Package daikon.inv.ternary.threeScalar
Class LinearTernaryFloat
- Object
-
- Invariant
-
- TernaryInvariant
-
- ThreeFloat
-
- LinearTernaryFloat
-
- All Implemented Interfaces:
Serializable
,Cloneable
public class LinearTernaryFloat extends ThreeFloat
Represents a Linear invariant over three double scalarsx
,y
, andz
, of the formax + by + cz + d = 0
. The constantsa
,b
,c
, andd
are mutually relatively prime, and the constanta
is always positive.- See Also:
- Serialized Form
-
-
Nested Class Summary
-
Nested classes/interfaces inherited from class Invariant
Invariant.ClassVarnameComparator, Invariant.ClassVarnameFormulaComparator, Invariant.InvariantComparatorForPrinting, Invariant.Match
-
-
Field Summary
Fields Modifier and Type Field Description LinearTernaryCoreFloat
core
static boolean
debugLinearTernary
static boolean
dkconfig_enabled
Boolean.-
Fields inherited from class Invariant
CONFIDENCE_JUSTIFIED, CONFIDENCE_NEVER, CONFIDENCE_UNJUSTIFIED, debug, debugFlow, debugGuarding, debugIsObvious, debugIsWorthPrinting, debugPrint, debugPrintEquality, dkconfig_confidence_limit, dkconfig_fuzzy_ratio, dkconfig_simplify_define_predicates, falsified, invariantEnabledDefault, isGuardingPredicate, min_mod_non_missing_samples, ppt, PROBABILITY_JUSTIFIED, PROBABILITY_NEVER, PROBABILITY_UNJUSTIFIED
-
-
Constructor Summary
Constructors Modifier Constructor Description protected
LinearTernaryFloat()
protected
LinearTernaryFloat(PptSlice ppt)
-
Method Summary
All Methods Static Methods Instance Methods Concrete Methods Modifier and Type Method Description InvariantStatus
add_modified(double x, double y, double z, int count)
Similar toThreeFloat.check_modified(double, double, double, int)
except that it can change the state of the invariant if necessary.InvariantStatus
check_modified(double x, double y, double z, int count)
Presents a sample to the invariant.LinearTernaryFloat
clone()
Do nothing special, Overridden to remove exception from declaration.protected double
computeConfidence()
This method computes the confidence that this invariant occurred by chance.boolean
enabled()
returns whether or not this invariant is enabledboolean
enoughSamples()
Returns true if the invariant has enough samples to have its computed constants well-formed.static @Nullable LinearTernaryFloat
find(PptSlice ppt)
static List<LinearTernaryFloat>
findAll(VarInfo vi)
String
format_using(OutputFormat format)
Return a printed representation of this invariant, in the given format.static LinearTernaryFloat
get_proto()
Returns the prototype invariant for LinearTernaryFloatLinearTernaryFloat
instantiate_dyn(PptSlice slice)
Instantiate the invariant on the specified slice.boolean
instantiate_ok(VarInfo[] vis)
LinearTernary is only valid on non-constant integral types.boolean
isActive()
Returns whether or not the invariant is currently active.boolean
isExact()
Subclasses should override.boolean
isExclusiveFormula(Invariant other)
Returns true iff the two invariants represent mutually exclusive mathematical formulas -- that is, if one of them is true, then the other must be false.@Nullable DiscardInfo
isObviousDynamically(VarInfo[] vis)
Return non-null if this invariant is necessarily true from a fact that can be determined dynamically (after checking data) -- for the given varInfos rather than the varInfos of this.boolean
isSameFormula(Invariant other)
Returns true iff the two invariants represent the same mathematical formula.@Nullable Invariant
merge(List<Invariant> invs, PptSlice parent_ppt)
Merge the invariants in invs to form a new invariant.boolean
mergeFormulasOk()
Returns whether or not it is possible to merge invariants of the same class but with different formulas when combining invariants from lower ppts to build invariants at upper program points.String
repr()
For printing invariants, there are two interfaces: repr gives a low-level representation (Invariant.repr_prob()
also prints the confidence), andInvariant.format()
gives a high-level representation for user output.protected Invariant
resurrect_done(int[] permutation)
Called on the new invariant just before resurrect() returns it to allow subclasses to fix any information they might have cached from the old Ppt and VarInfos.InvariantStatus
