import numpy as np
import gurobipy as gp
from gurobipy import GRB

def f(w):
    y = np.exp(w)
    return y
    # return w+np.ones(len(w))

def error_function(target, predicted):
    error = target - predicted
    return error.mean()


def print_solutions(model):
    for v in model.getVars():
        print('%s %g' % (v.varName, v.x))
    print('Obj: %g' % model.getObjective().getValue())

targets = np.array([3, 2, 1])

model = gp.Model("qp")
model.setParam("PSDTol", float('Inf'))

weight1 = model.addVar(lb=-100, ub=100, name="w_1", vtype=GRB.CONTINUOUS)
weight2 = model.addVar(lb=-100, ub=100, name="w_2", vtype=GRB.CONTINUOUS)
weight3 = model.addVar(lb=-100, ub=100, name="w_3", vtype=GRB.CONTINUOUS)
var1 = model.addVar(lb=-100, ub=100, name="var_1", vtype=GRB.CONTINUOUS)
var2 = model.addVar(lb=-100, ub=100, name="var_2", vtype=GRB.CONTINUOUS)
var3 = model.addVar(lb=-100, ub=100, name="var_3", vtype=GRB.CONTINUOUS)

var = np.array([var1, var2, var3])

obj = error_function(targets, var)
model.setObjective(obj, GRB.MINIMIZE)

model.addConstr(var1 == f(weight1), "c1")
model.addConstr(var2 == f(weight2), "c2")
model.addConstr(var3 == f(weight3), "c3")

model.optimize()
print_solutions(model)