1 2 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
|
// Copyright 2008 Michiaki Hamada
// Adapted from public domain code by Yi-Kuo Yu, NCBI
/**
* See lambda_calculator.h
*/
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include "nrutil.h"
int Alphsize;
#include "nrutil.cpp"
#include "ludcmp.cpp"
#include "lubksb.cpp"
#include "lambda_calculator.h"
#define Epsilon 1.0e-36
#define E_bound 1.0e-12
#define Infty 1000000.0
#define min(A, B) ((A) > (B) ? (B) : (A) )
#define max(A, B) ((A) > (B) ? (A) : (B) )
#define bool int
#define true 1
#define false 0
double Lambda_UB; //lambda upper bound
double r_max_m, c_max_m; //min of each row's (column's) max
void makematrix(const double **mat_b, double **a, double lambda);
typedef struct Lambda {
double min;
double max;
int flag; // 1 means there is a range, -1 means no solution possible.
} Lambda;
typedef struct Sum {
double value;
int flag; // 1 means no negative bg_freq, -1 means there is negative bg_freq
} Sum;
Lambda Find_JP(const double **mat_b, double la_min, double la_max, double **JP, double *p_in, double *q_in);
Sum Check_root(const double **mat_b, double **a, double lambda, double *p, double *q);
double Check_det(const double **mat_b, double **a, double lambda);
Sum Nail_lambda(const double **mat_b, int flag_sign, double lambda_min, double lambda_max, double *p, double *q,
double *la_add);
double Nail_det(const double **mat_b, int flag_sign, double lambda_min, double lambda_max);
bool Check_range(const double **mat_b);
double *Locate_det_zero(const double **mat_b, int *); //pointer to root list are returned with how many of them by int
double calculate_lambda(const double **mat_b, int alpha_size,
double *p, double *q) {
double **JP/*, *q, *p*/;
int k;
double *root_location;
int N_root;
Lambda Lambda_local;
Alphsize = alpha_size;
if (!Check_range(mat_b)) return -1.0;
root_location = Locate_det_zero(mat_b, &N_root);
if (root_location == NULL && N_root > 0) return -1.0;
//q=dvector(1,Alphsize);
//p=dvector(1,Alphsize);
JP = dmatrix(1, Alphsize, 1, Alphsize);
if (N_root == 0) {
Lambda_local = Find_JP(mat_b, 0, Lambda_UB, JP, p, q);
if (1 == Lambda_local.flag) { // sensible solution found
// Remember to find the right place to free the vectors
//free_dvector(p, 1,Alphsize);
//free_dvector(q, 1,Alphsize);
free(root_location);
free_dmatrix(JP, 1, Alphsize, 1, Alphsize);
return (Lambda_local.min + Lambda_local.max) / 2.0;
} else if (-1 == Lambda_local.flag) {
//printf("matrix pass first screening but no sensible solution found. :-( \n");
}
} else if (N_root > 0) {
//printf("N_root = %d for this matirx \n", N_root);
//for (i=0;i<N_root;i++) printf("N_root[%d] = %lf\n",i,root_location[i]);
for (k = 0; k <= N_root; k++) {
if (k == 0) {
Lambda_local.min = 0;
Lambda_local.max = root_location[0];
}
else if (k == N_root) {
Lambda_local.min = root_location[N_root - 1];
Lambda_local.max = Lambda_UB + Epsilon;
}
else {
Lambda_local.min = root_location[k - 1];
Lambda_local.max = root_location[k];
}
Lambda_local = Find_JP(mat_b, Lambda_local.min, Lambda_local.max, JP, p, q);
if (1 == Lambda_local.flag) { // sensible solution found
//free_dvector(p, 1,Alphsize);
//free_dvector(q, 1,Alphsize);
free(root_location);
free_dmatrix(JP, 1, Alphsize, 1, Alphsize);
return (Lambda_local.min + Lambda_local.max) / 2.0;
} else if (-1 == Lambda_local.flag) {
//printf("matrix pass first screening but still no sensible solution found. :-( \n");
}
}
}
// Remember to find the right place to free the vectors
//free_dvector(p, 1,Alphsize);
//free_dvector(q, 1,Alphsize);
free(root_location);
free_dmatrix(JP, 1, Alphsize, 1, Alphsize);
return -1.0;
}
bool Check_range(const double **mat_b) {
int pos_flag_r, neg_flag_r;
