| 12
 3
 4
 5
 6
 7
 8
 9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
 100
 101
 102
 103
 104
 105
 106
 107
 108
 109
 110
 111
 112
 113
 114
 115
 116
 117
 118
 119
 120
 121
 122
 123
 124
 125
 126
 127
 128
 129
 130
 131
 132
 133
 134
 135
 136
 137
 138
 139
 140
 141
 142
 143
 144
 145
 146
 147
 148
 149
 150
 151
 152
 153
 154
 155
 156
 157
 158
 159
 160
 161
 162
 163
 164
 165
 166
 167
 168
 169
 170
 171
 172
 173
 174
 175
 176
 177
 178
 179
 180
 181
 182
 183
 184
 185
 186
 187
 188
 189
 190
 191
 192
 193
 194
 195
 196
 197
 198
 199
 200
 201
 202
 203
 204
 205
 206
 207
 208
 209
 210
 211
 212
 213
 214
 215
 216
 217
 218
 219
 220
 221
 222
 223
 224
 225
 226
 227
 228
 229
 230
 231
 232
 233
 234
 235
 236
 237
 238
 239
 240
 241
 242
 243
 244
 245
 246
 247
 248
 249
 250
 251
 252
 253
 254
 255
 256
 257
 258
 259
 260
 261
 262
 263
 264
 265
 266
 267
 268
 269
 270
 271
 272
 273
 274
 275
 276
 277
 278
 279
 
 | % STK_SAMPLING_OLHS generates a random Orthogonal Latin Hypercube (OLH) sample
%
% CALL: X = stk_sampling_olhs (N)
%
%    generates a random Orthogonal Latin Hypercube (OLH) sample X, using the
%    construction of Ye (1998). The algorithm only works for sample sizes N
%    of the form 2^(R+1)+1, with R >= 1. Trying to generate an OLHS with a
%    value of N that is not of this form generates an error. The number of
%    factors is D = 2*R, and the OLHS is defined on [-1; 1]^D.
%
% CALL: X = stk_sampling_olhs (N, D)
%
%    does exactly the same thing, provided that there exists an integer R
%    such that N = 2^(R+1)+1 and D = 2*R (or D is empty).
%
% CALL: X = stk_sampling_olhs (N, D, BOX)
%
%    generates an OLHS on BOX. Again, D can be empty since the number of
%    factors can be deduced from N.
%
% CALL: X = stk_sampling_olhs (N, D, BOX, PERMUT)
%
%    uses a given permutation PERMUT, instead of a random permutation, to
%    initialize the construction of Ye (1998). As a result, the generated
%    OLHS is not random anymore. PERMUT must be a permutation of 1:2^R. If
%    BOX is empty, then the default domain [-1, 1]^D is used.
%
% NOTE: orthogonality
%
%    The samples generated by this functions are only orthogonal, stricty-
%    speaking, if BOX is a symmetric domain (e.g., [-1, 1] ^ D). Otherwise,
%    the generated samples should be called "uncorrelated".
%
% REFERENCE
%
%    Kenny Q. Ye, "Orthogonal Column Latin Hypercubes and Their
%    Application in Computer Experiments", Journal of the American
%    Statistical Association, 93(444), 1430-1439, 1998.
%    http://dx.doi.org/10.1080/01621459.1998.10473803
%
% See also: stk_sampling_randomlhs, stk_sampling_maximinlhs
% Copyright Notice
%
%    Copyright (C) 2017 CentraleSupelec
%    Copyright (C) 2012-2014 SUPELEC
%
%    Author:  Julien Bect  <julien.bect@centralesupelec.fr>
% Copying Permission Statement
%
%    This file is part of
%
%            STK: a Small (Matlab/Octave) Toolbox for Kriging
%               (http://sourceforge.net/projects/kriging)
%
%    STK is free software: you can redistribute it and/or modify it under
%    the terms of the GNU General Public License as published by the Free
%    Software Foundation,  either version 3  of the License, or  (at your
%    option) any later version.
%
%    STK is distributed  in the hope that it will  be useful, but WITHOUT
%    ANY WARRANTY;  without even the implied  warranty of MERCHANTABILITY
%    or FITNESS  FOR A  PARTICULAR PURPOSE.  See  the GNU  General Public
%    License for more details.
%
%    You should  have received a copy  of the GNU  General Public License
%    along with STK.  If not, see <http://www.gnu.org/licenses/>.
