public interface TunableEngine extends UCIEngine
HASH_OPTION_NAME, OWN_BOOK_OPTION_NAME, START_POSITION, THREADS_OPTION_NAME
Modifier and Type | Method and Description |
---|---|
double |
eval(Map<String,Double> gradientCache)
Statically evaluates the current position and records the gradient of the evaluation function w.r.t.
|
EngineParameters |
getParameters()
Returns the parameters which are used by the engine to control search, evaluation, and other aspects of game play.
|
default boolean |
isGradientDefined()
Specifies whether the gradient of the evaluation function is mathematically defined.
|
void |
notifyParametersChanged()
Notifies the engine that the parameters have changed and that if it uses cached values, it should reload them.
|
void |
setDeterministicEvaluationMode(boolean on)
Sets whether the engine should support deterministic static evaluation without the use of hash tables or any other mechanisms that
introduce non-determinism.
|
close, getAuthor, getDebugInfo, getHashLoadPermill, getName, getOptions, getSearchInfo, init, isInit, newGame, play, ponderHit, quit, search, setDebugMode, setHashSizeOption, setOption, setOption, setOwnBookOption, setPosition, setPosition, setThreadsOption, stop
EngineParameters getParameters()
void notifyParametersChanged()
void setDeterministicEvaluationMode(boolean on)
on
- Whether the engine should support deterministic evaluation mode.double eval(Map<String,Double> gradientCache)
gradientCache
- A map for recording the partial derivatives of the objective evaluation function (where a positive output means
white is in the lead and negative output means black is in the lead) with respect to the static evaluation parameters. The key
should be the name of the parameter and the value should be the derivative. It can be null if the calculation of the gradient is not
needed or if isGradientDefined()
returns false.default boolean isGradientDefined()
gradientCache
parameter of the eval(Map)
method.Copyright © 2020. All rights reserved.