int pos_flag_c, neg_flag_c;
double r_max, c_max; //max of each row (or column)
int L_r = 0, L_c = 0; // number of zero-score rows (columns)
// First make sure each row and column have both pos and neg entries
r_max_m = c_max_m = 100000000000.0;
for (int i = 1; i <= Alphsize; i++) {
r_max = 0;
c_max = 0;
pos_flag_r = -1;
neg_flag_r = -1;
pos_flag_c = -1;
neg_flag_c = -1;
for (int j = 1; j <= Alphsize; j++) {
if (mat_b[i][j] > 0) {
if (mat_b[i][j] > r_max) r_max = mat_b[i][j];
pos_flag_r = 1;
} else if (mat_b[i][j] < 0) neg_flag_r = 1;
if (mat_b[j][i] > 0) {
if (mat_b[j][i] > c_max) c_max = mat_b[j][i];
pos_flag_c = 1;
} else if (mat_b[j][i] < 0) neg_flag_c = 1;
}
if ((pos_flag_r == -1) || (neg_flag_r == -1) || (pos_flag_c == -1) || (neg_flag_c == -1)) {
if ((pos_flag_r == -1) && (neg_flag_r == -1)) {
printf("only zero score at row %d\n", i);
L_r++;
} else if ((pos_flag_c == -1) && (neg_flag_c == -1)) {
printf("only zero score at column %d\n", i);
L_c++;
} else {
//printf("all positive or all negative at row or column %d\n", i);
//printf("therefore invalid matrix. exit now. \n");
return false;
//exit(1);
}
}
if ((r_max < r_max_m) && (r_max > 0)) r_max_m = r_max;
if ((c_max < c_max_m) && (c_max > 0)) c_max_m = c_max;
}
// Find the upper bound for lambda
if (r_max_m > c_max_m) {
Lambda_UB = 1.1 * log(1.0 * Alphsize - L_r) / r_max_m;
} else {
Lambda_UB = 1.1 * log(1.0 * Alphsize - L_c) / c_max_m;
}
//printf("the upper bound for lambda is %lf\n", Lambda_UB);
return true;
}
double Check_det(const double **mat_b, double **a, double lambda) {
double d;
int i, /*j,*/ *indx;
indx = ivector(1, Alphsize);
makematrix(mat_b, a, lambda);
ludcmp(a, Alphsize, indx, &d);
for (i = 1; i <= Alphsize; i++) d *= a[i][i];
free_ivector(indx, 1, Alphsize);
return d; //returning the determinant
}
Sum Check_root(const double **mat_b, double **a, double lambda, double *p, double *q) {
double **y, /* *col,*/ d;
//double sum = 0.0;
int i, j;//, *indx;
Sum Sum_here;
y = dmatrix(1, Alphsize, 1, Alphsize);
//indx = ivector(1,Alphsize);
int indx[Alphsize + 1];
//col = dvector(1,Alphsize);
double col[Alphsize + 1];
makematrix(mat_b, a, lambda);
ludcmp(a, Alphsize, indx, &d);
Sum_here.value = 0.0;
for (i = 1; i <= Alphsize; i++) q[i] = 0.0;
for (j = 1; j <= Alphsize; j++) {
for (i = 1; i <= Alphsize; i++){
col[i] = 0.0;
}
col[j] = 1.0;
lubksb(a, Alphsize, indx, col);
p[j] = 0.0;
for (i = 1; i <= Alphsize; i++) {
y[i][j] = col[i];
Sum_here.value += y[i][j];
p[j] += y[i][j];
q[i] += y[i][j];
}
}
Sum_here.flag = 1;
for (i = 1; i < Alphsize; i++) {
if ((p[i] < 0) || (q[i] < 0)) {
Sum_here.flag = -1;
//printf("problematic freq. p[%d] = %.4f q[%d]=%.4f\n",i,p[i],i,q[i]);
}
}
free_dmatrix(y, 1, Alphsize, 1, Alphsize);
return Sum_here;
}
double *Locate_det_zero(const double **mat_b, int *N_root_add) {
double **a/*, *q, *p */; // a is the exponentiated matrix of socres, p and q are bg_freqs
int i/*,j,k*/;
int N; // number of points for first round
int flag_sign;
double lambda/*, l_tmp, sum, sum_min, sum_max */;
double lambda_root, dlambda /*, dsum=0.5 */;
//double *l_here, *s_here;
double root[5000];
double *root_temp;
//double error=0.000000000001;
int zero_monitor = 0; // record number of zeros found in the range
//int flag;
a = dmatrix(1, Alphsize, 1, Alphsize);
//Sum_local = (Sum *)malloc(sizeof(Sum));
//Lambda_local = (Lambda *)malloc(sizeof(Lambda));
N = 2 + max(400, ((int) (Lambda_UB - 0) / 0.005));
//printf("N = %d in Locate_det_zero\n", N);
dlambda = (Lambda_UB) / (N * 1.0);
//l_here = (double *)malloc((N+1)*sizeof(double));
//s_here = (double *)malloc((N+1)*sizeof(double));
double l_here[N + 1];
double s_here[N + 1];
for (i = 0; i < N; i++) {
lambda = (i + 1) * dlambda;
