simple LP does not work with picos

[cannon chii]

I am testing a linear programming functionality of PICOS.

    prob = pic.Problem()

    tau_pic = prob.add_variable("tau",1)

    print prob

    prob.solve(solver = "cvxopt")

The problem printed out is 

    optimization problem  (LP):

    1 variables, 0 affine constraints

    

    tau : (1, 1), continuous

    

    find vars

    such that

      []

    ---------------------

The solution should be assignment of tau. Yet it generates error:

    ---------------------------------------------------------------------------

    ValueError                                Traceback (most recent call last)

    <ipython-input-968-6bd799ff66cd> in <module>()

          7 print prob

          8 

    ----> 9 prob.solve(solver = "cvxopt")

    

    /home/me/anaconda2/lib/python2.7/site-packages/picos/problem.pyc in solve(self, **options)

       5002                         or self.options['solver'] == 'smcp'

       5003                         or self.options['solver'] == 'cvxopt'):

    -> 5004                     primals, duals, obj, sol = self._cvxopt_solve()

       5005 

       5006                 # For cplex

    

    /home/me/anaconda2/lib/python2.7/site-packages/picos/problem.pyc in _cvxopt_solve(self)

       5256                                          G, h, dims,

       5257                                          self.cvxoptVars['A'],

    -> 5258                                          self.cvxoptVars['b'])

       5259             probtype = 'ConeLP'

       5260 

    

    /home/me/anaconda2/lib/python2.7/site-packages/cvxopt/coneprog.pyc in conelp(c, G, h, dims, A, b, primalstart, dualstart, kktsolver, xnewcopy, xdot, xaxpy, xscal, ynewcopy, ydot, yaxpy, yscal, **kwargs)

        571     if kktsolver in defaultsolvers:

        572         if KKTREG is None and (b.size[0] > c.size[0] or b.size[0] + cdim_pckd < c.size[0]):

    --> 573            raise ValueError("Rank(A) < p or Rank([G; A]) < n")

        574         if kktsolver == 'ldl':

        575             factor = misc.kkt_ldl(G, dims, A, kktreg = KKTREG)

    

    ValueError: Rank(A) < p or Rank([G; A]) < n

Anyone know what might happen?  

[Martin]

CVXOPT's conelp() routine assumes that your problem has constraints (the problem is trivial if you don't).