def interpolate_pixtab(self, pixtab_df, update_params=False, *args, **kwargs):
print("Iniciando interpolador Amoeba-Borovicka...")
if update_params:
self.PARAMS = self.update_params(self.PARAMS, pixtab_df, *args, **kwargs)
self.altura = construct_altura_deg(*self.PARAMS)
self.azimut = construct_azimuth_deg(*self.PARAMS)
X0, Y0 = self.PARAMS[10:12]
print("Interpolando con cenith x: {}, y: {}...".format(X0, Y0))
df = self._interpolate_df()
print("Interpolacion finalizada.")
return df
def __init__(self, params = None, *args, **kwargs):
if params:
self.PARAMS = params
self.altura = construct_altura_deg(*self.PARAMS)
self.azimut = construct_azimuth_deg(*self.PARAMS)
def paso1(params, data, drop_outlayers=False):
"""
314 - V, S, D, P, Q, X0, Y0, a0
Primero haremos un acercamiento 314 para buscar
parametros C,A,F,E,ep que nos ayuden a reducir mejor los datos outlayers
"""
def construct_acumulate_alt_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
V, S, D, P, Q, X0, Y0, a0 = ctes
C, A, F, E, ep, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_alt_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
def construct_acumulate_az_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
V, S, D, P, Q, X0, Y0, a0 = ctes
C, A, F, E, ep, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_az_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
######################################################################################################
def amoeba(params, mega_alt_xi, mega_az_xi):
new_alt_params = optimize.fmin(mega_alt_xi, params)#, maxfun=100)
# new_az_params = optimize.fmin(mega_az_xi, params, maxfun=1100)
return new_alt_params
C, A, F, V, S, D, P, Q, E, ep, X0, Y0, a0 = params
init_params = [C, A, F, E, ep]
ctes = [V, S, D, P, Q, X0, Y0, a0]
# data = pd.read_csv("Data.pixtab", sep=" ")
data = data.dropna()
if drop_outlayers:
data = drop_outlayers_by_borovicka(data,
construct_altura_deg(*init_params[:3],
*ctes[:-3], *init_params[-2:], *ctes[-3:]), 5)
x_exp = data.x.values
y_exp = data.y.values
alt_catalogo = data.alt.values
az_catalogo = np.rad2deg(np.arccos(np.cos(np.deg2rad(data.az.values))))
mega_alt_xi = construct_acumulate_alt_xi(x_exp, y_exp, alt_catalogo, ctes)
mega_az_xi = construct_acumulate_az_xi(x_exp, y_exp, az_catalogo, ctes)
pbar = tqdm(total=1000, desc="C, A, F, E, ep")
result = amoeba(init_params, mega_alt_xi, mega_az_xi)
pbar.close()
result_1 = np.array([*result[:3], *ctes[:-3], *result[-2:], *ctes[-3:]])
print(result_1)
return result_1
def paso5(params, data, drop_outlayers=False):
"""
Ahora buscamos C,A,F reduciendo la altura.
Según lo que experimentamos no deberia afectar al azimut
"""
def construct_acumulate_alt_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
V, S, D, P, Q, E, ep, X0, Y0, a0 = ctes
C, A, F, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_alt_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
def construct_acumulate_az_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
V, S, D, P, Q, E, ep, X0, Y0, a0 = ctes
C, A, F, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_az_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
######################################################################################################
def amoeba(params, mega_alt_xi, mega_az_xi):
new_alt_params = optimize.fmin(mega_alt_xi, params)#, maxfun=100)
# new_az_params = optimize.fmin(mega_az_xi, params, maxfun=1100)
return new_alt_params
C, A, F, V, S, D, P, Q, E, ep, X0, Y0, a0 = params
init_params = [C, A, F]
ctes = [V, S, D, P, Q, E, ep, X0, Y0, a0]
# data = pd.read_csv("Data.pixtab", sep=" ")
data = data.dropna()
if drop_outlayers:
data = drop_outlayers_by_borovicka(data,
construct_altura_deg(*init_params, *ctes), 5)
x_exp = data.x.values
y_exp = data.y.values
alt_catalogo = data.alt.values
az_catalogo = np.rad2deg(np.arccos(np.cos(np.deg2rad(data.az.values))))
mega_alt_xi = construct_acumulate_alt_xi(x_exp, y_exp, alt_catalogo, ctes)
mega_az_xi = construct_acumulate_az_xi(x_exp, y_exp, az_catalogo, ctes)
pbar = tqdm(total=100, desc="C, A, F")
result = amoeba(init_params, mega_alt_xi, mega_az_xi)
pbar.close()
result_5 = np.array([*result, *ctes])
print(result_5)
return result_5
def paso3(params, data, drop_outlayers=False):
"""
Ahora buscamos E,X0,Y0,a0 quitando datos outlayers reduciendo azimut.
