CVXPY:目标不是DCP
CVXPY:目标不是DCP
提问于 2022-09-23 22:05:23
我试图解决标准支持向量机优化问题的对偶问题：
代码很简单：
X = cp.Parameter((n, 2))
X.value = np.einsum("j,jk->jk", df["y"].to_numpy(), df[["x1", "x2"]].to_numpy())
y = cp.Parameter(n)
y.value = df["y"].to_numpy()
w = cp.Variable(n)
ones = cp.Parameter(n)
ones.value = np.ones(n)
obj = cp.Maximize(-1/2 * w.T @ X @ X.T @ w + ones.T @ w)
constraints = [w >= 0,
w <= ones,
w.T @ y == 0]
prob = cp.Problem(obj, constraints)
prob.solve()但得到：
DCPError: Problem does not follow DCP rules. Specifically:
The objective is not DCP. Its following subexpressions are not:
Promote(-0.5, (200,)) @ var366 @ param364 @ param364.T @ var366回答 1
Stack Overflow用户
发布于 2022-09-26 06:36:24
cvxpy不能验证您的长乘法是DCP。相反，把它写成
obj = cp.Maximize(-1/2 * cp.sum_squares(X.T @ w) + ones.T @ w)页面原文内容由Stack Overflow提供。腾讯云小微IT领域专用引擎提供翻译支持
原文链接：
https://stackoverflow.com/questions/73833398
复制相关文章
相似问题
腾讯云开发者
