I am using scipy.optimize.minimize() from scipy to implement LBFGS method. But I am having the following problem:

Line search cannot locate an adequate point after MAXLS
  function and gradient evaluations.
  Previous x, f and g restored.
 Possible causes: 1 error in function or gradient evaluation;
                  2 rounding error dominate computation.

My code is the next one:

import scipy
import scipy.optimize 
import numpy as np
import matplotlib.pyplot as plt
import timeit
from funcion_objetivo import funcion_objetivo
from funcion_gradiente import gradiente

Nx = 210
Nz = 68  
tEnd = 2.5 
dt = 0.004
t = np.arange(0,tEnd,dt)
frec = 3
Nt = tEnd/dt 
Nt= int(Nt)
dz = 25
dx = 25
Sx1 = 25
Sx2 = 185
Sx3 = 105
Sx4 = 65
Sx5 = 145
Sz = 3
offset_max = 4000 
a = (np.pi*frec)**2
t0 = 1
g1 = (-2*a*(t-t0)*np.exp(-a*(t-t0)**2)).T
g1 = np.reshape(g1,(np.size(g1),1))

######### Modelo original
vp_ori = np.load('modelo_original.npy')
######### Modelo inicial
vp_ite = np.load('modelo_inicial1.npy')


start=timeit.default_timer()
p = scipy.optimize.minimize(funcion_objetivo, vp_ite.flatten(), 
                            method='L-BFGS-B', jac=True, args=(vp_ori,),
                            options={'disp':True, 'maxiter':5})
stop=timeit.default_timer()
print('time:', stop-start) 

vp_ori and vp_ite are matrix of (210, 68).
FWI_Grad and propagator are predefined functions. The first one returns a vector and the second one a tuple of len = 3 each contains a matrix, for example Pt_obs1[0].shape = (625, 165).

This is my objective function:

from propagator import propagator
import numpy as np
from funcion_gradiente import gradiente
from numba import jit

@jit
def funcion_objetivo(vp_ite, vp_ori):
    
    Sx1 = 25
    Sx2=185
    Sx3=105
    Sx4=65
    Sx5=145
    Sz=3
    dx=25
    dz=25
    dt=0.004
    frec=3
    a = (np.pi*frec)**2
    t0 = 1
    tEnd = 2.5 
    dt = 0.004
    t = np.arange(0,tEnd,dt)
    g1 = (-2*a*(t-t0)*np.exp(-a*(t-t0)**2)).T
    g1 = np.reshape(g1,(np.size(g1),1))
    
    
    Pt_obs1 = propagator(vp_ori, g1, Sx1, Sz, dx, dz, dt, 4000, frec)
    print('Punto observado 1 done')
    Pt_obs2 = propagator(vp_ori, g1, Sx2, Sz, dx, dz, dt, 4000, frec)
    print('Punto observado 2 done')
    Pt_obs3 = propagator(vp_ori, g1, Sx3, Sz, dx, dz, dt, 2125, frec)
    print('Punto observado 3 done')
    Pt_obs4 = propagator(vp_ori, g1, Sx4, Sz, dx, dz, dt, 3125, frec)
    print('Punto observado 4 done')
    Pt_obs5 = propagator(vp_ori, g1, Sx5, Sz, dx, dz, dt, 3125, frec)
    print('Punto observado 5 done')

