def get_objective_tensor(var_ids, sym_data):
"""Get objective tensor via gradient of c'x."""
xs = [
tf.Variable(tf.zeros(sym_data.var_sizes[var_id], dtype=tf.float32))
for var_id in var_ids
]
xs_map = dict(zip((var_id for var_id in var_ids), xs))
obj_t = expressions.tensor(sym_data.objective, xs_map)
# get gradient, handling None values
return vector_ops.vstack([
vector_ops.vec(ci) if ci is not None else vector_ops.vec(
tf.zeros(sym_data.var_sizes[var_ids[i]], dtype=tf.float32))
for i, ci in enumerate(tf.gradients(obj_t, xs))
])
def A(self, x):
xs = {}
for var_id, var_size in self.sym_data.var_sizes.items():
var_offset = self.sym_data.var_offsets[var_id]
idx = slice(var_offset, var_offset + var_size[0] * var_size[1])
xs[var_id] = vector_ops.mat(x[idx, :], var_size)
return vector_ops.vstack([
vector_ops.vec(expressions.tensor(Ai, xs)) for Ai in self.A_exprs
])
def get_constraint_tensors(constraints):
"""Get expression for Ax + b."""
A_exprs = [constr.expr for constr in tree_mat.prune_constants(constraints)]
b = vector_ops.vstack([
vector_ops.vec(
tf.constant(-tree_mat.mul(constr.expr, {}), dtype=tf.float32))
for constr in constraints
])
return A_exprs, b
def AT(self, y):
ys = []
offset = 0
for constr in self.constraints:
idx = slice(offset, offset + constr.size[0] * constr.size[1])
ys.append(vector_ops.mat(y[idx, :], constr.size))
offset += constr.size[0] * constr.size[1]
x_map = sum_dicts(
expressions.adjoint_tensor(Ai, ys[i])
for i, Ai in enumerate(self.A_exprs))
return vector_ops.vstack(
[vector_ops.vec(x_map[var_id]) for var_id in self.var_ids])