Como a0 esta más cerca del resultado pude que se demore menos.
(19min menos y mejor resultado)
"""
def construct_acumulate_alt_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
C, A, F, V, S, D, P, Q, ep = ctes
E, X0, Y0, a0, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_alt_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
def construct_acumulate_az_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
C, A, F, V, S, D, P, Q, ep = ctes
E, X0, Y0, a0, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_az_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
######################################################################################################
def amoeba(params, mega_alt_xi, mega_az_xi):
# new_alt_params = optimize.fmin(mega_alt_xi, params, maxfun=1100)
new_az_params = optimize.fmin(mega_az_xi, params)#, maxfun=100)
return new_az_params
C, A, F, V, S, D, P, Q, E, ep, X0, Y0, a0 = params
init_params = [E, X0, Y0, a0]
ctes = [C, A, F, V, S, D, P, Q, ep]
# data = pd.read_csv("Data.pixtab", sep=" ")
data = data.dropna()
if drop_outlayers:
data = drop_outlayers_by_borovicka(data,
construct_altura_deg(*ctes[:-1], init_params[0], ctes[-1],
*init_params[-3:]), 5)
x_exp = data.x.values
y_exp = data.y.values
alt_catalogo = data.alt.values
az_catalogo = np.rad2deg(np.arccos(np.cos(np.deg2rad(data.az.values))))
mega_alt_xi = construct_acumulate_alt_xi(x_exp, y_exp, alt_catalogo, ctes)
mega_az_xi = construct_acumulate_az_xi(x_exp, y_exp, az_catalogo, ctes)
pbar = tqdm(total=100, desc="E, X0, Y0, a0")
result = amoeba(init_params, mega_alt_xi, mega_az_xi)
pbar.close()
result_3 = np.array([*ctes[:-1], result[0], ctes[-1], *result[-3:]])
print(result_3)
return result_3
def paso2(params, data, drop_outlayers=False):
"""
Ahora buscamos a0 quitando datos outlayers reduciendo azimut.
Al ser solo a0 deberia ser mas fácil.
"""
def construct_acumulate_alt_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
C, A, F, V, S, D, P, Q, E, ep, X0, Y0 = ctes
a0, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_alt_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
def construct_acumulate_az_xi(x_list, y_list, z_list, ctes = None):
def acumulate_xi(params):
C, A, F, V, S, D, P, Q, E, ep, X0, Y0 = ctes
a0, = params
suma_xi = 0
for x, y, z in zip(x_list, y_list, z_list):
xi = construct_az_xi(int(x), int(y), int(z))
suma_xi += xi(C, A, F, V, S, D, P, Q, E, ep, float(X0), float(Y0), a0)
pbar.update()
return suma_xi
return acumulate_xi
######################################################################################################
def amoeba(params, mega_alt_xi, mega_az_xi):
# new_alt_params = optimize.fmin(mega_alt_xi, params, maxfun=1100)
new_az_params = optimize.fmin(mega_az_xi, params)#, maxfun=10)
return new_az_params
C, A, F, V, S, D, P, Q, E, ep, X0, Y0, a0 = params
init_params = [a0]
ctes = [C, A, F, V, S, D, P, Q, E, ep, X0, Y0]
# data = pd.read_csv("Data.pixtab", sep=" ")
data = data.dropna()
if drop_outlayers:
data = drop_outlayers_by_borovicka(data,
construct_altura_deg(*ctes, *init_params), 5)
x_exp = data.x.values
y_exp = data.y.values
alt_catalogo = data.alt.values
az_catalogo = np.rad2deg(np.arccos(np.cos(np.deg2rad(data.az.values))))
mega_alt_xi = construct_acumulate_alt_xi(x_exp, y_exp, alt_catalogo, ctes)
mega_az_xi = construct_acumulate_az_xi(x_exp, y_exp, az_catalogo, ctes)
pbar = tqdm(total=1000, desc="a0")
result = amoeba(init_params, mega_alt_xi, mega_az_xi)
pbar.close()
result_2 = np.array([*ctes, *result])
print(result_2)
return result_2
def xi(C, A, F, V, S, D, P, Q, E, ep, X0, Y0, a0):
alt_teorico = construct_altura_deg(C, A, F, V, S, D, P, Q, E, ep, X0, Y0, a0)
return (z - alt_teorico(x, y))**2