public void UsageBadVariableActionContext()
{
int numExceptionsCaught = 0;
int numExceptionsExpected = 5;
var tryCatchArgumentException = (Action <Action>)((action) => {
try
{
action();
}
catch (ArgumentException ex)
{
if (ex.ParamName.ToLower() == "ctx")
{
numExceptionsCaught++;
}
}
});
tryCatchArgumentException(() => {
var mwt = new MwtExplorer <TestContext>("test", new TestRecorder <TestContext>());
var policy = new TestPolicy <TestContext>();
var explorer = new EpsilonGreedyExplorer <TestContext>(policy, 0.2f);
mwt.ChooseAction(explorer, "key", new TestContext());
});
tryCatchArgumentException(() =>
{
var mwt = new MwtExplorer <TestContext>("test", new TestRecorder <TestContext>());
var policy = new TestPolicy <TestContext>();
var explorer = new TauFirstExplorer <TestContext>(policy, 10);
mwt.ChooseAction(explorer, "key", new TestContext());
});
tryCatchArgumentException(() =>
{
var mwt = new MwtExplorer <TestContext>("test", new TestRecorder <TestContext>());
var policies = new TestPolicy <TestContext> [2];
for (int i = 0; i < 2; i++)
{
policies[i] = new TestPolicy <TestContext>(i * 2);
}
var explorer = new BootstrapExplorer <TestContext>(policies);
mwt.ChooseAction(explorer, "key", new TestContext());
});
tryCatchArgumentException(() =>
{
var mwt = new MwtExplorer <TestContext>("test", new TestRecorder <TestContext>());
var scorer = new TestScorer <TestContext>(10);
var explorer = new SoftmaxExplorer <TestContext>(scorer, 0.5f);
mwt.ChooseAction(explorer, "key", new TestContext());
});
tryCatchArgumentException(() =>
{
var mwt = new MwtExplorer <TestContext>("test", new TestRecorder <TestContext>());
var scorer = new TestScorer <TestContext>(10);
var explorer = new GenericExplorer <TestContext>(scorer);
mwt.ChooseAction(explorer, "key", new TestContext());
});
Assert.AreEqual(numExceptionsExpected, numExceptionsCaught);
}
public void EndToEndSoftmax()
{
uint numActions = 10;
float lambda = 0.5f;
TestRecorder <SimpleContext> recorder = new TestRecorder <SimpleContext>();
TestScorer <SimpleContext> scorer = new TestScorer <SimpleContext>(numActions);
MwtExplorer <SimpleContext> mwtt = new MwtExplorer <SimpleContext>("mwt", recorder);
var explorer = new SoftmaxExplorer <SimpleContext>(scorer, lambda, numActions);
EndToEnd(mwtt, explorer, recorder);
}
public void Softmax()
{
uint numActions = 10;
float lambda = 0.5f;
uint numActionsCover = 100;
float C = 5;
TestRecorder <TestContext> recorder = new TestRecorder <TestContext>();
TestScorer <TestContext> scorer = new TestScorer <TestContext>(numActions);
MwtExplorer <TestContext> mwtt = new MwtExplorer <TestContext>("mwt", recorder);
var explorer = new SoftmaxExplorer <TestContext>(scorer, lambda, numActions);
uint numDecisions = (uint)(numActions * Math.Log(numActions * 1.0) + Math.Log(numActionsCover * 1.0 / numActions) * C * numActions);
uint[] actions = new uint[numActions];
Random rand = new Random();
for (uint i = 0; i < numDecisions; i++)
{
uint chosenAction = mwtt.ChooseAction(explorer, rand.NextDouble().ToString(), new TestContext()
{
Id = (int)i
});
actions[chosenAction - 1]++; // action id is one-based
}
for (uint i = 0; i < numActions; i++)
{
Assert.IsTrue(actions[i] > 0);
}
var interactions = recorder.GetAllInteractions();
Assert.AreEqual(numDecisions, (uint)interactions.Count);
for (int i = 0; i < numDecisions; i++)
{
Assert.AreEqual(i, interactions[i].Context.Id);
}
}
public void Softmax()
{
int numActions = 10;
float lambda = 0.5f;
int numActionsCover = 100;
float C = 5;
var scorer = new TestScorer <RegularTestContext>(1, numActions);
var explorer = new SoftmaxExplorer(lambda);
uint numDecisions = (uint)(numActions * Math.Log(numActions * 1.0) + Math.Log(numActionsCover * 1.0 / numActions) * C * numActions);
var contexts = new RegularTestContext[numDecisions];
for (int i = 0; i < numDecisions; i++)
{
contexts[i] = new RegularTestContext {
Id = i
};
}
SoftmaxWithContext(numActions, explorer, scorer, contexts);
}
public void SoftmaxFixedActionUsingVariableActionInterface()
{
int numActions = 10;
float lambda = 0.5f;
int numActionsCover = 100;
float C = 5;
var scorer = new TestScorer <VariableActionTestContext>(1, numActions);
var explorer = new SoftmaxExplorer(lambda);
int numDecisions = (int)(numActions * Math.Log(numActions * 1.0) + Math.Log(numActionsCover * 1.0 / numActions) * C * numActions);
var contexts = new VariableActionTestContext[numDecisions];
for (int i = 0; i < numDecisions; i++)
{
contexts[i] = new VariableActionTestContext(numActions)
