def setup(self):
time_units = self.options['time_units']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
i_step = self.options['i_step']
glm_B = self.options['glm_B']
glm_V = self.options['glm_V']
self.dy_dF = dy_dF = {}
self.add_input('h', units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
shape = state['shape']
F_name = get_name('F', state_name, i_step=i_step)
y_old_name = get_name('y_old', state_name, i_step=i_step)
y_new_name = get_name('y_new', state_name, i_step=i_step)
self.add_input(F_name, shape=(num_stages,) + shape,
units=get_rate_units(state['units'], time_units))
self.add_input(y_old_name, shape=(num_step_vars,) + shape,
units=state['units'])
self.add_output(y_new_name, shape=(num_step_vars,) + shape,
units=state['units'])
F_arange = np.arange(num_stages * size).reshape(
(num_stages,) + shape)
y_arange = np.arange(num_step_vars * size).reshape(
(num_step_vars,) + shape)
# -----------------
# (num_step_vars, num_stages,) + shape
rows = np.einsum('i...,j->ij...', y_arange, np.ones(num_stages, int)).flatten()
cols = np.zeros((num_step_vars, num_stages,) + shape, int).flatten()
self.declare_partials(y_new_name, 'h', rows=rows, cols=cols)
cols = np.einsum('j...,i->ij...', F_arange, np.ones(num_step_vars, int)).flatten()
self.declare_partials(y_new_name, F_name, rows=rows, cols=cols)
# -----------------
# (num_step_vars, num_step_vars,) + shape
data = np.einsum('ij,...->ij...',
glm_V, np.ones(shape)).flatten()
rows = np.einsum('i...,j->ij...', y_arange, np.ones(num_step_vars)).flatten()
cols = np.einsum('j...,i->ij...', y_arange, np.ones(num_step_vars)).flatten()
self.declare_partials(y_new_name, y_old_name, val=data, rows=rows, cols=cols)
def setup(self):
time_units = self.options['time_units']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
i_step = self.options['i_step']
glm_B = self.options['glm_B']
glm_V = self.options['glm_V']
self.dy_dF = dy_dF = {}
self.declare_partials('*', '*', dependent=False)
self.add_input('h', units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
y_old_name = get_name('y_old', state_name, i_step=i_step)
y_new_name = get_name('y_new', state_name, i_step=i_step)
for j_stage in range(num_stages):
F_name = get_name('F',
state_name,
i_step=i_step,
j_stage=j_stage)
self.add_input(F_name,
shape=(1, ) + state['shape'],
units=get_rate_units(state['units'],
time_units))
self.add_input(y_old_name,
shape=(num_step_vars, ) + state['shape'],
units=state['units'])
self.add_output(y_new_name,
shape=(num_step_vars, ) + state['shape'],
units=state['units'])
self.declare_partials(y_new_name, 'h', dependent=True)
y_arange = np.arange(num_step_vars * size).reshape(
(num_step_vars, size))
# num_step_vars, num_step_vars, size
data = np.einsum('ij,...->ij...', glm_V, np.ones(size)).flatten()
rows = np.einsum('i...,j->ij...', y_arange,
np.ones(num_step_vars, int)).flatten()
cols = np.einsum('j...,i->ij...', y_arange,
np.ones(num_step_vars, int)).flatten()
self.declare_partials(y_new_name,
y_old_name,
val=data,
rows=rows,
cols=cols)
for j_stage in range(num_stages):
F_name = get_name('F',
state_name,
i_step=i_step,
j_stage=j_stage)
vals = np.zeros((num_step_vars, 1, size))
rows = np.arange(num_step_vars * 1 * size)
cols = np.zeros((num_step_vars, 1 * size), int)
for ii_step in range(num_step_vars):
vals[ii_step, 0, :] = glm_B[ii_step, j_stage]
cols[ii_step, :] = np.arange(1 * size)
vals = vals.flatten()
cols = cols.flatten()
self.declare_partials(y_new_name, F_name, rows=rows, cols=cols)
dy_dF[state_name, j_stage] = vals
def setup(self):
time_units = self.options['time_units']
num_times = self.options['num_times']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
glm_A = self.options['glm_A']
glm_U = self.options['glm_U']
h_arange = np.arange(num_times - 1)
self.add_input('h_vec', shape=(num_times - 1), units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
shape = state['shape']
F_name = get_name('F', state_name)
y_name = get_name('y', state_name)
