sam::ResearchStrategy¶
Abstract class for different research strategies. More...
#include <ResearchStrategy.h>
Inherited by sam::DefaultResearchStrategy
Public Functions¶
| Name | |
|---|---|
| virtual | ~ResearchStrategy() =0 |
| ResearchStrategy() | |
| virtual bool | willStartHacking(std::optional< SubmissionPool > & subs) Indicates whether the researcher will start going to the hacking procedure. |
| virtual bool | willContinueHacking(Experiment * experiment, PolicyChain & pchain) Indicates whether the researcher will continue the hacking procedure. |
| virtual bool | willContinueHacking(std::optional< SubmissionPool > & sub, PolicyChain & pchain) |
| bool | willBeSubmitting(const std::optional< SubmissionPool > & sub, PolicyChain & pchain) |
| virtual bool | willContinueReplicating(SubmissionPool & subs) Decides whether the researcher is going to continue the replication process or not. |
| void | reset() Resets the internal state of the research strategy. |
| std::optional< SubmissionPool > | stashedSubmissions() Returns a copy of the stashed_submissions. |
| virtual std::optional< SubmissionPool > | selectOutcomeFromExperiment(Experiment * experiment, PolicyChainSet & pchain_set) =0 Returns an optional result of applying the PolicyChainSet on the given Experiment. |
| virtual std::optional< SubmissionPool > | selectOutcomeFromPool(SubmissionPool & spool, PolicyChainSet & pchain_set) =0 Returns an optional result of applying the PolicyChainSet on the given list of submissions. |
| void | saveOutcomes(Experiment & experiment, PolicyChain & pchain) |
| std::optional< SubmissionPool > | selectOutcome(Experiment & experiment, PolicyChainSet & pchain_set) Returns the result of applying the PolicyChainSet on the Experiment. |
| std::optional< SubmissionPool > | selectBetweenSubmissions(SubmissionPool & spool, PolicyChainSet & pchain_set) Returns the result of applying the PolicyChainSet on the submission pool. |
| std::unique_ptr< ResearchStrategy > | build(json & research_strategy_config) ResearchStrategy factory method. |
Public Attributes¶
Protected Attributes¶
| Name | |
|---|---|
| std::optional< SubmissionPool > | submission_candidates |
| SubmissionPool | stashed_submissions |
Detailed Description¶
class sam::ResearchStrategy;
Abstract class for different research strategies.
Note: This is being implemented as abstract class due to historical reasons. There is only one research strategy at the moment, but I'm considering the possibility of adding more and maybe redesigning this again.
Public Functions Documentation¶
function ~ResearchStrategy¶
virtual ~ResearchStrategy() =0
Pure destructors of ResearchStrategy. This is necessary for proper deconstruction of derived classes.
function ResearchStrategy¶
ResearchStrategy()
Besides constructing the base class, it also registers the Submission and DependentVariable in Lua for later use of the derived classes, e.g., DefaultResearchStrategy
function willStartHacking¶
inline virtual bool willStartHacking(
std::optional< SubmissionPool > & subs
)
Indicates whether the researcher will start going to the hacking procedure.
Parameters:
- subs The subs
Return: Returns true if researcher is going to start the hacking procedure
Reimplemented by: sam::DefaultResearchStrategy::willStartHacking
The default here is to not go for hacking if we already have one candidate; but this can be overridden in different research strategies
function willContinueHacking¶
inline virtual bool willContinueHacking(
Experiment * experiment,
PolicyChain & pchain
)
Indicates whether the researcher will continue the hacking procedure.
Parameters:
- experiment A reference to the experiment
Reimplemented by: sam::DefaultResearchStrategy::willContinueHacking
This will be used by hackTheResearch and uses the decision policy to decide whether the next hacking strategy is going to be executed !
function willContinueHacking¶
inline virtual bool willContinueHacking(
std::optional< SubmissionPool > & sub,
PolicyChain & pchain
)
Reimplemented by: sam::DefaultResearchStrategy::willContinueHacking
function willBeSubmitting¶
bool willBeSubmitting(
const std::optional< SubmissionPool > & sub,
PolicyChain & pchain
)
Parameters:
- subs A list of Submission(s)
- pchain A policy chain
Return: returns true if submission should be submitted to the Journal.
Indicates whether the Researcher is going to submit the given set of submissions or not.
Usually submission_decision_policies will be used to evaluate the quality of the submission pool. This checks if at least one of the submissions satisfy all of the given policies.
function willContinueReplicating¶
inline virtual bool willContinueReplicating(
SubmissionPool & subs
)
Decides whether the researcher is going to continue the replication process or not.
