Пример #1
0
    X = orth(X)
    return np.dot(X,np.dot(uu,X))
    
    
data = {}
init_params = {}


#init_params = 0.1*( ones(len(UU)) + 2.*R.randn(len(UU)) )
#init_params = UU

iii = 1
data  = [{ 'STA':STA[i] , 'STC':STC[i] } for i in range(iii)]
init_params = [{'theta':STA[i] * 0.05 , 'M':STC[iii] * 0.05} for i in range(iii)]

term = kolia_theano.term(init_params=init_params[0],differentiate=['f'],
                          f=quadratic_Poisson, barrier=barrier, ldet=ldet, eigs=eigs)

terms = [deepcopy(term) for i in range(iii)]

def syM(x):
    xx = x['M']
    x['M'] = (xx+np.transpose(xx))/2.
    return x

def symm(X):
    X = [ syM(x) for x in objective.inflate(X)]
    return objective.flatten(X)

def callback_one(ip,d):
    pp = objective.inflate(ip)
    print
Пример #2
0
quad = lambda x : 0.2*(x+1.)**2

(N_spikes,STA,STC), U, V1, bbar, Cb , STAB = LNLNP(T=10000,sigma=sigma,NL=quad,N=27)
Nsub, N     =  U.shape
NRGC, Nsub  =  V1.shape


iii = NRGC
data  = [{ 'STA':STA[i] , \
           'STC':STC[i] } for i in range(iii)]
init_params = [{'theta':data[i]['STA'] * 0.1 , \
                'M':data[i]['STC'] * 0.1} for i in range(iii)]


term = kolia_theano.term(init_params=init_params[0],differentiate=['f'],
                          f=quadratic_Poisson, barrier=eig_barrier)

class Objective: pass
objective = Objective()

def f(params,args):
    return np.sum([term.f(params[i*N*(N+1):(i+1)*N*(N+1)],args[i]) for i in range(iii)],0)

def barrier(params,args):
    return np.sum([term.barrier(params[i*N*(N+1):(i+1)*N*(N+1)],args[i]) for i in range(iii)],0)

def df(params,args):
    return np.concatenate([term.df(params[i*N*(N+1):(i+1)*N*(N+1)],args[i]) for i in range(iii)])

objective.f       = f
objective.df      = df