Y_out_name = get_name('Y_out', state_name)
Y_in_name = get_name('Y_in', state_name)
Y_arange = np.arange((num_times - 1) * num_stages * size).reshape(
(num_times - 1, num_stages,) + shape)
F_arange = np.arange((num_times - 1) * num_stages * size).reshape(
(num_times - 1, num_stages,) + shape)
y_arange = np.arange(num_times * num_step_vars * size).reshape(
(num_times, num_step_vars,) + shape)
self.add_input(F_name,
shape=(num_times - 1, num_stages,) + shape,
units=get_rate_units(state['units'], time_units))
self.add_input(y_name,
shape=(num_times, num_step_vars,) + shape,
units=state['units'])
self.add_input(Y_in_name, val=0.,
shape=(num_times - 1, num_stages,) + shape,
units=state['units'])
self.add_output(Y_out_name,
shape=(num_times - 1, num_stages,) + shape,
units=state['units'])
# -----------------
ones = -np.ones((num_times - 1) * num_stages * size)
arange = np.arange((num_times - 1) * num_stages * size)
self.declare_partials(Y_out_name, Y_in_name, val=ones, rows=arange, cols=arange)
# -----------------
# (num_times - 1, num_stages, num_stages,) + shape
rows = np.einsum('ij...,k->ijk...', Y_arange, np.ones(num_stages, int)).flatten()
cols = np.einsum('jk...,i->ijk...',
np.ones((num_stages, num_stages,) + shape, int), h_arange).flatten()
self.declare_partials(Y_out_name, 'h_vec', rows=rows, cols=cols)
cols = np.einsum('ik...,j->ijk...', F_arange, np.ones(num_stages, int)).flatten()
self.declare_partials(Y_out_name, F_name, rows=rows, cols=cols)
# -----------------
# (num_times - 1, num_stages, num_step_vars,) + shape
data = np.einsum('jk,i...->ijk...',
glm_U, np.ones((num_times - 1,) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...', Y_arange, np.ones(num_step_vars)).flatten()
cols = np.einsum('ik...,j->ijk...', y_arange[:-1, :, :], np.ones(num_stages)).flatten()
self.declare_partials(Y_out_name, y_name, val=data, rows=rows, cols=cols)
def setup(self):
time_units = self.options['time_units']
num_times = self.options['num_times']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
glm_B = self.options['glm_B']
glm_V = self.options['glm_V']
self.dy_dy = dy_dy = {}
self.dy_dy_inv = dy_dy_inv = {}
h_arange = np.arange(num_times - 1)
self.add_input('h_vec', shape=(num_times - 1), units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
shape = state['shape']
F_name = get_name('F', state_name)
y0_name = get_name('y0', state_name)
y_name = get_name('y', state_name)
y0_arange = np.arange(num_step_vars *
size).reshape((num_step_vars, ) + shape)
y_arange = np.arange(num_times * num_step_vars * size).reshape((
num_times,
num_step_vars,
) + shape)
F_arange = np.arange((num_times - 1) * num_stages * size).reshape((
num_times - 1,
num_stages,
) + shape)
self.add_input(F_name,
shape=(
num_times - 1,
num_stages,
) + shape,
units=get_rate_units(state['units'], time_units))
self.add_input(y0_name,
shape=(num_step_vars, ) + shape,
units=state['units'])
self.add_output(y_name,
shape=(
num_times,
num_step_vars,
) + shape,
units=state['units'])
# -----------------
# (num_times, num_step_vars,) + shape
data1 = np.ones(num_times * num_step_vars * size)
rows1 = np.arange(num_times * num_step_vars * size)
cols1 = np.arange(num_times * num_step_vars * size)
# (num_times - 1, num_step_vars, num_step_vars,) + shape
data2 = np.einsum('i...,jk->ijk...',
np.ones((num_times - 1, ) + shape),
-glm_V).flatten()
rows2 = np.einsum('ij...,k->ijk...', y_arange[1:, :, :],
np.ones(num_step_vars)).flatten()
cols2 = np.einsum('ik...,j->ijk...', y_arange[:-1, :, :],
np.ones(num_step_vars)).flatten()
data = np.concatenate([data1, data2])
rows = np.concatenate([rows1, rows2])
cols = np.concatenate([cols1, cols2])
dy_dy[state_name] = scipy.sparse.csc_matrix(
(data, (rows, cols)),
shape=(num_times * num_step_vars * size,
num_times * num_step_vars * size))
dy_dy_inv[state_name] = scipy.sparse.linalg.splu(dy_dy[state_name])
self.declare_partials(y_name,
y_name,
val=data,
rows=rows,
cols=cols)