Parameters:
- subs A list of submissions
Return: Returns true if the replication procedure should continue
Reimplemented by: sam::DefaultResearchStrategy::willContinueReplicating
function reset¶
inline void reset()
Resets the internal state of the research strategy.
function stashedSubmissions¶
inline std::optional< SubmissionPool > stashedSubmissions()
Returns a copy of the stashed_submissions.
Return: Returns an optional containing a copy of stashed_submissions
Todo: this can be improved
function selectOutcomeFromExperiment¶
virtual std::optional< SubmissionPool > selectOutcomeFromExperiment(
Experiment * experiment,
PolicyChainSet & pchain_set
) =0
Returns an optional result of applying the PolicyChainSet on the given Experiment.
Parameters:
- experiment The experiment
- pchain_set The policy chain set
Return: Returns a list of submissions, if any.
Reimplemented by: sam::DefaultResearchStrategy::selectOutcomeFromExperiment
function selectOutcomeFromPool¶
virtual std::optional< SubmissionPool > selectOutcomeFromPool(
SubmissionPool & spool,
PolicyChainSet & pchain_set
) =0
Returns an optional result of applying the PolicyChainSet on the given list of submissions.
Parameters:
- spool The submission pool
- pchain_set The policy chain set
Return: Returns a list of submissions, if any.
Reimplemented by: sam::DefaultResearchStrategy::selectOutcomeFromPool
function saveOutcomes¶
void saveOutcomes(
Experiment & experiment,
PolicyChain & pchain
)
Parameters:
- experiment a reference to the experiment
- pchain a policy chain, usually #stashing_policy
- experiment a reference to the experiment
- pchain a policy chain, usually stored in
stashing_policyin the config file
Creates and save all possible submissions from an experiment, if they pass the given policy predicate
Create and save all possible submissions from an experiment, if the satisfy all of the given policies in the pchain.
function selectOutcome¶
std::optional< SubmissionPool > selectOutcome(
Experiment & experiment,
PolicyChainSet & pchain_set
)
Returns the result of applying the PolicyChainSet on the Experiment.
Parameters:
- experiment The experiment
- pchain_set The policy chain set.
- experiment a reference to an experiment
- pchain_set a reference to a policy chain set
Return:
- Returns a list of submissions, if any.
- A list of submissions, if any.
Select an unique outcome from an experiment, if at some point, a PolicyChain finds a group of outcomes instead of a unique outcome, the selection will be saved and the will await for processing in a different stages.
We check all available PolicyChains in the given chain set, and will stop as soon as any chain returns something!
function selectBetweenSubmissions¶
std::optional< SubmissionPool > selectBetweenSubmissions(
SubmissionPool & spool,
PolicyChainSet & pchain_set
)
Returns the result of applying the PolicyChainSet on the submission pool.
Parameters:
- spool The spool
- pchain_set The policy chain set
- spool a collection of submissions collected in previous stages, e.g., selectOutcome
- pchain_set a set of policy chains
Return:
- Returns a list of submissions, if any.
- A list of submissions, if any.
Note: If submission_candidate is empty to this point, and we cannot find anything here, we'll continue with nothing, and the current experiment will most likely be discarded.
Select a unique submission from the given pool of submissions. If none of the submissions satisfies all the policies, we just return, and submission_candidate will not be rewritten.
If any of the pchains return something, we ignore the rest, and leave!
function build¶
static std::unique_ptr< ResearchStrategy > build(
json & research_strategy_config
)
ResearchStrategy factory method.
Parameters:
- research_strategy_config A JSON object containing information about each research strategy.
Return: Returns a unique pointer to the newly build ResearchStrategy.
Public Attributes Documentation¶
variable lua¶
sol::state lua;
variable initial_selection_policies¶
PolicyChainSet initial_selection_policies;
Todo: These guys should move to their own class, I don't have to keep everything here!
variable submission_decision_policies¶
PolicyChain submission_decision_policies;
variable between_stashed_selection_policies¶
PolicyChainSet between_stashed_selection_policies;
variable between_reps_policies¶
PolicyChainSet between_reps_policies;
variable will_not_start_hacking_decision_policies¶
PolicyChain will_not_start_hacking_decision_policies;
variable will_not_continue_replicating_decision_policy¶
PolicyChain will_not_continue_replicating_decision_policy;
variable stashing_policy¶
PolicyChain stashing_policy;
Protected Attributes Documentation¶
variable submission_candidates¶
std::optional< SubmissionPool > submission_candidates;
variable stashed_submissions¶
SubmissionPool stashed_submissions;
List of selected submissions collected by the researcher using the stashing_policy
Updated on 29 June 2021 at 16:13:46 CEST