*mcrng = gmx_rng_init(ir->expandedvals->lmc_seed);
if (!bStateFromCP)
{
- init_df_history_weights(dfhist,ir->expandedvals,ir->fepvals->n_lambda);
+ init_df_history_weights(dfhist, ir->expandedvals, ir->fepvals->n_lambda);
}
}
return bIfFlat;
}
-static gmx_bool CheckIfDoneEquilibrating(int nlim, t_expanded *expand, df_history_t *dfhist, gmx_large_int_t step)
+static gmx_bool CheckIfDoneEquilibrating(int nlim, t_expanded *expand, df_history_t *dfhist, gmx_int64_t step)
{
int i, totalsamples;
}
static gmx_bool UpdateWeights(int nlim, t_expanded *expand, df_history_t *dfhist,
- int fep_state, real *scaled_lamee, real *weighted_lamee, gmx_large_int_t step)
+ int fep_state, real *scaled_lamee, real *weighted_lamee, gmx_int64_t step)
{
real maxdiff = 0.000000001;
gmx_bool bSufficientSamples;
real clam_varm, clam_varp, clam_weightsm, clam_weightsp, clam_minvar;
real *lam_weights, *lam_minvar_corr, *lam_variance, *lam_dg;
double *p_k;
- double pks = 0;
+ double pks = 0;
real *numweighted_lamee, *logfrac;
int *nonzero;
real chi_m1_0, chi_p1_0, chi_m2_0, chi_p2_0, chi_p1_m1, chi_p2_m1, chi_m1_p1, chi_m2_p1;
if (lamnew > maxfep)
{
/* it's possible some rounding is failing */
- if (gmx_within_tol(remainder[fep_state],0,50*GMX_DOUBLE_EPS))
+ if (gmx_within_tol(remainder[fep_state], 0, 50*GMX_DOUBLE_EPS))
{
/* numerical rounding error -- no state other than the original has weight */
lamnew = fep_state;
/* print out the weights to the log, along with current state */
extern void PrintFreeEnergyInfoToFile(FILE *outfile, t_lambda *fep, t_expanded *expand, t_simtemp *simtemp, df_history_t *dfhist,
- int fep_state, int frequency, gmx_large_int_t step)
+ int fep_state, int frequency, gmx_int64_t step)
{
int nlim, i, ifep, jfep;
real dw, dg, dv, dm, Tprint;
extern int ExpandedEnsembleDynamics(FILE *log, t_inputrec *ir, gmx_enerdata_t *enerd,
t_state *state, t_extmass *MassQ, int fep_state, df_history_t *dfhist,
- gmx_large_int_t step, gmx_rng_t mcrng,
+ gmx_int64_t step, gmx_rng_t mcrng,
rvec *v, t_mdatoms *mdatoms)
/* Note that the state variable is only needed for simulated tempering, not
Hamiltonian expanded ensemble. May be able to remove it after integrator refactoring. */