# -----------------
# (num_step_vars,) + shape
data = -np.ones((num_step_vars, ) + shape).flatten()
rows = y_arange[0, :, :].flatten()
cols = y0_arange.flatten()
self.declare_partials(y_name,
y0_name,
val=data,
rows=rows,
cols=cols)
# -----------------
# (num_times - 1, num_step_vars, num_stages,) + shape
rows = np.einsum('ij...,k->ijk...', y_arange[1:, :, :],
np.ones(num_stages)).flatten()
cols = np.einsum('jk...,i->ijk...',
np.ones((
num_step_vars,
num_stages,
) + shape), h_arange).flatten()
self.declare_partials(y_name, 'h_vec', rows=rows, cols=cols)
cols = np.einsum('ik...,j->ijk...', F_arange,
np.ones(num_step_vars)).flatten()
self.declare_partials(y_name, F_name, rows=rows, cols=cols)
def setup(self):
time_units = self.options['time_units']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
i_stage = self.options['i_stage']
i_step = self.options['i_step']
glm_A = self.options['glm_A']
glm_U = self.options['glm_U']
self.declare_partials('*', '*', dependent=False)
self.add_input('h', units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
y_old_name = get_name('y_old',
state_name,
i_step=i_step,
i_stage=i_stage)
Y_name = get_name('Y', state_name, i_step=i_step, i_stage=i_stage)
for j_stage in range(i_stage):
F_name = get_name('F',
state_name,
i_step=i_step,
i_stage=i_stage,
j_stage=j_stage)
self.add_input(F_name,
shape=(1, ) + state['shape'],
units=get_rate_units(state['units'],
time_units))
self.add_input(y_old_name,
shape=(num_step_vars, ) + state['shape'],
units=state['units'])
self.add_output(Y_name,
shape=(1, ) + state['shape'],
units=state['units'])
vals = np.zeros((num_step_vars, size))
rows = np.zeros((num_step_vars, size), int)
cols = np.arange(num_step_vars * size)
for ii_step in range(num_step_vars):
vals[ii_step, :] = glm_U[i_stage, ii_step]
rows[ii_step, :] = np.arange(size)
vals = vals.flatten()
rows = rows.flatten()
self.declare_partials(Y_name, 'h', dependent=True)
self.declare_partials(Y_name,
y_old_name,
val=vals,
rows=rows,
cols=cols)
for j_stage in range(i_stage):
F_name = get_name('F',
state_name,
i_step=i_step,
i_stage=i_stage,
j_stage=j_stage)
arange = np.arange(size)
self.declare_partials(Y_name, F_name, rows=arange, cols=arange)
def setup(self):
time_units = self.options['time_units']
num_times = self.options['num_times']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
glm_B = self.options['glm_B']
glm_V = self.options['glm_V']
self.mtx_lu_dict = {}
self.mtx_y0_dict = {}
self.mtx_h_dict = {}
self.mtx_hf_dict = {}
self.num_y0_dict = {}
self.num_F_dict = {}
self.num_y_dict = {}
h_arange = np.arange(num_times - 1)
num_h = num_times - 1
self.add_input('h_vec', shape=(num_times - 1), units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
shape = state['shape']
y0_name = get_name('y0', state_name)
F_name = get_name('F', state_name)
y_name = get_name('y', state_name)
# --------------------------------------------------------------------------------
y0_arange = np.arange(num_step_vars *
size).reshape((num_step_vars, ) + shape)
F_arange = np.arange((num_times - 1) * num_stages * size).reshape((
num_times - 1,
num_stages,
) + shape)
y_arange = np.arange(num_times * num_step_vars * size).reshape((
num_times,
num_step_vars,
) + shape)
num_y0 = np.prod(y0_arange.shape)
num_F = np.prod(F_arange.shape)
num_y = np.prod(y_arange.shape)
# --------------------------------------------------------------------------------
self.add_input(y0_name,
shape=(num_step_vars, ) + shape,
units=state['units'])
self.add_input(F_name,
shape=(
num_times - 1,
num_stages,
) + shape,
units=get_rate_units(state['units'], time_units))
self.add_output(y_name,
shape=(
num_times,
num_step_vars,
) + shape,
units=state['units'])
# --------------------------------------------------------------------------------
data_list = []
rows_list = []
cols_list = []
# y identity
data = np.ones(num_y)
rows = np.arange(num_y)
cols = np.arange(num_y)
data_list.append(data)
rows_list.append(rows)
cols_list.append(cols)
