def sym_grad_vector_basis(self, p=None):
phi = self.basis(p=p)
zero = sp.ZeroMatrix(1, phi.cols)
A00 = sp.BlockMatrix([[phi.diff(self.x), zero]])
A11 = sp.BlockMatrix([[zero, phi.diff(self.y)]])
A01 = sp.BlockMatrix([[phi.diff(self.y) / 2, phi.diff(self.x) / 2]])
return A00, A11, A01
def matrix_E(self, p=None):
p = self.p if p is None else p # p >=2
c = np.repeat(np.arange(1, p), np.arange(1, p))
ldof0 = self.number_of_dofs(p=p - 2)
E00 = sp.zeros(ldof0, ldof0)
E10 = sp.zeros(ldof0, ldof0)
for i in range(ldof0):
E00[i, i] = self.h**2 / c[i]
E10[i, i] = self.h**2 / c[i]
if p == 2:
return sp.BlockMatrix([[E00], [E10]])
else:
ldof1 = self.number_of_dofs(p=p - 3)
E01 = sp.zeros(ldof0, ldof1)
E11 = sp.zeros(ldof0, ldof1)
idx = self.diff_index_1(p=p - 2)
x = idx['x']
y = idx['y']
for i in range(ldof1):
E01[y[0][i], i] = sp.Rational(y[1][i] * self.h**2, c[y[0][i]])
E11[x[0][i], i] = sp.Rational(-x[1][i] * self.h**2, c[x[0][i]])
return sp.BlockMatrix([[E00, E01], [E10, E11]])
def get_tf_sp(self):
self.T0S = sp.BlockMatrix([
[self.R0S, sp.zeros(3, 1)], [sp.zeros(1, 3),
sp.eye(1)]
]).as_explicit() # Transformation function for parallel system.
self.T0S_M = sp.BlockMatrix([
[self.R0S_M, sp.zeros(3, 1)], [sp.zeros(1, 3),
sp.eye(1)]
]).as_explicit() # Transformation function for parallel system.
def _construct_force_vector(self):
# Ti_e = 2E(Pi).T * (M + (ui x Fi))
Fi = self.Fi(self.t)
Ti = self.Ti(self.t)
Ti_e = 2*E(self.Pi).T * (Ti + Skew(self.ui)*Fi)
self._Qi = sm.BlockMatrix([[Fi], [Ti_e]])
Fj = -Fi
Tj = -Ti
Tj_e = 2*E(self.Pj).T * (Ti + Skew(self.uj)*Fj)
self._Qj = sm.BlockMatrix([[Fj], [Tj_e]])
def get_foreman():
# Basis transformation
v_n = lambda i, N: sympy.Matrix( # noqa: E731
[1 if i == n else 0 for n in range(N)])
S, X, Y, Z = [v_n(i, 4) for i in range(4)]
up, dw = [v_n(i, 2) for i in range(2)]
X = sympy.I * X
Y = sympy.I * Y
Z = sympy.I * Z
molenkamp_basis = [
kr(up, S),
kr(dw, S),
+(1 / sympy.sqrt(2)) * kr(up, X + sympy.I * Y),
+(1 / sympy.sqrt(6)) * (kr(dw, X + sympy.I * Y) - kr(up, 2 * Z)),
-(1 / sympy.sqrt(6)) * (kr(up, X - sympy.I * Y) + kr(dw, 2 * Z)),
-(1 / sympy.sqrt(2)) * kr(dw, X - sympy.I * Y),
+(1 / sympy.sqrt(3)) * (kr(dw, X + sympy.I * Y) + kr(up, Z)),
+(1 / sympy.sqrt(3)) * (kr(up, X - sympy.I * Y) - kr(dw, Z)),
]
subs_notation = {
Ec: Ev + E0,
L: -(g1 + 4 * g2) * (hbar**2 / 2 / m0),
M: -(g1 - 2 * g2) * (hbar**2 / 2 / m0),
Np: -(3 * g3 + (3 * kappa + 1)) * (hbar**2 / 2 / m0),
Nm: -(3 * g3 - (3 * kappa + 1)) * (hbar**2 / 2 / m0),
Ac: (g0 * hbar**2 / 2 / m0),
}
Hs = spin_orbit()
Hcc = sympy.Matrix([Ec + kx * Ac * kx + ky * Ac * ky + kz * Ac * kz])
Hcv = +sympy.I * sympy.Matrix([[P * kx, P * ky, P * kz]])
Hvc = -sympy.I * sympy.Matrix([kx * P, ky * P, kz * P])
data = [[valence_term(i, j) for j in range(3)] for i in range(3)]
Hvv = sympy.Matrix(data)
H4 = sympy.BlockMatrix([[Hcc, Hcv], [Hvc, Hvv]])
H8 = sympy.Matrix(sympy.BlockDiagMatrix(H4, H4)) + Hs
dag = lambda x: x.conjugate().transpose() # noqa: E731
U_dag = sympy.Matrix(sympy.BlockMatrix([molenkamp_basis]))
U = dag(U_dag)
hamiltonian = U * H8 * dag(U)
hamiltonian = hamiltonian.subs(subs_notation)
hamiltonian = hamiltonian + (Ev - Delta / 3) * sympy.diag(
0, 0, 1, 1, 1, 1, 1, 1)
return sympy.ImmutableMatrix(hamiltonian.expand())