s_here[i] = Check_det(mat_b, a, lambda);
l_here[i] = lambda;
}
if (s_here[0] < 0.0) flag_sign = -1;
if (s_here[0] > 0.0) flag_sign = 1;
if (fabs(s_here[0]) / exp(l_here[0] * (r_max_m + c_max_m) / 2.0) <= Epsilon) {
root[zero_monitor++] = l_here[0];
flag_sign = 0;
}
for (i = 1; i < N; i++) {
if ((flag_sign != 0) && (fabs(s_here[i]) > Epsilon)) {
if (s_here[i - 1] * s_here[i] < 0) {
//printf("occurring at regular places\n");
lambda_root = Nail_det(mat_b, flag_sign, l_here[i - 1], l_here[i]);
root[zero_monitor++] = lambda_root;
flag_sign = -flag_sign; // the flag switch sign after one sol found
//printf("a (regular) root of det found at %12.10f, i= %d\n", lambda_root,i);
}
} else {
if (s_here[i] < 0.0) flag_sign = -1;
if (s_here[i] > 0.0) flag_sign = 1;
if (fabs(s_here[i]) / exp(l_here[i] * (r_max_m + c_max_m) / 2.0) <= Epsilon) {
root[zero_monitor++] = l_here[i];
}
}
}
//printf("total number of solution found in range is %d\n", i_monitor);
root_temp = (double *) malloc(zero_monitor * sizeof(double));
*N_root_add = zero_monitor;
if (zero_monitor > 0) {
if (zero_monitor >= N / 4) {
//printf("It is likely that uniform zero determinant is occurring.\n");
//printf("number of small det points = %d out of %d, exit now....\n",zero_monitor, N);
free(root_temp);
return NULL;
//exit(1);
}
for (i = 0; i < zero_monitor; i++) {
root_temp[i] = root[i];
//printf("root_location[%d] = %lf\n",i,root_temp[i]);
}
}
free_dmatrix(a, 1, Alphsize, 1, Alphsize);
return root_temp;
}
Lambda Find_JP(const double **mat_b, double la_min, double la_max, double **JP, double *p_in, double *q_in) {
double **a, *q, *p; // a is the exponentiated matrix of socres, p and q are bg_freqs
int i, j/*,k*/;
int N; // number of points for first round
double lambda/*, l_tmp, sum, sum_min, sum_max*/;
double lambda_max, lambda_min, lambda_final, dlambda/*, dsum=0.5*/;
//double *l_here, *s_here;
//double error=0.000000000000001;
//int validity_flag; // 1 means valid, -1 means not valid.
int flag_sign; // 1 means small lambda sum > 1, -1 means otherwise
int flag_done = -1; // 1 means find sensible solution, -1 means sensible not found
int i_monitor = 0; // record number of solution found in the range, including nonsense ones
int j_monitor;
Lambda Lambda_local;
//Sum *Sum_local;
Sum Sum_local;
lambda_min = la_min;
lambda_max = la_max;
q = q_in;
p = p_in;
a = dmatrix(1, Alphsize, 1, Alphsize);
//Sum_local = (Sum *)malloc(sizeof(Sum));
//Lambda_local = (Lambda *)malloc(sizeof(Lambda));
N = 2 + max(400, ((int) (lambda_max - lambda_min) / 0.005));
//printf("N = %d in Find_JP\n", N);
dlambda = (lambda_max - lambda_min) / (N * 1.0);
//l_here = (double *)malloc((N+1)*sizeof(double));
//s_here = (double *)malloc((N+1)*sizeof(double));
double l_here[N + 1];
double s_here[N + 1];
//printf("lambda_min enter = %12.10e, lambda_max = %12.10f\n", lambda_min, lambda_max);
for (i = 0; i < N - 1; i++) {
lambda = lambda_min + (i + 1) * dlambda;
makematrix(mat_b, a, lambda);
Sum_local = Check_root(mat_b, a, lambda, p, q);
l_here[i] = lambda;
s_here[i] = Sum_local.value - 1.0;
//printf("scan %d th time in Find_JP\n",i );
}
//printf("finish first time scanining in Find_JP\n");
if (s_here[0] < 0.0) flag_sign = -1;
else if (s_here[0] > 0.0) flag_sign = 1;
else if (s_here[0] == 0.0) { //needs refined definition on flag_sign
printf("enter the exact hit, rarely occurs other than when lambda = 0 \n");
j_monitor = 1;
flag_sign = 0;
while ((flag_sign == 0) && (j_monitor < N)) {
Sum_local = Check_root(mat_b, a, l_here[0] + j_monitor * dlambda / N, p, q);
if (Sum_local.value > 1.0) {
flag_sign = 1;
} else if (Sum_local.value < 1.0) {
flag_sign = -1;
}
j_monitor++;
}
}
for (i = 1; i < N; i++) { // should be N-1 ???
