lin_stat_gen_detailed_data.m

function detailed_data = lin_stat_gen_detailed_data(model,model_data)
%function detailed_data = lin_stat_gen_detailed_data(model,model_data)
%
% function computing the high-dimensional parameter independent 
% data required for rb-treatment of a fem-problem, comprises all
% expensive computations

% B. Haasdonk 22.2.2011

detailed_data = model_data;

% change mode of model to components:
old_mode = model.decomp_mode;
model.decomp_mode = 1; % == components

% the following generates cell arrays of components
[A_comp,f_comp] = model.operators(model,model_data);
Q_A = length(A_comp);
Q_f = length(f_comp);
H = length(f_comp{1}); % number of dofs
if model.compute_output_functional
  l_comp = model.operators_output(model,detailed_data);
  Q_l = length(l_comp);
end;

% transform cell arrays to matrices for efficiency
detailed_data.f_comp = zeros(H,Q_f);
for q = 1:Q_f
  detailed_data.f_comp(:,q) = f_comp{q};
end;
if model.compute_output_functional
  detailed_data.l_comp = zeros(H,Q_l);
  for q = 1:Q_l
    detailed_data.l_comp(:,q) = l_comp{q};
  end;
end;
% detect overall sparsity pattern: 
Asum = A_comp{1};
for q = 2:Q_A
  Asum = Asum + A_comp{q};
  ind = find(Asum);
  siz = size(Asum);
  [i,j] = ind2sub(siz,ind);
  detailed_data.A_size = siz; 
  detailed_data.A_row_indices = i; 
  detailed_data.A_column_indices = j; 
end;
for q = 1:Q_A
  Aq = A_comp{q};
  detailed_data.A_comp(:,q) = Aq(ind);
end;

% generate reduced basis
detailed_data = rb_basis_generation(model,detailed_data);

% set back old mode
model.decomp_mode = old_mode;