Trial

class Trial

A trial contains a number of TrialStages. The Acceptance is computed as the stages are enacted, and then sent to Criteria to decide if the trial is accepted or rejected.

Subclassed by feasst::TrialAVB2Half, feasst::TrialAVB4, feasst::TrialAdd, feasst::TrialAddAVB, feasst::TrialAddAVBDivalent, feasst::TrialAddMultiple, feasst::TrialBeta, feasst::TrialFactory, feasst::TrialGibbsParticleTransferOneWay, feasst::TrialGibbsVolumeTransfer, feasst::TrialGrowLinear, feasst::TrialModel, feasst::TrialMorph, feasst::TrialMorphExpanded, feasst::TrialMove, feasst::TrialRemove, feasst::TrialRemoveAVB, feasst::TrialRemoveAVBDivalent, feasst::TrialRemoveMultiple, feasst::TrialRotateCluster, feasst::TrialSwapSites, feasst::TrialTranslateCluster, feasst::TrialVolume

Public Functions

double weight() const: Return the unnormalized relative probability of selection of this trial with respect to all trials.

double weight_per_number_fraction() const: Return the weight per number of particles of given type.

std::vector<int> number_fraction_exclude_type() const: Return the particle types excluded in number fraction weights.

void add_stage(std::shared_ptr<TrialSelect> select, std::shared_ptr<Perturb> perturb, argtype *stage_args): Add a stage which includes selection and perturbation with arguments.

void add_stage(std::shared_ptr<TrialSelect> select, std::shared_ptr<Perturb> perturb): Same as above, but without arguments for stage.

void add_stage(std::shared_ptr<TrialStage> stage): Same as above, but by copying an existing stage.

void set(const int index, std::shared_ptr<TrialStage> stage): Set a stage, as can be done just before each attempt.

int num_stages() const: Number of stages.

const TrialStage &stage(const int index) const: Return a stage.

int64_t num_success() const: Number of successful attempts.

int64_t num_attempts() const: Number of attempts.

int64_t num_auto_reject() const: Number of automatic rejections.

void increment_num_attempts(): Increment the number of attempts for acceptance.

double acceptance() const: Return the ratio of the number of successful attempts and total attempts. Return -1 if no attempts were performed.

virtual void reset_stats(): Reset trial statistics.

virtual std::string status_header() const: Return the header description for the status of the trial (e.g., acceptance, etc).

std::string name_or_description() const: Return the class name (or description if name is “Trial”)

virtual std::string status() const: Return the status of the trial (e.g., acceptance, etc).

virtual void tune(): Tune parameters.

virtual void precompute(Criteria *criteria, System *system): Precompute quantities before simulation for optimization.

void set(std::shared_ptr<TrialCompute> compute): Set the computation of the trial and acceptance.

const TrialCompute &compute() const: Return TrialCompute.

void revert(System *system, Criteria *criteria): Revert all stages in reverse order.

virtual bool attempt(Criteria *criteria, System *system, Random *random): Attempt a trial. Return true if accepted.

const std::string &description() const: Return the description, as used in Log.

void set_description(const std::string &description): Set the description.

Arguments

Trial(argtype args = argtype())

args:

weight: unnormalized relative probability of selection of this trial with respect to all trials (default: 1).
weight_per_number_fraction: if > 0, the weight is continuously updated to (weight_per_number_fraction * number of TrialSelect::particle_type in first TrialStage / total number of particles). If <= 0, the given weight above is fixed to that value (default: -1). For a multicomponent simulation, weight_per_number_fraction with a separate trial for each mobile particle_type ensures equal probability of selecting any particles of those types regardless of the number of particles of each type, while still allowing for different tunable parameters for each particle_type.
number_fraction_exclude_type[i]: if >= 0 (default: -1), exclude this particle type from the total number in fractions of weight_per_number_fraction. If there is a fixed particle type in a multicomponent grand canonical ensemble, avoid changing the relative weight of translation vs insert/delete and breaking detailed balance. The “[i]” is to be substituted for an integer 0, 1, 2,… If only one excluded type, then the “[i]” is optional.
print_num_accepted: if true (default: false), print the number of accepted trials.