def get_rotation_matrix(self):
self.jyhat = -self.XOE # make x-axis of joystick to be on XOE plane. Since x & z-axis used to make xoe plane and y-axis has to perpendicular to both,
# y-axis is thus perpendicular to XOE, i.e. the norm of XOE.
# make x-axis to be on xoe plane, and hence to find y axis, cross(jyz,xhat) or y=z X x. therefore jyhat=-(XOE), !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
self.jxhat = self.jyhat.cross(
self.jzhat
) # already unit vector, since jzhat and jyhat are perpendicular and are unit vectors.
self.R0S = sp.BlockMatrix([[self.jxhat, self.jyhat,
self.jzhat]]).as_explicit()
self.R0S_M = sp.BlockMatrix(
[[self.jxhat_M, self.jyhat_M,
self.jzhat_M]]).as_explicit() # todo: check
def __init__(self, *args, **kwargs):
name = 'ground'
super().__init__(name)
self.P_ground = quatrenion('Pg_%s' % self.name,
format_as=r'{Pg_{%s}}' % self.name)
self.normalized_pos_equation = sm.BlockMatrix(
[[self.R], [self.P - self.P_ground]])
self.normalized_vel_equation = sm.BlockMatrix([[zero_matrix(3, 1)],
[zero_matrix(4, 1)]])
self.normalized_acc_equation = sm.BlockMatrix([[zero_matrix(3, 1)],
[zero_matrix(4, 1)]])
self.normalized_jacobian = sm.BlockMatrix(
[[sm.Identity(3), zero_matrix(3, 4)],
[zero_matrix(4, 3), sm.Identity(4)]])
def hamilton(M):
text = "Metoda Cayleya-Hamiltona<br>"
eigen_values, eigen_vectors = eigen(M)
text += 'Eigenvalues: $' + sp.latex(eigen_values) + ' = ' + sp.latex(
eigen_values.evalf(5)) + '$<br>'
eigen_values = eigen_values.evalf(5)
x = sp.symbols(
str(['a' + str(i)
for i in range(len(eigen_values))])[1:-1].replace("'", ""))
if len(eigen_values) == 1:
x = [x]
A = sp.ones(len(x), 1)
for i in range(1, len(x)):
A = sp.Matrix(sp.BlockMatrix([A, sp.HadamardPower(eigen_values, i)]))
b = sp.Matrix(sp.HadamardPower(sp.E, eigen_values * t))
result = sp.Matrix(list(sp.linsolve((A, b), *x))[0])
text += "$"+sp.latex(b)+' = '+sp.latex(A)+'\\bullet'+sp.latex(sp.Matrix(x))+'\\implies' + \
sp.latex(sp.Matrix(x))+' = '+sp.latex(result) + "$<br>"
s = sp.diag(*tuple([1] * len(x))) * result[0, 0]
for i in range(1, len(x)):
s += M * result[i, 0]
text += "$e^{\\mathbb{A}t} = " + sp.latex(sp.simplify(s)) + "$<br><br>"
return s, text
def _preare_second_phase(self):
self.non_basic_variables = [
i for i in self.non_basic_variables
if i not in self.artificial_variables
]
A = self.T_star[:, self.non_basic_variables]
I = sy.eye(len(self.basic_variables))
b = self.T_star[:, -1]
T = sy.Matrix(sy.BlockMatrix([[A, I, b]]))
var_names = np.array(self.variable_names)
non_basic_varnames = list(var_names[self.non_basic_variables])
basic_varnames = list(var_names[self.basic_variables])
new_varnames = non_basic_varnames + basic_varnames
t = sy.zeros(1, len(new_varnames) + 1)
for var_name in self.objective:
index = new_varnames.index(var_name)
t[0, index] = -self.objective[var_name]
for var_name in self.objective:
index = basic_varnames.index(var_name)
t = t + self.objective[var_name] * T[index, :]
# print(t)
# print(T)
# print(self.variable_names)
# print(self.basic_variables)
# print(self.non_basic_variables)
self.t = t
self.T = T
self.running_variable_names = new_varnames
def M(i, n=None):
"""Elimination theory matrix: i^th matrix in the recursion with n external legs. Default is i = n."""