# V blocks: (num_times - 1) x num_step_var x num_step_var x ...
data = np.einsum('jk,i...->ijk...', -glm_V,
np.ones((num_times - 1, ) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...', y_arange[1:, :, :],
np.ones(num_step_vars, int)).flatten()
cols = np.einsum('ik...,j->ijk...', y_arange[:-1, :, :],
np.ones(num_step_vars, int)).flatten()
data_list.append(data)
rows_list.append(rows)
cols_list.append(cols)
# concatenate
data = np.concatenate(data_list)
rows = np.concatenate(rows_list)
cols = np.concatenate(cols_list)
mtx = scipy.sparse.csc_matrix((data, (rows, cols)),
shape=(num_y, num_y))
self.mtx_lu_dict[state_name] = scipy.sparse.linalg.splu(mtx)
# --------------------------------------------------------------------------------
data = np.ones(num_y0)
rows = y_arange[0, :, :].flatten()
cols = np.arange(num_y0)
self.mtx_y0_dict[state_name] = scipy.sparse.csc_matrix(
(data, (rows, cols)), shape=(num_y, num_y0))
# --------------------------------------------------------------------------------
data = np.ones(num_F)
rows = np.arange(num_F)
cols = np.einsum('i,j...->ij...', h_arange,
np.ones((num_stages, ) + shape, int)).flatten()
self.mtx_h_dict[state_name] = scipy.sparse.csc_matrix(
(data, (rows, cols)), shape=(num_F, num_h))
# --------------------------------------------------------------------------------
# B blocks: (num_times - 1) x num_step_vars x num_stage x ...
data = np.einsum('jk,i...->ijk...', glm_B,
np.ones((num_times - 1, ) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...', y_arange[1:, :, :],
np.ones(num_stages, int)).flatten()
cols = np.einsum('ik...,j->ijk...', F_arange,
np.ones(num_step_vars, int)).flatten()
self.mtx_hf_dict[state_name] = scipy.sparse.csc_matrix(
(data, (rows, cols)), shape=(num_y, num_F))
def setup(self):
time_units = self.options['time_units']
num_times = self.options['num_times']
num_stages = self.options['num_stages']
num_step_vars = self.options['num_step_vars']
glm_A = self.options['glm_A']
glm_U = self.options['glm_U']
glm_B = self.options['glm_B']
glm_V = self.options['glm_V']
self.mtx_y0_dict = {}
self.mtx_dict = {}
self.mtx_h_dict = {}
h_arange = np.arange(num_times - 1)
num_h = num_times - 1
self.add_input('h_vec', shape=(num_times - 1), units=time_units)
for state_name, state in iteritems(self.options['states']):
size = np.prod(state['shape'])
shape = state['shape']
y0_name = get_name('y0', state_name)
F_name = get_name('F', state_name)
Y_in_name = get_name('Y_in', state_name)
Y_out_name = get_name('Y_out', state_name)
# --------------------------------------------------------------------------------
y0_arange = np.arange(num_step_vars * size).reshape((num_step_vars,) + shape)
F_arange = np.arange((num_times - 1) * num_stages * size).reshape(
(num_times - 1, num_stages,) + shape)
Y_arange = np.arange((num_times - 1) * num_stages * size).reshape(
(num_times - 1, num_stages,) + shape)
y_arange = np.arange(num_times * num_step_vars * size).reshape(
(num_times, num_step_vars,) + shape)
num_y0 = np.prod(y0_arange.shape)
num_F = np.prod(F_arange.shape)
num_Y = np.prod(Y_arange.shape)
num_y = np.prod(y_arange.shape)
# --------------------------------------------------------------------------------
self.add_input(y0_name,
shape=(num_step_vars,) + shape,
units=state['units'])
self.add_input(F_name,
shape=(num_times - 1, num_stages,) + shape,
units=get_rate_units(state['units'], time_units))
self.add_input(Y_in_name, val=0.,
shape=(num_times - 1, num_stages,) + shape,
units=state['units'])
self.add_output(Y_out_name,
shape=(num_times - 1, num_stages,) + shape,
units=state['units'])
# -----------------
self.declare_partials(Y_out_name, 'h_vec')
self.declare_partials(Y_out_name, y0_name)
self.declare_partials(Y_out_name, F_name)
# -----------------
ones = -np.ones((num_times - 1) * num_stages * size)
arange = np.arange((num_times - 1) * num_stages * size)
self.declare_partials(Y_out_name, Y_in_name, val=ones, rows=arange, cols=arange)
