def __init__(self, m, k, formulation='penalizedl1', nonnegative=False):
if not cvx_available:
raise RuntimeError(
'Cannot initialize when cvxpy is not available.')
# Initialize parameters and variables
A = cvx.Parameter((m, k))
b = cvx.Parameter((m, 1))
l = cvx.Parameter(nonneg=True)
x = cvx.Variable((k, 1))
# Create the problem
if formulation == 'penalizedl1':
obj_func = 0.5 * cvx.sum_squares(cvx.matmul(A, x) -
b) + l * cvx.norm1(x)
constraints = []
elif formulation == 'constrainedl1':
obj_func = cvx.sum_squares(cvx.matmul(A, x) - b)
constraints = [cvx.norm1(x) <= l]
elif formulation == 'constrainedl2':
obj_func = cvx.norm1(x)
constraints = [cvx.norm(cvx.matmul(A, x) - b) <= l]
else:
raise ValueError("Unknown formulation '{0}'.".format(formulation))
if nonnegative:
constraints.append(x >= 0)
problem = cvx.Problem(cvx.Minimize(obj_func), constraints)
self._formulation = formulation
self._A = A
self._b = b
self._l = l
self._x = x
self._obj_func = obj_func
self._constraints = constraints
self._problem = problem
def findPiercingNumber(self):
# Finding piercing number using an integer linear program (ILP) formulation
# (a linear program (LP) that is constrained to have integer solutions)
# First, find constraint matrix/data for LP used to find piercing number
# Candidate piercing points are the endpoints of the intervals
_, orderedendpts = self.listSetEndpoints()
N = self.numVoters
M = 2 * N
Mat = np.empty([N, M])
for i in np.arange(0, N):
for j, p in enumerate(orderedendpts):
A = self.approvalSets[i]
Mat[i, j] = A.isPointInSet(p)
# Next, vector of coefficients of objective function
c = np.ones(M)
## Solve ILP using cvxpy
x = cp.Variable(M, integer=True)
objective = cp.Minimize(cp.matmul(c, x))
constraints = [cp.matmul(Mat, x) >= np.ones(N), 0 <= x, x <= 1]
prob = cp.Problem(objective, constraints)
val = prob.solve(solver='GLPK_MI')
piercingNumber = int(np.round(val))
piercingSet = np.transpose(orderedendpts)[np.round(x.value) > 0]
return piercingNumber, piercingSet, x.value, Mat
def estimate_parameters(y, mat1, mat2, gamma=0.1, beta=0.1):
th1 = cvx.Variable(mat1.shape[1])
th2 = cvx.Variable(mat2.shape[1])
if gamma is None:
gamma = 2 * np.sqrt(2 * np.log(len(y)))
cost = (cvx.sum_squares(y - cvx.matmul(mat1, th1) - cvx.matmul(mat2, th2))
+ gamma * cvx.norm1(th2) + beta * cvx.norm(th1[1:]))
problem = cvx.Problem(cvx.Minimize(cost))
problem.solve()
# th1.value[np.abs(th1.value) <= 1e-2 * th1.value[0]] = 0
# th2.value[np.abs(th2.value) <= 1e-2 * np.max(th2.value)] = 0
# sparsity_pattern_1 = np.abs(th1.value) <= 1e-2
sparsity_pattern_2 = np.abs(th2.value) <= 1e-2
cost = (cvx.sum_squares(y - cvx.matmul(mat1, th1) - cvx.matmul(mat2, th2))
+ beta * cvx.norm(th1[1:]))
constraints = []
# if np.sum(sparsity_pattern_1) > 0:
# constraints.append(th1[sparsity_pattern_1] == 0)
if np.sum(sparsity_pattern_2) > 0:
constraints.append(th2[sparsity_pattern_2] == 0)
problem = cvx.Problem(cvx.Minimize(cost), constraints)
problem.solve()
# th1.value[np.abs(th1.value) < 1e-4] = 0
th2.value[np.abs(th2.value) < 1e-4] = 0
return th1.value, th2.value
def LinProg(A, alpha, r):
x = cp.Variable(shape=(len(state_action_pairs), 1), name="x")
constraints = [cp.matmul(A, x) == alpha, x >= 0]
objective = cp.Maximize(cp.matmul(r, x))
problem = cp.Problem(objective, constraints)
return x, problem.solve()
def find_X_sol_agent(Y_s, Q, num_tasks, total_num_agents):
X_sol = cp.Variable((num_tasks, total_num_agents), boolean=True)
mismatch = Y_s - cp.matmul(X_sol, Q)
obj = cp.Minimize(cp.pnorm(mismatch, 2))
constraints = [cp.matmul(X_sol.T, np.ones([num_tasks, 1])) <= np.ones([total_num_agents , 1]),
cp.matmul(X_sol, np.ones([total_num_agents, 1])) == 3*np.ones([num_tasks, 1])]
opt_prob = cp.Problem(obj, constraints)
opt_prob.solve()
X_candidate = np.round(X_sol.value)
return X_candidate
def __init__(self, nb_measure, alpha=0.95):
self.h = cp.Variable(nb_measure, nonneg=True)
self.p = cp.Parameter(nb_measure, nonneg=True)
self.v = cp.Parameter(nb_measure)
objective = cp.Maximize(cp.matmul(self.v, cp.multiply(self.h, self.p)))
constraints = [
self.h <= 1 / (1 - alpha),
cp.matmul(self.h, self.p) == 1
]
self.problem = cp.Problem(objective, constraints)
def __init__(self, u):
self.u = u
self.v = np.array([1.0 / (np.log(2 + i)) for i, _ in enumerate(u)])
self.v[range(10, len(self.v))] = 0
self.P = cp.Variable((len(u), len(u)))
self.I = np.ones((len(u),))
self.constraints = [cp.matmul(self.I.transpose(), self.P) == self.I.transpose(),
cp.matmul(self.P, self.I) == self.I,
0 <= self.P, self.P <= 1
]
self.all_constraints = None
def __init__(self, nb_measure, q=2, c=1):
self.h = cp.Variable(nb_measure, nonneg=True)
self.p = cp.Parameter(nb_measure, nonneg=True)
self.v = cp.Parameter(nb_measure)
self.g = cp.multiply(self.p, 1 + self.h - cp.matmul(self.h, self.p))
objective = cp.Maximize(cp.matmul(self.v, self.g))
constraints = [
self.g >= 0,
cp.matmul(cp.power(self.h, q), self.p) <= c**q
]
self.problem = cp.Problem(objective, constraints)
def get_output(r, alpha):
r = np.array(r)
alpha = np.array(alpha)
x = cp.Variable(shape=(100, 1), name="x")
constraints = [cp.matmul(A, x) == alpha, x >= 0]
objective = cp.Maximize(cp.matmul(r, x))
problem = cp.Problem(objective, constraints)
solution = problem.solve()
for i in range(len(x.value)):
solutionx.append(x.value[i][0])
return solution
def __init__(self, nb_measure, alpha=0.95):
self.alpha = alpha
self.z = cp.Variable(nb_measure, nonneg=True)
self.p = cp.Parameter(nb_measure, nonneg=True)
self.v = cp.Parameter(nb_measure)
dkl = cp.matmul(self.p, -cp.entr(self.z))
objective = cp.Maximize(cp.matmul(self.p, cp.multiply(self.v, self.z)))
constraints = [
dkl <= np.log(1 / (1 - self.alpha)),
cp.matmul(self.p, self.z) == 1
]
self.problem = cp.Problem(objective, constraints)