问题描述
我试图通过定义一个 NxN 对称矩阵和一个 N 维偏置向量来解决带有 qobj 的 Gekko 的二次二元优化问题。然而，我观察到计算时间低得令人怀疑：2e-2s 用于解决 N=20 和 0.05s 解决 200 维问题。此外，我的迭代次数从未超过 3-4 次。我在这里遗漏了什么吗？
import numpy as np
N = 20
#create square symmetric matrix for the quadratic term
b = np.random.normal(0,1,(N,N))
Q = (b + b.T)/2
#bias vector for linear term
c=np.random.normal(0,N)
from gekko import GEKKO
m = GEKKO(remote=False)
z = m.Array(m.Var,N,integer=True,lb=0,ub=1,value=1)
m.qobj(c,A=Q,x=z,otype='min')
m.solve(disp=True)
感谢任何建议！
解决方法
切换到 APPT 求解器以获取混合整数解。
m.options.SOLVER=1
这是完整的脚本。 MIQP 问题通常非常快。如果问题是具有混合整数元素 (MINLP) 的非线性 (NLP)，它会显着减慢。
import numpy as np
N = 200
#create square symmetric matrix for the quadratic term
b = np.random.normal(0,1,(N,N))
Q = (b + b.T)/2
#bias vector for linear term
c=np.random.normal(0,N)
from gekko import GEKKO
m = GEKKO(remote=False)
z = m.Array(m.Var,N,integer=True,lb=0,ub=1,value=1)
m.qobj(c,A=Q,x=z,otype='min')
m.options.SOLVER=1
m.solve(disp=True)
print(z)
[[0.0] [1.0] [1.0] [0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [0.0] [1.0] [1.0]
[1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [0.0] [1.0] [1.0] [1.0]
[1.0] [1.0] [1.0] [1.0] [0.0] [1.0] [0.0] [1.0] [1.0] [0.0] [0.0] [0.0]
[0.0] [1.0] [1.0] [0.0] [0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [0.0] [1.0]
[1.0] [1.0] [0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [0.0]
[0.0] [1.0] [0.0] [0.0] [0.0] [0.0] [1.0] [1.0] [0.0] [0.0] [0.0] [0.0]
[1.0] [1.0] [0.0] [0.0] [1.0] [1.0] [0.0] [1.0] [1.0] [0.0] [1.0] [1.0]
[0.0] [1.0] [0.0] [0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0]
[0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0]
[1.0] [1.0] [1.0] [0.0] [0.0] [0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [1.0]
[0.0] [0.0] [1.0] [1.0] [0.0] [1.0] [0.0] [1.0] [0.0] [1.0] [1.0] [1.0]
[0.0] [0.0] [0.0] [0.0] [1.0] [0.0] [0.0] [1.0] [1.0] [0.0] [0.0] [1.0]
[1.0] [1.0] [0.0] [1.0] [0.0] [0.0] [1.0] [1.0] [1.0] [1.0] [0.0] [1.0]
[0.0] [0.0] [0.0] [0.0] [0.0] [0.0] [1.0] [1.0] [1.0] [1.0] [1.0] [0.0]
[0.0] [0.0] [1.0] [1.0] [1.0] [1.0] [0.0] [1.0] [0.0] [0.0] [1.0] [1.0]
[0.0] [0.0] [0.0] [1.0] [1.0] [1.0] [0.0] [1.0] [0.0] [1.0] [0.0] [0.0]
[0.0] [1.0] [1.0] [1.0] [0.0] [1.0] [1.0] [0.0]]