I am fitting my data to least sq fitting. I give constraints for the fitting though as shown below. 'Experiment_norm' is not defined but it is containing my data and calling my data to the code wasn't an issue so I did not put it in my question here. But one thing that I meant to ask in my constraints were to say that p[3], p[4], p[5], p[6], and p[7] to be less than 1. But when I call for the optimized parameters for p[3], p[4], p[5], p[6], and p[7], they are greater than 1. Am I doing something wrong?
def myerr(p, xdata, ydata):
return sum((ydata - calculate_spectrum(p,xdata))**2)
con = ({'type': 'eq', 'fun': lambda p: p[11]+p[12]+p[13]-1}, {'type': 'ineq', 'fun': lambda p: p[11]}, {'type': 'ineq', 'fun': lambda p: p[12]},{'type': 'ineq', 'fun': lambda p: p[13]},{'type': 'ineq', 'fun': lambda p: p[3]-1},{'type': 'ineq', 'fun': lambda p: p[4]-1},{'type': 'ineq', 'fun': lambda p: p[5]-1},{'type': 'ineq', 'fun': lambda p: p[6]-1},{'type': 'ineq', 'fun': lambda p: p[7]-1})
p0 = [m,z,3000,0.3,0.3,0.1,0.2,0.3,0.4,0.4,0.4,0.2,0.5,0.6]
p_opt = scipy.optimize.minimize(myerr, p0, args=(mz,experiment_norm), constraints = con)
From the documentation:
Equality constraint means that the constraint function result is to be zero whereas inequality means that it is to be non-negative.
Let's look at one of your constraints:
{'type': 'ineq', 'fun': lambda p: p[3]-1}
Non-negative means that p[3] - 1 must be greater or equal to zero. So p[3] - 1 >= 0 or p[3] >= 1. You did not specify the contstraint to be less than one but to be greater than one! What you want instead is this:
{'type': 'ineq', 'fun': lambda p: 1-p[3]}
Note that you can also group multiple contstraints together for simpler code:
{'type': 'ineq', 'fun': lambda p: [1-p[3], 1-p[4], 1-p[5], 1-p[6], 1-p[7]]}
or if p is a numpy array:
{'type': 'ineq', 'fun': lambda p: 1-p[3:8]}