setup(LinearBinaryFloat lb, VarInfo con_var, double con_val)
InvariantStatus
setup(OneOfFloat oo, VarInfo v1, double con1, VarInfo v2, double con2)
-
Methods inherited from class ThreeFloat
add, add_unmodified, check, check_unmodified, valid_types, var1, var2, var3
-
Methods inherited from class Invariant
add_sample, asInvClass, checkRep, clear_falsified, clone_and_permute, conf_is_ge, confidence_and, confidence_and, confidence_or, createGuardedInvariant, createGuardingPredicate, falsify, find, format, format_classname, format_too_few_samples, format_unimplemented, formatFuzzy, get_comparability, get_ni_suppressions, getConfidence, getGuardingList, getGuardingList, hasUninterestingConstant, instantiate, is_false, is_ni_suppressed, isAllPrestate, isEqualityComparison, isObvious, isObviousDynamically, isObviousDynamically_SomeInEquality, isObviousDynamically_SomeInEqualityHelper, isObviousStatically, isObviousStatically, isObviousStatically_AllInEquality, isObviousStatically_SomeInEquality, isObviousStatically_SomeInEqualityHelper, isReflexive, isSameInvariant, isValidEscExpression, isValidExpression, isWorthPrinting, justified, log, log, logDetail, logOn, match, permute, prob_and, prob_and, prob_is_ge, prob_or, repCheck, repr_prob, resurrect, simplify_format_double, simplify_format_long, simplify_format_string, state_match, toString, toString, transfer, usesVar, usesVar, usesVarDerived, varNames
-
-
-
-
Field Detail
-
dkconfig_enabled
public static boolean dkconfig_enabled
Boolean. True iff LinearTernary invariants should be considered.
-
debugLinearTernary
public static final boolean debugLinearTernary
- See Also:
- Constant Field Values
-
core
@Unused(when=Prototype.class) public LinearTernaryCoreFloat core
-
-
Constructor Detail
-
LinearTernaryFloat
protected LinearTernaryFloat(PptSlice ppt)
-
LinearTernaryFloat
protected LinearTernaryFloat()
-
-
Method Detail
-
get_proto
public static LinearTernaryFloat get_proto()
Returns the prototype invariant for LinearTernaryFloat
-
enabled
public boolean enabled()
returns whether or not this invariant is enabled
-
instantiate_ok
public boolean instantiate_ok(VarInfo[] vis)
LinearTernary is only valid on non-constant integral types.- Overrides:
instantiate_ok
in classInvariant
- See Also:
Invariant.valid_types(VarInfo[])
-
instantiate_dyn
public LinearTernaryFloat instantiate_dyn( LinearTernaryFloat this, PptSlice slice)
Instantiate the invariant on the specified slice.- Specified by:
instantiate_dyn
in classInvariant
- Returns:
- the new invariant
-
clone
@SideEffectFree public LinearTernaryFloat clone(@GuardSatisfied LinearTernaryFloat this)
Description copied from class:Invariant
Do nothing special, Overridden to remove exception from declaration.
-
resurrect_done
protected Invariant resurrect_done(int[] permutation)
Description copied from class:Invariant
Called on the new invariant just before resurrect() returns it to allow subclasses to fix any information they might have cached from the old Ppt and VarInfos.- Overrides:
resurrect_done
in classTernaryInvariant
- Parameters:
permutation
- the permutation- Returns:
- the permuted invariant
-
repr
public String repr(@GuardSatisfied LinearTernaryFloat this)
Description copied from class:Invariant
For printing invariants, there are two interfaces: repr gives a low-level representation (Invariant.repr_prob()
also prints the confidence), andInvariant.format()
gives a high-level representation for user output.
-
format_using
@SideEffectFree public String format_using(@GuardSatisfied LinearTernaryFloat this, OutputFormat format)
Description copied from class:Invariant
Return a printed representation of this invariant, in the given format.- Specified by:
format_using
in classInvariant
-
isActive
@Pure public boolean isActive()
Description copied from class:Invariant
Returns whether or not the invariant is currently active. This is used to identify those invariants that require a certain number of points before they actually do computation (eg, LinearBinary)This is used during suppresion. Any invariant that is not active cannot suppress another invariant.