    Pt_mod1 = propagator(vp_ite, g1, Sx1, Sz, dx, dz, dt, 4000, frec)
    print('Punto modelado 1 done')
    Pt_mod2 = propagator(vp_ite, g1, Sx2, Sz, dx, dz, dt, 4000, frec)
    print('Punto modelado 2 done')
    Pt_mod3 = propagator(vp_ite, g1, Sx3, Sz, dx, dz, dt, 2125, frec)
    print('Punto modelado 3 done')
    Pt_mod4 = propagator(vp_ite, g1, Sx4, Sz, dx, dz, dt, 3125, frec)
    print('Punto modelado 4 done')
    Pt_mod5 = propagator(vp_ite, g1, Sx5, Sz, dx, dz, dt, 3125, frec)
    print('Punto modelado 5 done')
        
    r1 = np.sum(Pt_mod1[0] - Pt_obs1[0])
    r2 = np.sum(Pt_mod2[0] - Pt_obs2[0])
    r3 = np.sum(Pt_mod3[0] - Pt_obs3[0])
    r4 = np.sum(Pt_mod4[0] - Pt_obs4[0])
    r5 = np.sum(Pt_mod5[0] - Pt_obs5[0])
    
    r = (r1**2 + r2**2 + r3**2 + r4**2 + r5**2)
    norma = 0.5*np.sqrt(r)
    print('la norma es: ', norma)
    
    grad = gradiente(vp_ite, vp_ori)
    print('el gradiente es: ', grad)
    
    return norma, grad

This is my gradient function:

from FWI_GRAD import FWI_GRAD
import numpy as np
from numba import jit
from propagator import propagator

@jit
def gradiente(vp_ite, vp_ori):
    Nx = 210
    Nz = 68  
    dz = 25
    Sx1 = 25
    Sx2=185
    Sx3=105
    Sx4=65
    Sx5=145
    Sz=3
    dx=25
    dz=25
    dt=0.004
    frec=3
    a = (np.pi*frec)**2
    t0 = 1
    tEnd = 2.5 
    dt = 0.004
    t = np.arange(0,tEnd,dt)
    g1 = (-2*a*(t-t0)*np.exp(-a*(t-t0)**2)).T
    g1 = np.reshape(g1,(np.size(g1),1))
    dt = 0.004
    frec = 3
    beta = 1
    Nt = tEnd/dt 
    Nt= int(Nt)
    
    Pt_obs1 = propagator(vp_ori, g1, Sx1, Sz, dx, dz, dt, 4000, frec)
    print('Punto observado 1 done')
    Pt_obs2 = propagator(vp_ori, g1, Sx2, Sz, dx, dz, dt, 4000, frec)
    print('Punto observado 2 done')
    Pt_obs3 = propagator(vp_ori, g1, Sx3, Sz, dx, dz, dt, 2125, frec)
    print('Punto observado 3 done')
    Pt_obs4 = propagator(vp_ori, g1, Sx4, Sz, dx, dz, dt, 3125, frec)
    print('Punto observado 4 done')
    Pt_obs5 = propagator(vp_ori, g1, Sx5, Sz, dx, dz, dt, 3125, frec)
    print('Punto observado 5 done')
        
    f1,grad1 = FWI_GRAD(vp_ite, Nx, Nz, Nt, g1, Sx1, Sz, dx, dz, dt, 4000, frec, Pt_obs1[0], 1) 
    f2,grad2 = FWI_GRAD(vp_ite, Nx, Nz, Nt, g1, Sx2, Sz, dx, dz, dt, 4000, frec, Pt_obs2[0], 2) 
    f3,grad3 = FWI_GRAD(vp_ite, Nx, Nz, Nt, g1, Sx3, Sz, dx, dz, dt, 2125, frec, Pt_obs3[0], 3)
    f4,grad4 = FWI_GRAD(vp_ite, Nx, Nz, Nt, g1, Sx4, Sz, dx, dz, dt, 3125, frec, Pt_obs4[0], 4) 
    f5,grad5 = FWI_GRAD(vp_ite, Nx, Nz, Nt, g1, Sx5, Sz, dx, dz, dt, 3125, frec, Pt_obs5[0], 5) 

    grad = beta*grad1 + beta*grad2 + beta*grad3 + beta*grad4 + beta*grad5
    
    return grad

What are the possible reasons that I should investigate? Do you know what is the error? I would appreciate any help.

I tried to return from objective function just ‘norma’ and set jac=gradiente but it did not work.