,W) for t in rtimes]])
        rp = []
        for r in range(len(rtimes)):
            wsqt = m.sqrt(rweights[r])
            for c in range(3):
                am[r,c] *= wsqt
            rp.append(rprices[r]*wsqt)
        (A,B,C),resid,rank,sigma = linalg.lstsq(am,rp)
        return A,B,C

    def residual_wosc(M,Tc,Phi,W):
        A,B,C = solve_ABC(M,Tc,Phi,W)
        rvals = [f_wosc(t,A,B,C,M,Tc,Phi,W) for t in rtimes]
        return sum([(v-p)*(v-p)*w for v,p,w in zip(rvals,rprices,rweights)])

def f (x):
    return np.linalg.norm (x**2 - 2)

x0 = np.zeros(10)
p = NLP (f, x0)
r = p.solve ('ralg')
-----------

This didn't work at all.
r.xf
Out[7]: array([ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.])

2) Here is my numpy version
>>> print numpy.version.version
2.0.0.dev-6441c2a

print scipy.version.version
0.10.0.dev
]
2) using easy_install I get the below

Below is my error

sudo python setup.py install
/Library/Python/2.6/site-packages/scipy-0.10.0.dev_20110527-py2.6-macosx-10.6-universal.egg/scipy/__init__.py:97: UserWarning: Numpy 1.5.0 or above is recommended for this version of scipy (detected version 1.2.1)
  UserWarning)
Segmentation fault

I am usually using pip to install python packages.
pip install openopt fails for me:

$ pip search openopt
openopt                   - A python module for numerical optimization
$ sudo pip install openopt
Password:
Downloading/unpacking openopt
  Could not find any downloads that satisfy the requirement openopt
No distributions at all found for openopt
Storing complete log in /Users/deil/.pip/pip.log

Is this a problem of my setup or is pip generally not supported by openopt?

Thanks!
Christoph

When running the MILP example referenced in http://openopt.org/MILP with the GLPK solver, the solver quits without giving any result. Any statements after the p.solve('glpk') are ignored, but I get no error message. All the output I can see is the following:

--------------------------------------------------
solver: glpk   problem: unnamed    type: MILP   goal: min
iter    objFunVal    log10(maxResidual)   
    0  0.000e+00               2.36

Does anyone have an idea of what's wrong? Any hints will be much appreciated.

Best,

Roger

I am a PhD student at Indian Institute of Technology. I am formulating an optimization problem which is most likely will be a MINLP one. It is related to one issue in optimal operation of electric power distribution system. I hope that openopt will be useful for me. I have try some examples and like them.

Thanks a lot for accepting me as a member of this forum.

Sincerely Yours,

Novalio Daratha
Research Scholar at Power System Engineering Group
Electrical Engineering Dept.
Indian Institute of Technology Roorkee
Roorkee, Uttaranchal, India

    Seems like you are using FuncDesigner from Enthought Python Distribution;
    consider using free GPL-licensed alternatives
    PythonXY (http://www.pythonxy.com) or
    Sage (http://sagemath.org) instead.
   

    Seems like you are using OpenOpt from Enthought Python Distribution;
    consider using free GPL-licensed alternatives
    PythonXY (http://www.pythonxy.com) or
    Sage (http://sagemath.org) instead.
   
--------------------------------------------------
solver: scipy_fsolve   problem: unnamed    type: NLSP
iter    objFunVal   
    0  8.650e+04
Traceback (most recent call last):
  File "systemSolve.py", line 26, in <module>
    r = S.solve(startPoint, fixedVars=fixedVars)
  File "/home/nwagner/local/lib/python2.6/site-packages/FuncDesigner-0.33-py2.6.egg/FuncDesigner/ooSystem.py", line 96, in solve
    return self._solve(*args, **kwargs)
  File "/home/nwagner/local/lib/python2.6/site-packages/FuncDesigner-0.33-py2.6.egg/FuncDesigner/ooSystem.py", line 178, in _solve
    return p.solve() if kwargs.get('manage', False) in (False, 0) else p.manage()
  File "/home/nwagner/local/lib/python2.6/site-packages/openopt-0.33-py2.6.egg/openopt/kernel/baseProblem.py", line 229, in solve
    return runProbSolver(self, *args, **kwargs)
  File "/home/nwagner/local/lib/python2.6/site-packages/openopt-0.33-py2.6.egg/openopt/kernel/runProbSolver.py", line 220, in runProbSolver
    solver(p)
  File "/home/nwagner/local/lib/python2.6/site-packages/openopt-0.33-py2.6.egg/openopt/solvers/scipy_optim/scipy_fsolve_oo.py", line 17, in __solver__
    xf = fsolve(p.f, p.x0, fprime=p.df, xtol = p.xtol, maxfev = p.maxFunEvals, warning = (p.iprint>=0))
TypeError: fsolve() got an unexpected keyword argument 'warning'

what is the difference between scipy's fmin_bfgs and openopt's scipy_bfgs ?
fmin_bfgs doesn't work in my example.
Any pointer would be appreciated.

Nils

I guess it should be

Sparse = cvxopt.base.spmatrix instead of  Sparse = cvxopt.spmatrix

in line 6 of cvxopt_misc.py

Nils