# --------------------------------------------------------------------------------
# mtx_y0: num_stages x num_step_vars x ...
data = np.ones((num_step_vars,) + shape).flatten()
rows = y_arange[0, :, :].flatten()
cols = y0_arange.flatten()
mtx_y0 = scipy.sparse.csc_matrix((data, (rows, cols)), shape=(num_y, num_y0)).toarray()
# --------------------------------------------------------------------------------
# mtx_A: (num_times - 1) x num_stages x num_stages x ...
data = np.einsum('jk,i...->ijk...',
glm_A, np.ones((num_times - 1,) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...',
Y_arange, np.ones(num_stages, int)).flatten()
cols = np.einsum('ik...,j->ijk...',
F_arange, np.ones(num_stages, int)).flatten()
mtx_A = scipy.sparse.csc_matrix((data, (rows, cols)), shape=(num_Y, num_F)).toarray()
# --------------------------------------------------------------------------------
# mtx_B: (num_times - 1) x num_step_vars x num_stages x ...
data = np.einsum('jk,i...->ijk...',
glm_B, np.ones((num_times - 1,) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...',
y_arange[1:, :, :], np.ones(num_stages, int)).flatten()
cols = np.einsum('ik...,j->ijk...',
F_arange, np.ones(num_step_vars, int)).flatten()
mtx_B = scipy.sparse.csc_matrix((data, (rows, cols)), shape=(num_y, num_F)).toarray()
# --------------------------------------------------------------------------------
# mtx_U: (num_times - 1) x num_stages x num_step_vars x ...
data = np.einsum('jk,i...->ijk...',
glm_U, np.ones((num_times - 1,) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...',
Y_arange, np.ones(num_step_vars, int)).flatten()
cols = np.einsum('ik...,j->ijk...',
y_arange[:-1, :, :], np.ones(num_stages, int)).flatten()
mtx_U = scipy.sparse.csc_matrix((data, (rows, cols)), shape=(num_Y, num_y)).toarray()
# --------------------------------------------------------------------------------
# mtx_y
data_list = []
rows_list = []
cols_list = []
# identity
data = np.ones(num_y)
rows = np.arange(num_y)
cols = np.arange(num_y)
data_list.append(data); rows_list.append(rows); cols_list.append(cols)
# (num_times - 1) x num_step_var x num_step_var x ...
data = np.einsum('jk,i...->ijk...',
-glm_V, np.ones((num_times - 1,) + shape)).flatten()
rows = np.einsum('ij...,k->ijk...',
y_arange[1:, :, :], np.ones(num_step_vars, int)).flatten()
cols = np.einsum('ik...,j->ijk...',
y_arange[:-1, :, :], np.ones(num_step_vars, int)).flatten()
data_list.append(data); rows_list.append(rows); cols_list.append(cols)
# concatenate
data = np.concatenate(data_list)
rows = np.concatenate(rows_list)
cols = np.concatenate(cols_list)
mtx_y = scipy.sparse.csc_matrix((data, (rows, cols)), shape=(num_y, num_y))
mtx_y_inv = scipy.sparse.linalg.splu(mtx_y)
# --------------------------------------------------------------------------------
# mtx_h
data = np.ones(num_F)
rows = np.arange(num_F)
cols = np.einsum('i,j...->ij...',
h_arange, np.ones((num_stages,) + shape, int)).flatten()
mtx_h = scipy.sparse.csc_matrix((data, (rows, cols)), shape=(num_F, num_h)).toarray()
# --------------------------------------------------------------------------------
self.mtx_y0_dict[state_name] = mtx_U.dot(mtx_y_inv.solve(mtx_y0))
self.mtx_dict[state_name] = mtx_A + mtx_U.dot(mtx_y_inv.solve(mtx_B))
self.mtx_h_dict[state_name] = mtx_h