if (flag_sign == 0) {
printf("flag_sign = 0 \n");
exit(1);
}
if (s_here[i - 1] * s_here[i] < 0) {
lambda_min = l_here[i - 1];
lambda_max = l_here[i];
Sum_local = Nail_lambda(mat_b, flag_sign, lambda_min, lambda_max, p, q, &lambda_final);
if (Sum_local.flag == 1) {
i = N;
flag_done = 1;
Lambda_local.flag = 1;
Lambda_local.min = lambda_final, Lambda_local.max = lambda_final;
}
flag_sign = -flag_sign; // the flag switch sign after one sol found
i_monitor++;
}
}
if (flag_done == 1) {
// Write correct JP to the matrix
makematrix(mat_b, a, lambda_final);
for (i = 1; i <= Alphsize; i++) {
for (j = 1; j <= Alphsize; j++) {
JP[i][j] = a[i][j] * p[i] * q[j];
}
}
free_dmatrix(a, 1, Alphsize, 1, Alphsize);
return Lambda_local;
} else if (flag_done == -1) {
//printf("no sensible solution in the plausible x range: (%lf,%lf)\n", la_min, la_max);
Lambda_local.flag = -1;
Lambda_local.min = 0;
Lambda_local.max = Infty;
return Lambda_local;
}
// never come here
return Lambda_local;
}
Sum Nail_lambda(const double **mat_b, int flag_sign, double lambda_min, double lambda_max, double *p, double *q,
double *lam_add) {
double **a;
double lambda;
//Sum *Sum_local;
Sum Sum_local;
a = dmatrix(1, Alphsize, 1, Alphsize);
//Sum_local = (Sum *)malloc(sizeof(Sum));
lambda = (lambda_min + lambda_max) / 2.0;
Sum_local = Check_root(mat_b, a, lambda, p, q);
while (fabs(Sum_local.value - 1.0) > E_bound) {
if (flag_sign * (Sum_local.value - 1.0) < 0) lambda_max = lambda;
else if (flag_sign * (Sum_local.value - 1.0) > 0) lambda_min = lambda;
// Added by MCF to avoid infinite loop:
if (lambda == (lambda_min + lambda_max) / 2.0) {
Sum_local.flag = -1;
break;
}
lambda = (lambda_min + lambda_max) / 2.0;
Sum_local = Check_root(mat_b, a, lambda, p, q);
}
free_dmatrix(a, 1, Alphsize, 1, Alphsize);
*lam_add = lambda;
return Sum_local;
}
double Nail_det(const double **mat_b, int flag_sign, double lambda_min, double lambda_max) {
double **a;
double lambda;
double value;
a = dmatrix(1, Alphsize, 1, Alphsize);
lambda = (lambda_min + lambda_max) / 2.0;
value = Check_det(mat_b, a, lambda);
while ((fabs(value) > E_bound) && (lambda > 0)) {
if (flag_sign * (value) < 0) lambda_max = lambda;
else if (flag_sign * (value) > 0) lambda_min = lambda;
lambda = (lambda_min + lambda_max) / 2.0;
value = Check_det(mat_b, a, lambda);
}
free_dmatrix(a, 1, Alphsize, 1, Alphsize);
return lambda;
}
void makematrix(const double **mat_b, double **a, double lambda) {
int i, j;
for (i = 1; i <= Alphsize; i++)
for (j = 1; j <= Alphsize; j++) {
*(*(a + i) + j) = exp(lambda * mat_b[i][j]);
}
}
|