if n is None:
n = i
zs = punctures(n)
if i < 3:
raise Exception("Elimination theory matrix: invalid argument.")
elif i == 3:
return sympy.Matrix([])
elif i == 4:
return hms(n)
else:
M_im1 = M(i - 1, n)
M_im1_diff = sympy.diff(M_im1, zs[i - 3 - 1])
M_im1 = M_im1.subs(
zs[i - 3 - 1], 0
) # just a trick to simplify calculation, nothing to do with Mobius
rows, columns = M_im1.shape
M_i = sympy.Matrix(
sympy.BlockMatrix(
tuple(
tuple([
M_im1 if _i == _j else M_im1_diff if _i == _j +
1 else sympy.zeros(rows, columns)
for _i in range(i - 3)
]) for _j in range(i - 4))))
return M_i
def Psi(n):
"""Psi anti-symmetric matrix (reduced)."""
ks = sympy.symbols('k1:{}'.format(n + 1))
es = sympy.symbols('e1:{}'.format(n + 1))
zs = sympy.symbols('z1:{}'.format(n + 1))
# A Matrix
A = sympy.zeros(n, n)
for a in range(n):
for b in range(a + 1, n):
A[a, b] = 2 * ks[a] * ks[b] / (zs[a] - zs[b])
A[b, a] = -A[a, b]
# B Matrix
B = sympy.zeros(n, n)
for a in range(n):
for b in range(a + 1, n):
B[a, b] = 2 * es[a] * es[b] / (zs[a] - zs[b])
B[b, a] = -B[a, b]
# C Matrix
C = sympy.zeros(n, n)
for a in range(n):
for b in range(n):
if a == b:
C[a, b] = -sum([
2 * es[a] * ks[j] / (zs[a] - zs[j])
for j in range(n) if j != a
])
else:
C[a, b] = 2 * es[a] * ks[b] / (zs[a] - zs[b])
# Psi - from block matrix
Psi = sympy.Matrix(sympy.BlockMatrix(((A, -C.T), (C, B))))
# drop rows & cols 1 and 2 (i.e. 0 and 1)
Psi = Psi[range(2, 2 * n), range(2, 2 * n)]
# simplify operations involving infinity
for i in range(1, n):
Psi = Psi.subs(((zs[0] - zs[i], zs[0]), (zs[i] - zs[0], -zs[0]),
(zs[i] + zs[0], zs[0]), (-zs[i] - zs[0], -zs[0])))
# pull out z1 ** -2 --- Pfaffian goes aas root of this
for i in range(2 * n - 2):
if Psi[-n, i] != 0:
Psi[-n, i] = Psi[-n, i] * zs[0]
if Psi[i, -n] != 0:
Psi[i, -n] = Psi[i, -n] * zs[0]
# Mobius fix
Psi = Psi.subs(((zs[0], sympy.oo), (zs[1], 1), (zs[n - 1], 0)))
# simplify
for i in range(2, n):
Psi = Psi.subs(((sympy.oo - zs[i], sympy.oo), ))
Psi = Psi.subs(((-sympy.oo + zs[i], -sympy.oo), ))
Psi = Psi.subs(((sympy.oo + zs[i], sympy.oo), ))
return Psi
def Displace(d_row):
d1, d2, d3 = d_row
I = smp.Matrix.eye(3)
Z = smp.zeros(1, 3)
p = smp.Matrix([[d1], [d2], [d3]])
blockMatrix = smp.BlockMatrix([[I, p], [Z, smp.Matrix([1])]])
return smp.Matrix(blockMatrix)
def pos_level_equations(self):
"""
A block matrix that stores the set of matrix/vector equations that
represents the position constraints equations. The shape of the blocks
is (nve, 1), where the scalar shape is (nc, 1).
