dbc.stitch(and_csp)

[rog77]

On colab, I copy over the file dwave-examples/factoring-notebook/blob/master/01-factoring-overview.ipynb from your github, though the open menu.

I insert a code cell at the top with:

!pip install dwave-ocean-sdk
!pip install dwavebinarycsp[maxgap]

(the second line is suggested by an error if the notebook is run without it)

I restart the runtime from the cell on colab, as per instructions from pip.

When I run the rest of the notebook, it hangs on:
and_bqm = dbc.stitch(and_csp)

With the following messages:

VisibleDeprecationWarning Traceback (most recent call last)
in ()
----> 1 and_bqm = dbc.stitch(and_csp)
2 and_bqm.remove_offset()
3
4 print("Linear coefficients:\n\t{}".format(
5 {key: round(val, 2) for key, val in and_bqm.linear.items()}))

10 frames
/usr/local/lib/python3.7/dist-packages/dwavebinarycsp/compilers/stitcher.py in stitch(csp, min_classical_gap, max_graph_size)
180 # try to use the penaltymodel ecosystem
181 try:
--> 182 pmodel = pm.get_penalty_model(spec)
183 except pm.ImpossiblePenaltyModel:
184 # hopefully adding more variables will make it possible

/usr/local/lib/python3.7/dist-packages/penaltymodel/core/interface.py in get_penalty_model(specification)
71 for factory in iter_factories():
72 try:
---> 73 pm = factory(specification)
74 except ImpossiblePenaltyModel as e:
75 # information about impossible models should be propagated

/usr/local/lib/python3.7/dist-packages/penaltymodel/lp/interface.py in get_penalty_model(specification)
57 linear_energy_ranges=specification.ising_linear_ranges,
58 quadratic_energy_ranges=quadratic_ranges,
---> 59 min_classical_gap=specification.min_classical_gap)
60 except ValueError:
61 raise pm.exceptions.FactoryException("Specification is for too large of a model")

/usr/local/lib/python3.7/dist-packages/penaltymodel/lp/generation.py in generate_bqm(graph, table, decision_variables, linear_energy_ranges, quadratic_energy_ranges, min_classical_gap, catch_warnings)
176 try:
177 result = linprog(cost_weights.flatten(), A_eq=noted_matrix, b_eq=noted_bound,
--> 178 A_ub=unnoted_matrix, b_ub=unnoted_bound, bounds=bounds)
179 except (OptimizeWarning, LinAlgWarning) as e:
180 raise ValueError('Penaltymodel-lp has a bad matrix')

/usr/local/lib/python3.7/dist-packages/scipy/optimize/_linprog.py in linprog(c, A_ub, b_ub, A_eq, b_eq, bounds, method, callback, options, x0)
552 x, status, message, iteration = _linprog_ip(
553 c, c0=c0, A=A, b=b, callback=callback,
--> 554 postsolve_args=postsolve_args, **solver_options)
555 elif meth == 'revised simplex':
556 x, status, message, iteration = _linprog_rs(

/usr/local/lib/python3.7/dist-packages/scipy/optimize/_linprog_ip.py in _linprog_ip(c, c0, A, b, callback, postsolve_args, maxiter, tol, disp, alpha0, beta, sparse, lstsq, sym_pos, cholesky, pc, ip, permc_spec, **unknown_options)
1123 lstsq, sym_pos, cholesky,
1124 pc, ip, permc_spec, callback,
-> 1125 postsolve_args)
1126
1127 return x, status, message, iteration

/usr/local/lib/python3.7/dist-packages/scipy/optimize/_linprog_ip.py in _ip_hsd(A, b, c, c0, alpha0, beta, maxiter, disp, tol, sparse, lstsq, sym_pos, cholesky, pc, ip, permc_spec, callback, postsolve_args)
753 d_x, d_y, d_z, d_tau, d_kappa = _get_delta(
754 A, b, c, x, y, z, tau, kappa, gamma, eta,
--> 755 sparse, lstsq, sym_pos, cholesky, pc, ip, permc_spec)
756
757 if ip: # initial point

/usr/local/lib/python3.7/dist-packages/scipy/optimize/_linprog_ip.py in _get_delta(A, b, c, x, y, z, tau, kappa, gamma, eta, sparse, lstsq, sym_pos, cholesky, pc, ip, permc_spec)
319
320 # [4] 8.12 and "Let alpha be the maximal possible step..." before 8.23
--> 321 alpha = _get_step(x, d_x, z, d_z, tau, d_tau, kappa, d_kappa, 1)
322 if ip: # initial point - see [4] 4.4
323 gamma = 10

/usr/local/lib/python3.7/dist-packages/scipy/optimize/_linprog_ip.py in _get_step(x, d_x, z, d_z, tau, d_tau, kappa, d_kappa, alpha0)
372 alpha_z = alpha0 * np.min(z[i_z] / -d_z[i_z]) if np.any(i_z) else 1
373 alpha_kappa = alpha0 * kappa / -d_kappa if d_kappa < 0 else 1
--> 374 alpha = np.min([1, alpha_x, alpha_tau, alpha_z, alpha_kappa])
375 return alpha
376

<array_function internals> in amin(*args, **kwargs)

/usr/local/lib/python3.7/dist-packages/numpy/core/fromnumeric.py in amin(a, axis, out, keepdims, initial, where)
2829 """
2830 return _wrapreduction(a, np.minimum, 'min', axis, None, out,
-> 2831 keepdims=keepdims, initial=initial, where=where)
2832
2833

/usr/local/lib/python3.7/dist-packages/numpy/core/fromnumeric.py in _wrapreduction(obj, ufunc, method, axis, dtype, out, **kwargs)
85 return reduction(axis=axis, out=out, **passkwargs)
86
---> 87 return ufunc.reduce(obj, axis, dtype, out, **passkwargs)
88
89

VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray

** I get identical errors using dwavebinarycsp.stitch on my own PC as well.

Is this a code issue, or human error on my part?

Thanks!

[JoelPasvolsky]

Hi @rog77,

Can you try replacing your !pip install dwavebinarycsp[maxgap] line with

!pip install scipy==1.5.4
!pip install penaltymodel-maxgap==0.5.4

Also note that you might need to also install bokeh, as can be seen in the https://github.com/dwave-examples/factoring-notebook/blob/master/requirements.txt file.

[rog77]

That sorted it, thanks!