I need to constrain a variable that is argument of optimization below a certain value EQ<1000. #function that produces the variable is the following:
def fEQ(iEQ, iC, iTATM, index):
if index==0:
return iC[0]
else:
return iEQ[0]/psi*iTATM[index]**2
t = np.arange(1, 101)
NT = len(t)
EQ=np.zeros(NT)
#The objective function (omitted some functions to make some clarity - they are all interrelated in reality)
def fOBJ(x,sign,iI, iCPRICE,iEQ,iPERIODU,iCEMUTOTPER,iRI,iNT):
iMIU = x[0:NT]
iS = x[NT:(2*NT)]
for i in range(iNT):
iCPRICE[i] = fCPRICE(iMIU,i)
iI[i] = fI(iS,iY,i)
iEQ[i]=fEQ(iEQ,iC,iTATM,i)
iPERIODU[i] = fPERIODU(iC,iEQ,il,i)
iCEMUTOTPER[i] = fCEMUTOTPER(iPERIODU,il,i)
iRI = fRI(iCPC,i)
resUtility = np.zeros(1)
fUTILITY(iCEMUTOTPER, resUtility)
return sign*resUtility[0]
#Optimization
result = opt.minimize(fOBJ, x_start, args=(-1.0,CPRICE,I,EQ,PERIODU,CEMUTOTPER,RI,NT), method='SLSQP',bounds = tuple(bnds),options={'disp': True, 'maxiter':1000 }, constraints={'type':'ineq','fun':lambda EQ: EQ[NT]-1000})
Output of the optimization is "Positive directional derivative for linesearch (Exit mode 8)", it cannot find an optimum. Hovewer I have the suspect that my constraint is not fully working. Any suggestions?