"""
return sm.BlockMatrix(self._pos_level_equations)
def perp_grad_space_basis(self, p=None):
p = self.p if p is None else p
if p > 1:
phi = self.basis(p=p - 1)
return sp.BlockMatrix([phi[2] * phi, -phi[1] * phi])
else:
return sp.ZeroMatrix(2, 1)
def jacobian_j(self):
"""
A block matrix that stores the jacobian of the constraint equations
with respect to the body_j generalized coordinates. The shape of the
blocks is (nve, 2), where the scalar shape is (nc, 7).
"""
return sm.BlockMatrix(self._jacobian_j)
def acc_level_equations(self):
"""
A block matrix that stores the set of matrix/vector equations that
represents the right hand-side of the acceleration level of the
constraints equations. The shape of the blocks is (nve, 1), where the
scalar shape is (nc, 1).
"""
return sm.BlockMatrix(self._acc_level_equations)
def construct_matrix_from_block_matrix(A_expr):
A_expr = sympy.BlockMatrix(A_expr)
A_collapsed_expr = sympy.Matrix.zeros(A_expr.rows, A_expr.cols)
for r in range(A_expr.rows):
for c in range(A_expr.cols):
A_collapsed_expr[r, c] = A_expr[r, c]
return A_collapsed_expr
def test_BlockMatrix():
n = sy.Symbol('n', integer=True)
A, B, C, D = [sy.MatrixSymbol(name, n, n) for name in 'ABCD']
At, Bt, Ct, Dt = map(theano_code_, (A, B, C, D))
Block = sy.BlockMatrix([[A, B], [C, D]])
Blockt = theano_code_(Block)
solutions = [tt.join(0, tt.join(1, At, Bt), tt.join(1, Ct, Dt)),
tt.join(1, tt.join(0, At, Ct), tt.join(0, Bt, Dt))]
assert any(theq(Blockt, solution) for solution in solutions)
def _construct_force_vector(self):
dij = self.joint.dij
distance = sm.sqrt(dij.T * dij)
unit_vector = dij / distance
defflection = self.LF - distance[0, 0]
velocity = (unit_vector.T * self.joint.dijd)
velocity = sm.sqrt(velocity.T * velocity)[0, 0]
self.Fi = unit_vector * (self.Fs(defflection) - self.Fd(velocity))
Ti_e = 2 * G(self.Pi).T * (self.Ti + Skew(self.ui).T * self.Fi)
self._Qi = sm.BlockMatrix([[self.Fi], [Ti_e]])
self.Fj = -self.Fi
Tj_e = 2 * G(self.Pj).T * (self.Tj + Skew(self.uj).T * self.Fj)
self._Qj = sm.BlockMatrix([[self.Fj], [Tj_e]])
def HomogeneousInverse(T):
R = T[0:3, 0:3]
p = T[0:3, 3]
R_inv = smp.Matrix.transpose(R)
p_inv = -R_inv * p
blockMatrix = smp.BlockMatrix([[R_inv, p_inv],
[smp.zeros(1, 3),
smp.Matrix([1])]])
return smp.Matrix(blockMatrix)
def ScrewTose3(S):
S1 = smp.Matrix(S).reshape(6, 1)
w = S1[0:3, 0].reshape(1, 3)
v = S1[3:, 0].reshape(3, 1)
wSkew = SkewSymmetric(w)
blockMatrix = smp.BlockMatrix([[wSkew, v],
[smp.zeros(1, 3),
smp.Matrix([0])]])
return smp.Matrix(blockMatrix)
def _construct_force_vector(self):
dij = (self.Ri + self.ui - self.Rj - self.uj)
dijd = (self.Rdi + self.Bui*self.Pdi - self.Rdj - self.Buj*self.Pdj)
bush_trasformation = self.mi_bar.A.T * self.Ai.T
F_bush_i = self.Fs(bush_trasformation, dij) \
+ self.Fd(bush_trasformation, dijd)
self.Fi = self.Ai * self.mi_bar.A * -F_bush_i
Ti_e = - 2*E(self.Pi).T * Skew(self.ui).T * self.Fi
#Ti_e = -(self.Ai * Skew(self.ui_bar) * 2*G(self.Pi)).T * self.Fi
self._Qi = sm.BlockMatrix([[self.Fi], [Ti_e]])
self.Fj = -self.Fi
Tj_e = - 2*E(self.Pj).T * Skew(self.uj).T * self.Fj