function [x, aux] = stk_sampling_olhs (n, d, box, permut, extended)
if nargin > 5,
    stk_error ('Too many input arguments.', 'TooManyInputArgs');
end
% Read argument dim
if (nargin < 2) || ((nargin < 3) && (isempty (d)))
    d = 1;  % Default dimension
elseif (nargin > 2) && (~ isempty (box))
    d = size (box, 2);
end
if nargin < 5, extended = false; end
%%% PROCESS INPUT ARGUMENTS %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% recover the "order" r from the value of n
r = floor(log2(n - 1) - 1);
n_ = 2^(r + 1) + 1;
if (r <= 0) || (abs(n - n_) > eps),
    errmsg = 'n must be an integer of the form 2^(r+1) + 1 with r > 0';
    stk_error(errmsg, 'InvalidArgument');
end
n = n_;
if ~extended
    % check that d has the correct value for a "full" Ye98-OLHS
    % (other values of d can be reached by removing columns from a full OLHS)
    if (nargin > 1) && ~isempty(d),
        if d ~= 2 * r,
            errmsg = 'Incorrect value of d, please read the documentation...';
            stk_error(errmsg, 'InvalidArgument');
        end
    else
        d = 2 * r;
    end
else
    % check that d has the correct value for a "full" Cioppa-Lucas(2007) NOLHS
    % (other values of d can be reached by removing columns from a full NOLHS)
    if (nargin > 1) && ~isempty(d),
        if d ~= r + 1 + nchoosek(r, 2),
            errmsg = 'Incorrect value of d, please read the documentation...';
            stk_error(errmsg, 'InvalidArgument');
        end
    else
        d = r + 1 + nchoosek(r, 2);
    end
end
% number of "positive levels"
q = 2^r; % = (n - 1)/2
% read argument 'box'
if (nargin < 3) || isempty (box)
    box = stk_hrect (repmat ([-1; 1], 1, d));  % build a default box
else
    box = stk_hrect (box);  % convert input argument to a proper box
end
% permutation
if (nargin < 4) || isempty(permut),
    permut = randperm(q)';
else
    permut = permut(:);
    if ~isequal(sort(permut), (1:q)'),
        errmsg = sprintf('permut should be a permutation of 1:%d.', q);
        stk_error(errmsg, 'InvalidArgument');
    end
end
%%% CONSTRUCT AN "ORTHOGONAL" LHS FOLLOWING PROCESS INPUT ARGUMENTS %%%%%%%%%%%
% Construct permutation matrices A1, A2, ..., Ar
A = cell(1, r);
for i = 1:r,
    Ai = 1;
    for j = 1:i,
        Z  = zeros(size(Ai));
        Ai = [Z Ai; Ai Z]; %#ok<AGROW>
    end
    for j = (i+1):r,
        Z  = zeros(size(Ai));
        Ai = [Ai Z; Z Ai]; %#ok<AGROW>
    end
    A{i} = Ai;
end
% Construct the matrix M
M = [permut zeros(q, 2*r-1)];
for j = 1:r, % from column 2 to column r+1
    M(:, j+1) = A{j} * permut;
end
if ~extended, % OLHS / Ye (1998)
    for j = 1:(r-1), % from column r+2 to column 2*r
        M(:, j+r+1) = A{j} * A{r} * permut;
    end
else % NOLHS / Cioppa & Lucas(2007)
    col = r + 2;
    for j = 1:(r-1),
        for k = (j + 1):r,
            M(:, col) = A{j} * A{k} * permut;
            col = col + 1;
        end
    end
end
% Construct the matrix S
S = ones(q, 2*r);
for j = 1:r, % from column 2 to column r+1
    aj = 1;
    for l = r:(-1):1,
        if l == r - j + 1,
            aj = [-aj; aj]; %#ok<AGROW>
        else
            aj = [aj; aj]; %#ok<AGROW>
        end
    end
    S(:, j+1) = aj;
end
if ~extended, % OLHS / Ye (1998)
    for j = 1:(r-1), % from column r+2 to column 2*r
        S(:, r+1+j) = S(:, 2) .* S(:, j+2);
    end
else % NOLHS / Cioppa & Lucas(2007)
    col = r + 2;
    for j = 1:(r - 1),
        for k = (j + 1):r,
            S(:, col) = S(:, j+1) .* S(:, k+1);
            col = col + 1;
        end
    end
end
% Construct the matrix T
T = M .* S;
% Construct the OLHS X (with integer levels -q, ..., 0, ... +q)
x_integer_levels = [T; zeros(1, d); -T];
%%% CONVERT TO THE REQUESTED BOX %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Convert to positive integer levels (1 ... n)
x = x_integer_levels + q + 1;
% Convert to [0; 1]-valued levels
x = (2*x - 1) / (2*n);
% And, finally, convert to box
x = stk_dataframe (stk_rescale (x, [], box), box.colnames);
% Note: the results reported in Cioppa & Lucas correspond to the scaling
%   x = struct('a', stk_rescale(x, [min(x); max(x)], box));
%%% OUTPUT SOME AUXILIARY DATA IF REQUESTED %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if nargout > 1,
    aux = struct('M', M, 'S', S, 'X', x_integer_levels);
end
end % function
%%
% Check error for incorrect number of input arguments
%!shared x, n, d, box, permut
%! n = 5; d = 2; box = [0 0; 1, 1]; permut = 1:2;
%!error x = stk_sampling_olhs();
%!test  x = stk_sampling_olhs(n);
%!test  x = stk_sampling_olhs(n, d);
%!test  x = stk_sampling_olhs(n, d, box);
%!test  x = stk_sampling_olhs(n, d, box, permut);
%!error x = stk_sampling_olhs(n, d, box, permut, pi);
%%
% Check that the output is a dataframe
% (all stk_sampling_* functions should behave similarly in this respect)
%!assert (isa (x, 'stk_dataframe'));
%%
% Check that column names are properly set, if available in box
%!assert (isequal (x.colnames, {}));
%!test
%! cn = {'W', 'H'};  box = stk_hrect (box, cn);
%! x = stk_sampling_olhs (n, d, box);
%! assert (isequal (x.colnames, cn));
%%
% Check output argument
%!test
%! for r = 1:5
%!
%!   n = 2 ^ (r + 1) + 1;  d = 2 * r;
%!   x = stk_sampling_olhs (n, d);
%!
%!   assert (isequal (size (x), [n d]));
%!
%!   box = repmat ([-1; 1], 1, d);
%!   assert (stk_is_lhs (x, n, d, box));
%!
%!   x = double (x);  w = x' * x;
%!   assert (stk_isequal_tolabs (w / w(1,1), eye (d)));
%!
%! end
 |