[Help-gsl] Siman problem copying best_values

[Alessandro Nascimento]

Hi GSL folks,

I am trying a SA optimization with this code:

/**********************************************************************************************
double Optimizer::Simulated_Annealing(Mol2* Lig){
gsl_rng* r;
r = gsl_rng_alloc(gsl_rng_ranlxs2);
srand(time(NULL));
gsl_rng_set(r, (rand() % 100));

gsl_siman_params_t params = { 200, 200, 1.0, 1.0, 0.001, 1.1, 0.00001};

double f_minimum = evaluate_energy(Lig, Lig->xyz);
printf("Evaluated energy: %7.3f ------> %7.3f\n", f_minimum,
Lig->xyz[0][0]);

gsl_siman_solve(r, Lig, siman_energy, siman_new_step, siman_change,
siman_print_data, NULL, NULL, NULL, sizeof(Mol2), params);

f_minimum = evaluate_energy(Lig, Lig->xyz);
printf("Evaluated energy: %7.3f ------> %7.3f\n", f_minimum,
Lig->xyz[0][0]);

***************************************************************************************************/

The energy function also calls my function "evaluate_energy" cpp function:

/**********************************************************************************************
double Optimizer::siman_energy(void *data){
double f;
Mol2* Lig2 = (Mol2*) data;
f = evaluate_energy(Lig2, Lig2->xyz);
return (f);
}
***************************************************************************************************/

The funny thing is that I can NOT get the best energies achievied with SA
after gsl_siman_solve.In contrary, I have something like:


Evaluated energy:   7.540 ------>  13.196
#-iter  #-evals   temperature     position   energy
    0       101         0.001  11.508      -7.16879      -7.16879
    1       201   0.000909091   6.181       -37.311       -37.311
    2       301   0.000826446   9.292      -85.9492      -85.9492
    3       401   0.000751315   0.529      -85.9492      -85.9492
    4       501   0.000683013   2.821      -85.9492      -85.9492
    5       601   0.000620921   0.771      -85.9492      -85.9492
    6       701   0.000564474   1.260      -85.9492      -85.9492
    7       801   0.000513158 -12.741      -85.9492      -85.9492
    8       901   0.000466507  -9.173      -85.9492      -85.9492
    9      1001   0.000424098 -15.450      -85.9492      -85.9492
   10      1101   0.000385543  -8.908      -85.9492      -85.9492
   11      1201   0.000350494 -21.362      -85.9492      -85.9492
   12      1301   0.000318631 -21.844      -85.9492      -85.9492
   13      1401   0.000289664 -20.422      -85.9492      -85.9492
   14      1501   0.000263331 -20.524      -85.9492      -85.9492
   15      1601   0.000239392 -21.456      -85.9492      -85.9492
   16      1701   0.000217629 -29.786      -85.9492      -85.9492
   17      1801   0.000197845 -28.451      -85.9492      -85.9492
   18      1901   0.000179859 -34.339      -85.9492      -85.9492
   19      2001   0.000163508 -34.233      -85.9492      -85.9492
   20      2101   0.000148644 -35.589      -85.9492      -85.9492
   21      2201   0.000135131 -24.893      -85.9492      -85.9492
   22      2301   0.000122846 -28.700      -85.9492      -85.9492
   23      2401   0.000111678 -22.464      -85.9492      -85.9492
   24      2501   0.000101526 -23.738      -85.9492      -85.9492
   25      2601    9.2296e-05 -27.228      -85.9492      -85.9492
   26      2701   8.39055e-05 -34.404      -85.9492      -85.9492
   27      2801   7.62777e-05 -26.407      -85.9492      -85.9492
   28      2901   6.93433e-05 -17.783      -85.9492      -85.9492
   29      3001   6.30394e-05 -19.237      -85.9492      -85.9492
   30      3101   5.73086e-05 -33.685      -85.9492      -85.9492
   31      3201   5.20987e-05 -21.991      -85.9492      -85.9492
   32      3301   4.73624e-05 -28.721      -85.9492      -85.9492
   33      3401   4.30568e-05 -37.596      -85.9492      -85.9492
   34      3501   3.91425e-05 -31.584      -85.9492      -85.9492
   35      3601   3.55841e-05 -39.479      -85.9492      -85.9492
   36      3701   3.23492e-05 -36.385      -85.9492      -85.9492
   37      3801   2.94083e-05 -40.311      -85.9492      -85.9492
   38      3901   2.67349e-05 -32.598      -85.9492      -85.9492
   39      4001   2.43044e-05 -39.861      -85.9492      -85.9492
   40      4101   2.20949e-05 -40.343      -85.9492      -85.9492
   41      4201   2.00863e-05 -41.349      -85.9492      -85.9492
   42      4301   1.82603e-05 -36.646      -85.9492      -85.9492
   43      4401   1.66002e-05 -26.142      -85.9492      -85.9492
   44      4501   1.50911e-05 -28.222      -85.9492      -85.9492
   45      4601   1.37192e-05 -21.963      -85.9492      -85.9492
   46      4701    1.2472e-05 -32.856      -85.9492      -85.9492
   47      4801   1.13382e-05 -36.224      -85.9492      -85.9492
   48      4901   1.03074e-05 -39.059      -85.9492      -85.9492
Evaluated energy:   0.881 ------> -39.059

Looks like the SA algorithm is working as expected but my data is not being
copied back to *x0_p, as it should.

Any clues?

PS: The entire code can be visualized in
http://code.google.com/p/libela/source/browse/trunk/iMcLiBELa/src/Optimizer.cpp

Thanks in advance,

--nascimento