#Tj_e = -(self.Aj * Skew(self.uj_bar) * 2*G(self.Pj)).T * self.Fj
self._Qj = sm.BlockMatrix([[self.Fj], [Tj_e]])
def r_lambda(M):
_N = M[:-1, :-1]
nb_rows, nb_cols = _N.shape
fac = sp.factorial
eN = sum([_N**i / fac(i) for i in range(0, n + 1)],
start=sp.zeros(*_N.shape))
return sp.Matrix(
sp.BlockMatrix(
[[eN, sp.zeros(nb_rows, 1)],
[sp.zeros(1, nb_cols),
sp.Matrix([sp.exp(M[-1, -1])])]]))
def r_lambda(M):
_N = M[:-1, :-1]
nb_rows, nb_cols = _N.shape
hnm = _h(n, m)(_N).evalf()
knm = _k(n, m)(_N).evalf()
iknm = sp.refine(invert(knm), sp.Q.integer(k))
eN = sp.refine(sp.simplify(hnm) * iknm, sp.Q.integer(k))
return sp.Matrix(
sp.BlockMatrix(
[[eN, sp.zeros(nb_rows, 1)],
[sp.zeros(1, nb_cols),
sp.Matrix([sp.exp(M[-1, -1])])]]))
def test_BlockMatrix():
n = sympy.Symbol('n', integer=True)
A = sympy.MatrixSymbol('A', n, n)
B = sympy.MatrixSymbol('B', n, n)
C = sympy.MatrixSymbol('C', n, n)
D = sympy.MatrixSymbol('D', n, n)
At, Bt, Ct, Dt = map(theano_code, (A, B, C, D))
Block = sympy.BlockMatrix([[A, B], [C, D]])
Blockt = theano_code(Block)
solutions = [tt.join(0, tt.join(1, At, Bt), tt.join(1, Ct, Dt)),
tt.join(1, tt.join(0, At, Ct), tt.join(0, Bt, Dt))]
assert any(theq(Blockt, solution) for solution in solutions)
def balancer(left_side, right_side): #Algorithm based on: http://www.logical.ai/chemistry/html/chem-nullspace.html
"""Takes in left_side, which is a list of tuples (left side is number of molecules (0 if unknown), right side is the molecule itself)."""
#Test to ensure each side has the same elements
left_element_list = []
for mol in left_side: #Loads atoms into list
for atom in mol[1].atoms:
if atom[0] not in left_element_list:
left_element_list.append(atom[0])
right_element_list = []
for mol in right_side: #Loads atoms into other list
for atom in mol[1].atoms:
if atom[0] not in right_element_list:
right_element_list.append(atom[0])
left_element_list.sort() #Sorts lists into order (so they can be compared)
right_element_list.sort()
if left_element_list != right_element_list:
raise Exception("Different elements on left and right side of chemical equation")
molecule_list = left_side + right_side
#Create array
element_matrix = sympy.zeros(len(left_element_list), len(left_side) + len(right_side))
for i in range(0, len(molecule_list)):
for atom in molecule_list[i][1].atoms:
element_matrix[left_element_list.index(atom[0]), i] = atom[1] #Uses the left side list to determine the order that different elements go into the matrix- it is arbitrary, as long as it is consistent
element_matrix = element_matrix.T
element_matrix = sympy.BlockMatrix([element_matrix, sympy.eye(len(left_side + right_side))], axis=1)
element_matrix_rref = sympy.Matrix(element_matrix).rref()
solution_list = []
for i in range(0, element_matrix_rref[0].shape[0]): #Loop through list, and test if ansewers are solutions according to the algorithm (i.e. their pivots are further right than the number of atoms added)
if element_matrix_rref[1][i] >= len(left_element_list):
solution_list.append(sympy.Matrix(element_matrix_rref[0].row(i)))
if len(solution_list) == 0:
return None
#Make negatives positive
