Making parameterized EGM objectives DPP

[BlockWaving]

Hi, Thanks for sharing this awesome package!

I am trying to optimize the gamma of the EGM problem (Please see link:) as follow:

# set gamma as cp.Parameter
gamma = cp.Parameter(1, nonneg=True)
gamma_tch = torch.log(torch.ones(1))
k = len(mus)
def Objs(w):
    u = cvx.vstack([cvx.log(pi[i])
      - gamma * mus[i] @ w
      + (gamma**2/2) * cvx.quad_form(w, Sigmas[i]) for i in range(k)])
    return cvx.log_sum_exp(u)

w = cvx.Variable(n)
objective = cvx.Minimize(Objs(w))
constraints = [ w >= 0, cvx.sum(w) == 1 ]
egm_prob = cvx.Problem(objective, constraints)

cvxpylayer = CvxpyLayer(egm_prob, [torch.exp(gamma_tch)], [w]),

But the egm_prob can not pass the DPP test:
ValueError: Problem must be DPP.

Please is there any ways to work around this issue?

Thanks!