Revision 1109 trunk/src/df1b2separable/df1b2imp.cpp
df1b2imp.cpp (revision 1109)  

2  2 
* $Id$ 
3  3 
* 
4  4 
* Author: David Fournier 
5 
* Copyright (c) 20082012 Regents of the University of California


5 
* Copyright (c) 20082012 Regents of the University of California 

6  6 
*/ 
7  7 
/** 
8  8 
* \file 
...  ...  
29  29 
gradient_structure::set_YES_DERIVATIVES(); 
30  30 
int nvar=x.size()+u0.size()+u0.size()*u0.size(); 
31  31 
independent_variables y(1,nvar); 
32 


32  
33  33 
// need to set random effects active together with whatever 
34  34 
// init parameters should be active in this phase 
35 
initial_params::set_inactive_only_random_effects();


36 
initial_params::set_active_random_effects();


37 
/*int onvar=*/initial_params::nvarcalc();


35 
initial_params::set_inactive_only_random_effects(); 

36 
initial_params::set_active_random_effects(); 

37 
/*int onvar=*/initial_params::nvarcalc(); 

38  38 
initial_params::xinit(y); // get the initial values into the 
39  39 
// do we need this next line? 
40  40 
y(1,xs)=x; 
...  ...  
55  55 
for (j=1;j<=us;j++) 
56  56 
y(ii++)=Hess(i,j); 
57  57  
58 
dvar_vector vy=dvar_vector(y);


58 
dvar_vector vy=dvar_vector(y); 

59  59 
initial_params::stddev_vscale(d,vy); 
60  60 
dvar_matrix vHess(1,us,1,us); 
61  61 
ii=xs+us+1; 
...  ...  
85  85 
for (int is=1;is<=nsamp;is++) 
86  86 
{ 
87  87 
dvar_vector tau=ch*pmin>lapprox>epsilon(is); 
88 


88  
89  89 
vy(xs+1,xs+us).shift(1)+=tau; 
90  90 
initial_params::reset(vy); // get the values into the model 
91  91 
vy(xs+1,xs+us).shift(1)=tau; 
...  ...  
112  112  
113  113  
114  114 
dvariable min_vf=min(sample_value); 
115 
vf=min_vflog(mean(exp(min_vfsample_value)));


116 
vf=us*0.91893853320467241;


117 


115 
vf=min_vflog(mean(exp(min_vfsample_value))); 

116 
vf=us*0.91893853320467241; 

117  
118  118 
int sgn=0; 
119  119 
dvariable ld; 
120  120 
if (ad_comm::no_ln_det_choleski_flag) 
...  ...  
132  132 
vy(xs+1,xs+us).shift(1)=u0; 
133  133 
initial_params::reset(vy); // get the values into the model 
134  134 
gradient_structure::set_YES_DERIVATIVES(); 
135 


135  
136  136 
ii=1; 
137  137 
for (i=1;i<=xs;i++) 
138  138 
xadjoint(i)=g(ii++); 
Also available in: Unified diff