About cooling schedule about openjij

[K-Suzuki-Jij]

The cooling schedule does not seem to be consistent with changing vartype

import openjij as oj
model = oj.BinaryPolynomialModel({(0,1):-1}, "SPIN")
res = oj.SASampler().sample_hubo(model.to_hising(), "SPIN")
res.info['schedule']
{'beta_max': 2.302585092994046, 'beta_min': 0.34657359027997264, 'num_sweeps': 1000}

res = oj.SASampler().sample_hubo(model.to_hubo(), "BINARY")
res.info['schedule']
{'beta_max': 1.5350567286626973, 'beta_min': 0.11552453009332421, 'num_sweeps': 1000}

Above two cases should have the same cooling schedule.
This inconsistency also occurs in sample_ising and sample_qubo

[j-i-k-o]

Detailed log shows that the right-bottom element of augmented ising matrix is 1, which makes the min_delta_energy always 1.

>>> import openjij as oj
>>> res = oj.SASampler().sample_ising({}, {(0,1):-1, (0,2):+1})
ising interaction: [[ 0. -1.  1.  0.]
 [ 0.  0.  0.  0.]
 [ 0.  0.  0.  0.]
 [ 0.  0.  0.  1.]]
ising_interaction[ising_interaction > 0]: [1. 1.]
abs_bias: [2. 0. 0. 1.]
abs_bias[abs_bias > 0]: [2. 1.]
>>> res.info
{'system': [], 'sampling_time': 344.903999575763, 'execution_time': 291.5839995694114, 'list_exec_times': array([291.58399957]), 'schedule': {'beta_max': 4.605170185988092, 'beta_min': 0.34657359027997264, 'num_sweeps': 1000}}
>>> res = oj.SASampler().sample_ising({}, {(0,1):-2, (0,2):+2})
ising interaction: [[ 0. -2.  2.  0.]
 [ 0.  0.  0.  0.]
 [ 0.  0.  0.  0.]
 [ 0.  0.  0.  1.]]
ising_interaction[ising_interaction > 0]: [2. 1.]
abs_bias: [4. 0. 0. 1.]
abs_bias[abs_bias > 0]: [4. 1.]

[K-Suzuki-Jij]

The cooling schedule does not seem to be consistent with changing vartype

import openjij as oj
model = oj.BinaryPolynomialModel({(0,1):-1}, "SPIN")
res = oj.SASampler().sample_hubo(model.to_hising(), "SPIN")
res.info['schedule']
{'beta_max': 2.302585092994046, 'beta_min': 0.34657359027997264, 'num_sweeps': 1000}

res = oj.SASampler().sample_hubo(model.to_hubo(), "BINARY")
res.info['schedule']
{'beta_max': 1.5350567286626973, 'beta_min': 0.11552453009332421, 'num_sweeps': 1000}

Above two cases should have the same cooling schedule. This inconsistency also occurs in sample_ising and sample_qubo

Hard to fix this right now, so we'll do it in the future.

[K-Suzuki-Jij]

The first problem has been fixed in v0.4.5.
I created a new issue about the second problem.