def hess(self, x, z=None, **kwargs):
"""Return the dense Hessian of the objective at x."""
if z is None:
z = np.zeros(self.ncon)
xz = np.concatenate((x, z))
H = adolc.hessian(self._lag_trace_id, xz)
return H[:self.nvar, :self.nvar]
def hessLag_adolc(x, lagrange, obj_factor, flag, user_data=None):
if flag:
result = (given['rind'], given['cind'])
else:
result = np.ravel(adolc.hessian(3, x)[given['rind'],
given['cind']],
order="C")
return result
def __call__(self, x0, lagrange, obj_factor, flag, user_data = None):
if flag:
return (self.rind, self.cind)
else:
x = np.hstack([x0,lagrange,obj_factor])
result = adolc.hessian(lID,x)
result1 = result[:nvar,:nvar]
result = None
result = sps.triu(result1,format='coo')
return result.data
def __init__(self, x0, lagrange, obj_factor):
x = np.hstack([x0,lagrange,obj_factor])
result = adolc.hessian(lID,x)
print 'hess initialized'
result1 = result[:nvar,:nvar]
result = None
result = sps.triu(result1,format='coo')
result1 = None
self.rind = result.row
self.cind = result.col
self.values = result.data
# Only keep hess values w/ respect to x, not lagrange/obj_factor
self.nnz = result.nnz
def ppRpWpW_adolc(self, sim, W, area):
tag = self.ppRpWpW_tag
if not self.ppRpWpW_traced:
aW = adouble(W.flatten(order='F'))
aarea = adouble(area)
adolc.trace_on(tag)
adolc.independent(aW)
aW3 = np.reshape(aW, W.shape, order='F')
pRpW = self.pRpW_adolc(sim, aW3, area)
pRpW.flatten(order='F')
adolc.dependent(pRpW)
adolc.trace_off()
return adolc.hessian(self.ppRpWpW_tag, W.flatten(order='F'))
def hessian(self,x):
return adolc.hessian(self.tape_number,
np.ravel(x))\
.reshape(x.shape\
+ x.shape)
adolc.independent(x) y = f(x) adolc.dependent(y) adolc.trace_off() # point at which the derivatives should be evaluated x = random((N,D)) print '\n\n' print 'Sympy function = function check (should be almost zero)' print f(x) - sym_f(x) print '\n\n' print 'Sympy vs Hand Derived Gradient check (should be almost zero)' print df(x) - sym_df(x) print 'Sympy vs Ad Derived Gradient check (should be almost zero)' print adolc.gradient(0, numpy.ravel(x)).reshape(x.shape) - sym_df(x) print '\n\n' print 'Sympy vs Hand Derived Hessian check (should be almost zero)' print ddf(x) - sym_ddf(x) print 'Sympy vs Ad Derive Hessian check (should be almost zero)' print adolc.hessian(0, numpy.ravel(x)).reshape(x.shape + x.shape) - sym_ddf(x)
def hessianA_taped(self, XP):
return adolc.hessian(self.adolcID, XP)
adolc.independent(x)
y = f(x)
adolc.dependent(y)
adolc.trace_off()
# point at which the derivatives should be evaluated
x = random((N,D))
print('\n\n')
print('Sympy function = function check (should be almost zero)')
print(f(x) - sym_f(x))
print('\n\n')
print('Sympy vs Hand Derived Gradient check (should be almost zero)')
print(df(x) - sym_df(x))
print('Sympy vs Ad Derived Gradient check (should be almost zero)')
print(adolc.gradient(0, numpy.ravel(x)).reshape(x.shape) - sym_df(x))
print('\n\n')
print('Sympy vs Hand Derived Hessian check (should be almost zero)')
print(ddf(x) - sym_ddf(x))
print('Sympy vs Ad Derive Hessian check (should be almost zero)')
print(adolc.hessian(0, numpy.ravel(x)).reshape(x.shape + x.shape) - sym_ddf(x))
def hessian(self, x):
return adolc.hessian(0,x)
trace_on(0)
for n in range(N):
for m in range(N):
independent(aA[n,m])
aC = inv(aA)[0]
ay = trace(aC)
dependent(ay)
trace_off()
adolc_num_locations.append(tapestats(0)['NUM_LOCATIONS'])
adolc_num_operations.append(tapestats(0)['NUM_OPERATIONS'])
t_end = time()
adolc_taping_times.append(t_end-t_start)
t_start = time()
H1 = adolc.hessian(0,A.ravel())
t_end = time()
print('adolc needs %0.6f seconds'%(t_end - t_start))
adolc_times.append(t_end-t_start)
# with ALGOPY
t_start = time()
cg = CGraph()
FA = zeros((2,N**2,N,N))
for n in range(N):
for m in range(N):
FA[0,n*N+m,:,:] = A[:,:]
FA[1,n*N+m,n,m] = 1.
def dense_hess(self, x, z, **kwargs):
"Return the Hessian of the objective at x in dense format."
return adolc.hessian(self._obj_trace_id, x)
def hessian(self, x):
return adolc.hessian(0, x)
trace_on(0)
for n in range(N):
for m in range(N):
independent(aA[n, m])
aC = inv(aA)[0]
ay = trace(aC)
dependent(ay)
trace_off()
adolc_num_locations.append(tapestats(0)['NUM_LOCATIONS'])
adolc_num_operations.append(tapestats(0)['NUM_OPERATIONS'])
t_end = time()
adolc_taping_times.append(t_end - t_start)
t_start = time()
H1 = adolc.hessian(0, A.ravel())
t_end = time()
print('adolc needs %0.6f seconds' % (t_end - t_start))
adolc_times.append(t_end - t_start)
# with ALGOPY
t_start = time()
cg = CGraph()
FA = zeros((2, N**2, N, N))
for n in range(N):
for m in range(N):
FA[0, n * N + m, :, :] = A[:, :]
FA[1, n * N + m, n, m] = 1.