for solution in solution_list:
for i in range(0, len(solution)):
if solution[i] < 0:
solution[i] = solution[i] * -1
#Turn fractions into whole numbers
for i in range(0, len(solution_list)):
while True:
largest_denominator = 1 #Finds largest denominator, then multiplies by that until all denominators are 1
for j in range(0, solution_list[i].shape[1]):
if solution_list[i][j].q != 1:
if solution_list[i][j].q > largest_denominator:
largest_denominator = solution_list[i][j].q
if largest_denominator != 1:
solution_list[i] = solution_list[i] * largest_denominator
else:
break
solution = solution_list[0].tolist()[0] #Convert to list
solution = solution[len(left_element_list):]
return solution
def _construct_force_vector(self):
dij = (self.Ri + self.ui - self.Rj - self.uj)
dijd = (self.Rdi + self.Bui*self.Pdi - self.Rdj - self.Buj*self.Pdj)
dij_bush_i = self.mi_bar.A.T * self.Ai.T * dij
dijd_bush_i = self.mi_bar.A.T * self.Ai.T * dijd
F_bush_i = (self.Kt*sm.Identity(3) * dij_bush_i) \
+ (self.Ct*sm.Identity(3) * dijd_bush_i)
self.Fi = self.Ai * self.mi_bar.A * -F_bush_i
Ti_e = - 2*E(self.Pi).T * Skew(self.ui).T * self.Fi
self._Qi = sm.BlockMatrix([[self.Fi], [Ti_e]])
dij_bush_j = self.mj_bar.A.T * self.Aj.T * dij
dijd_bush_j = self.mj_bar.A.T * self.Aj.T * dijd
F_bush_j = (self.Kt*sm.Identity(3) * dij_bush_j) \
+ (self.Ct*sm.Identity(3) * dijd_bush_j)
self.Fj = self.Aj * self.mj_bar.A * F_bush_j
Tj_e = - 2*E(self.Pj).T * Skew(self.uj).T * self.Fj
self._Qj = sm.BlockMatrix([[self.Fj], [Tj_e]])
def _construct_force_vector(self):
dij = (self.Ri + self.ui - self.Rj - self.uj)
dijd = (self.Rdi + self.Bui*self.Pdi - self.Rdj - self.Buj*self.Pdj)
# Fs = K(l - l0) + C*ld + Fa
l = sm.sqrt(dij.T*dij)[0,0]
l0 = self.LF
unit_vector = dij/l
ld = -((unit_vector).T * dijd)
defflection = l0 - l
Fs = -self.Fs(defflection) - self.Fd(ld)
self.Fi = -Fs * unit_vector
Ti_e = Fs * 2*E(self.Pi).T * Skew(self.ui).T * unit_vector
self.Fj = -self.Fi
Tj_e = -Fs * 2*E(self.Pj).T * Skew(self.uj).T * unit_vector
self._Qi = sm.BlockMatrix([[self.Fi], [Ti_e]])
self._Qj = sm.BlockMatrix([[self.Fj], [Tj_e]])
def my_exp(L, t, k):
"""
In fact we don't need to a `sp.MatrixSymbol` exponential
We only need to return a matrix of function : (`matrix_ij(dt,k)`)_{i,j}
"""
nb_cols, nb_rows = L.shape
matExp = sp.Matrix(
sp.BlockMatrix([[
sp.Matrix([[
sp.Function("matrixExpr{}{}".format(i, j))(t, k)
for j in range(nb_cols - 1)
] for i in range(nb_rows - 1)]),
sp.zeros(nb_rows - 1, 1)
], [sp.zeros(1, nb_cols - 1),
sp.Matrix([sp.exp(t * L[-1, -1])])]]))
return matExp
def funcMatExp(L, t, k):
"""
In fact we don't need to compute the real $e^{tL}$ with a classic
exponential function, or with Pade approximant or Taylor serie, we
only need to manipulate the matrix of IT functions.
This function returns the matrix of IT functions.
"""
nb_cols, nb_rows = L.shape
B = sp.Matrix([[
sp.Function("matrixExpr{}{}".format(i, j))(t, k)
for j in range(nb_cols - 1)
] for i in range(nb_rows - 1)])
matExp = sp.Matrix(
sp.BlockMatrix(
[[B, sp.zeros(nb_rows - 1, 1)],
[sp.zeros(1, nb_cols - 1),
sp.Matrix([sp.exp(t * L[-1, -1])])]]))
return matExp