def solve(self, init, x):
A = np.zeros([2, 2])
A[0][0] = special.lpmn(self.order, self.degree,
self.function(x[0]))[0][self.order][self.degree]
A[0][1] = special.lqmn(self.order, self.degree,
self.function(x[0]))[0][self.order][self.degree]
A[1][0] = self.function.deriv(1)(x[0]) * special.lpmn(
self.order, self.degree, self.function(
x[0]))[1][self.order][self.degree]
A[1][1] = self.function.deriv(1)(x[0]) * special.lqmn(
self.order, self.degree, self.function(
x[0]))[1][self.order][self.degree]
print(A)
c = np.linalg.solve(A, init)
self.c = c
y = np.zeros((len(x), 2))
for i in range(len(x)):
for j in range(2):
if j == 0:
y[i][j] = \
c[0]*special.lpmn(self.order, self.degree, self.function(x[i]))[j][self.order][self.degree] + \
c[1]*special.lqmn(self.order, self.degree, self.function(x[i]))[j][self.order][self.degree]
elif j == 1:
y[i][j] = \
c[0]*self.function.deriv(1)(x[i])*special.lpmn(self.order, self.degree, self.function(x[i]))[j][self.order][self.degree] + \
c[1]*self.function.deriv(1)(x[i])*special.lqmn(self.order, self.degree, self.function(x[i]))[j][self.order][self.degree]
self.x = x
self.y = y
return y
def cmp_LPQL(Nx,xmin,xmax, maxm,maxl, ist, Sol, Ck, Modd):
Xx=1-xmin+logspace(log10(xmin),log10(xmax),Nx)
# For rombohedral method we need the change of variable to equidistant mesh
# We can change the variable Int[f(x),{x,1,Inf}] = Int[ exp(t) f(1-xmin+exp(t)), {t, log(xmin),inf}]
# Here t is distributed in linear mesh with spacing dxi, and exp(t)=x-1+xmin, because we
# constructued logarithmic mesh by x=1-xmin+exp(t)
Expu = Xx-1.+xmin
dxi = (log(Expu[-1])-log(Expu[0]))/(len(Expu)-1.)
Qlm_all = array([special.lqmn(maxm,maxl,x)[0] for x in Xx])
Plm_all = array([special.lpmn(maxm,maxl,x)[0] for x in Xx])
Qlm_all[0]=0.0 # This diverges, hence we set it to zero
LPL = zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LQL = zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LPxL= zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LQxL= zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LLP = zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LLQ = zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LLPx= zeros((len(ist),maxl+1,len(Xx)),dtype=float)
LLQx= zeros((len(ist),maxl+1,len(Xx)),dtype=float)
for i,(i1,i2) in enumerate(ist):
(R1,Ene1,m1,p1,A1) = Sol[i1]
(R2,Ene2,m2,p2,A2) = Sol[i2]
print 'i1=', i1, 'i2=', i2, 'm1=', m1, 'm2=', m2
m = abs(m1-m2)
Qlm = transpose(Qlm_all[:,m])
Plm = transpose(Plm_all[:,m])
LL = array([Lambda(x,m1,p1,Ck[i1]) * Lambda(x,m2,p2,Ck[i2]) for ix,x in enumerate(Xx)])
for il in range(m,maxl+1):
odd = (Modd[i1]+Modd[i2]+il+m)%2
if odd: continue
LLQ [i,il,:] = LL * Qlm[il,:]
LLQx[i,il,:] = LLQ[i,il,:] * Xx**2
LLP [i,il,:] = LL * Plm[il,:]
LLPx[i,il,:] = LLP[i,il,:] * Xx**2
for ix in range(len(Xx)):
LPL [i,il,ix] = simps(LLP [i,il,:ix+1],Xx[:ix+1])
LPxL[i,il,ix] = simps(LLPx[i,il,:ix+1],Xx[:ix+1])
LQL [i,il,ix] = simps(LLQ [i,il,ix:],Xx[ix:])
LQxL[i,il,ix] = simps(LLQx[i,il,ix:],Xx[ix:])
return (LPL, LQL, LPxL, LQxL, LLP, LLQ, LLPx, LLQx)
def legendre_q_via_lqmn(n, x):
return lqmn(0, n, x)[0][0,-1]
def legendre_q_via_lqmn(n, x):
return lqmn(0, n, x)[0][0,-1]
def lqnm(n, m, z):
return sc.lqmn(int(m.real), int(n.real), z)[0][-1,-1]
def lqnm(n, m, z):
return sc.lqmn(m, n, z)[0][-1,-1]
def fin_potential(domain, x):
"""Compute the potential matrix for a finite domain.
Arguments:
domain (FinDomain) target domain
Returns:
matrix (np.array(fns, fns)) calculated potential matrix
"""
# Rename some useful quantities
max_m = domain.functions
d = x - (min(domain.position) + domain.halfwidth)
# Setup storage for the output
matrix = zeros((max_m, max_m), dtype=double)
iterator = np.nditer(matrix, flags=["multi_index"], op_flags=["readwrite"])
if abs(1 - abs(d) / domain.halfwidth) < 1e-14:
while not iterator.finished:
m, n = iterator.multi_index
mu = max(m, n) + 1
nu = min(m, n) + 1
integral = (sqrt((mu + 1.5) * (nu + 1.5)) / (2 * domain.halfwidth)) * sqrt(
(nu * (nu + 1) * (nu + 2) * (nu + 3))
/ (mu * (mu + 1) * (mu + 2) * (mu + 3))
)
if d < 0:
iterator[0] = (-1) ** (m + n) * integral
else:
iterator[0] = integral
iterator.iternext()
else:
while not iterator.finished:
m, n = iterator.multi_index
m += 1
n += 1
normalisation = (
2.0
* sqrt(
np.float64((m + 1.5) * (n + 1.5))
/ np.float64(
m
* (m + 1)
* (m + 2)
* (m + 3)
* n
* (n + 1)
* (n + 2)
* (n + 3)
)
)
/ domain.halfwidth
)
if d > 1:
legendre = (
lpmn(2, min(m, n) + 1, d / domain.halfwidth)[0][2][-1]
* lqmn(2, max(m, n) + 1, d / domain.halfwidth)[0][2][-1]
)
else:
legendre = (
lpmn(2, min(m, n) + 1, d / domain.halfwidth)[0][2][-1]
* (-1) ** max(m, n)
* lqmn(2, max(m, n) + 1, abs(d / domain.halfwidth))[0][2][-1]
)
if d < 0:
iterator[0] = -normalisation * legendre
else:
iterator[0] = normalisation * legendre
iterator.iternext()
return matrix
def lqnm(n, m, z):
return sc.lqmn(int(m.real), int(n.real), z)[0][-1, -1]
def lqnm(n, m, z):
return sc.lqmn(m, n, z)[0][-1, -1]