{
Id = i
};
}
SoftmaxWithContext(numActions, explorer, scorer, contexts);
}
public void SoftmaxScores()
{
uint numActions = 10;
float lambda = 0.5f;
TestRecorder <TestContext> recorder = new TestRecorder <TestContext>();
TestScorer <TestContext> scorer = new TestScorer <TestContext>(numActions, uniform: false);
MwtExplorer <TestContext> mwtt = new MwtExplorer <TestContext>("mwt", recorder);
var explorer = new SoftmaxExplorer <TestContext>(scorer, lambda, numActions);
Random rand = new Random();
mwtt.ChooseAction(explorer, rand.NextDouble().ToString(), new TestContext()
{
Id = 100
});
mwtt.ChooseAction(explorer, rand.NextDouble().ToString(), new TestContext()
{
Id = 101
});
mwtt.ChooseAction(explorer, rand.NextDouble().ToString(), new TestContext()
{
Id = 102
});
var interactions = recorder.GetAllInteractions();
Assert.AreEqual(3, interactions.Count);
for (int i = 0; i < interactions.Count; i++)
{
// Scores are not equal therefore probabilities should not be uniform
Assert.AreNotEqual(interactions[i].Probability, 1.0f / numActions);
Assert.AreEqual(100 + i, interactions[i].Context.Id);
}
}
public void SoftmaxScores()
{
int numActions = 10;
float lambda = 0.5f;
var recorder = new TestRecorder <RegularTestContext>();
var scorer = new TestScorer <RegularTestContext>(1, numActions, uniform: false);
//var mwtt = new MwtExplorer<RegularTestContext>("mwt", recorder);
var explorer = new SoftmaxExplorer(lambda);
var mwtt = MwtExplorer.Create("mwt", numActions, recorder, explorer, scorer);
Random rand = new Random();
mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = 100
});
mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = 101
});
mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = 102
});
var interactions = recorder.GetAllInteractions();
Assert.AreEqual(3, interactions.Count);
for (int i = 0; i < interactions.Count; i++)
{
// Scores are not equal therefore probabilities should not be uniform
Assert.AreNotEqual(interactions[i].Probability, 1.0f / numActions);
Assert.AreEqual(100 + i, interactions[i].Context.Id);
}
// Verify that policy action is chosen all the time
RegularTestContext context = new RegularTestContext {
Id = 100
};
List <float> scores = scorer.MapContext(context).Value.ToList();
float maxScore = 0;
int highestScoreAction = 0;
for (int i = 0; i < scores.Count; i++)
{
if (maxScore < scores[i])
{
maxScore = scores[i];
highestScoreAction = i + 1;
}
}
explorer.EnableExplore(false);
for (int i = 0; i < 1000; i++)
{
int chosenAction = mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = (int)i
});
Assert.AreEqual(highestScoreAction, chosenAction);
}
}
static void ExploreSoftmax <TContext>
(
string appId,
int policyType,
JToken configPolicy,
float lambda,
int numActions,
string[] experimentalUnitIdList,
TContext[] contextList,
string outputFile
)
{
var recorder = new StringRecorder <TContext>();
bool isVariableActionContext = typeof(IVariableActionContext).IsAssignableFrom(typeof(TContext));
switch (policyType)
{
case 0: // fixed all-equal scorer
{
var scorerScore = configPolicy["Score"].Value <int>();
var scorer = new TestScorer <TContext>(scorerScore, numActions);
var explorer = new SoftmaxExplorer(lambda);
var mwt = isVariableActionContext ?
MwtExplorer.Create(appId, new VariableActionProvider <TContext>(), recorder, explorer, scorer) :
MwtExplorer.Create(appId, numActions, recorder, explorer, scorer);
for (int i = 0; i < experimentalUnitIdList.Length; i++)
{
int numActionsVariable = isVariableActionContext ? ((IVariableActionContext)contextList[i]).GetNumberOfActions() : int.MaxValue;
mwt.ChooseAction(experimentalUnitIdList[i], contextList[i]);
}
File.AppendAllText(outputFile, recorder.GetRecording());
break;
}
case 1: // integer-progression scorer
{
var scorerStartScore = configPolicy["Start"].Value <int>();
var scorer = new TestScorer <TContext>(scorerStartScore, numActions, uniform: false);
var explorer = new SoftmaxExplorer(lambda);
var mwt = isVariableActionContext ?
MwtExplorer.Create(appId, new VariableActionProvider <TContext>(), recorder, explorer, scorer) :
MwtExplorer.Create(appId, numActions, recorder, explorer, scorer);
for (int i = 0; i < experimentalUnitIdList.Length; i++)
{
int numActionsVariable = isVariableActionContext ? ((IVariableActionContext)contextList[i]).GetNumberOfActions() : int.MaxValue;
mwt.ChooseAction(experimentalUnitIdList[i], contextList[i]);
}
File.AppendAllText(outputFile, recorder.GetRecording());
break;
}
}
}