-
setup
public InvariantStatus setup(LinearBinaryFloat lb, VarInfo con_var, double con_val)
-
setup
public InvariantStatus setup(OneOfFloat oo, VarInfo v1, double con1, VarInfo v2, double con2)
-
check_modified
public InvariantStatus check_modified(double x, double y, double z, int count)
Description copied from class:ThreeFloat
Presents a sample to the invariant. Returns whether the sample is consistent with the invariant. Does not change the state of the invariant.- Specified by:
check_modified
in classThreeFloat
count
- how many identical samples were observed in a row. For example, three calls to check_modified with a count parameter of 1 is equivalent to one call to check_modified with a count parameter of 3.- Returns:
- whether or not the sample is consistent with the invariant
-
add_modified
public InvariantStatus add_modified(double x, double y, double z, int count)
Description copied from class:ThreeFloat
Similar toThreeFloat.check_modified(double, double, double, int)
except that it can change the state of the invariant if necessary. If the invariant doesn't have any state, then the implementation should simply callThreeFloat.check_modified(double, double, double, int)
. This method need not check for falsification; that is done by the caller.- Specified by:
add_modified
in classThreeFloat
-
enoughSamples
public boolean enoughSamples(@GuardSatisfied LinearTernaryFloat this)
Description copied from class:Invariant
Returns true if the invariant has enough samples to have its computed constants well-formed. Is overridden in classes like LinearBinary/Ternary and Upper/LowerBound.- Overrides:
enoughSamples
in classInvariant
- Returns:
- true if the invariant has enough samples to have its computed constants well-formed
-
computeConfidence
protected double computeConfidence()
Description copied from class:Invariant
This method computes the confidence that this invariant occurred by chance. Clients should callInvariant.getConfidence()
instead.This method need not check the value of field "falsified", as the caller does that.
- Specified by:
computeConfidence
in classInvariant
- Returns:
- confidence of this invariant
- See Also:
Invariant.getConfidence()
-
isExact
@Pure public boolean isExact()
Description copied from class:Invariant
Subclasses should override. An exact invariant indicates that given all but one variable value, the last one can be computed. (I think that's correct, anyway.) Examples are IntComparison (when only equality is possible), LinearBinary, FunctionUnary. OneOf is treated differently, as an interface. The result of this method does not depend on whether the invariant is justified, destroyed, etc.
-
isObviousDynamically
@Pure public @Nullable DiscardInfo isObviousDynamically(VarInfo[] vis)
Description copied from class:Invariant
Return non-null if this invariant is necessarily true from a fact that can be determined dynamically (after checking data) -- for the given varInfos rather than the varInfos of this. Conceptually, this means, "Is this invariant dynamically obvious if its VarInfos were switched with vis?" Intended to be overriden by subclasses so they can filter invariants after checking; the overriding method should first call "super.isObviousDynamically(vis)". Since this method is dynamic, it should only be called after all processing.- Overrides:
isObviousDynamically
in classInvariant
-
isSameFormula
@Pure public boolean isSameFormula(Invariant other)
Description copied from class:Invariant
Returns true iff the two invariants represent the same mathematical formula. Does not consider the context such as variable names, confidences, sample counts, value counts, or related quantities. As a rule of thumb, if two invariants format the same, this method returns true. Furthermore, in many cases, if an invariant does not involve computed constants (as "x>c" and "y=ax+b" do for constants a, b, and c), then this method vacuously returns true.- Specified by:
isSameFormula
in classInvariant
- Parameters:
other
- the invariant to compare to this one- Returns:
- true iff the two invariants represent the same mathematical formula. Does not consider
-
isExclusiveFormula
@Pure public boolean isExclusiveFormula(Invariant other)
Description copied from class:Invariant
Returns true iff the two invariants represent mutually exclusive mathematical formulas -- that is, if one of them is true, then the other must be false. This method does not consider the context such as variable names, confidences, sample counts, value counts, or related quantities.- Overrides:
isExclusiveFormula
in classInvariant
- Parameters:
other
- the other invariant to compare to this one- Returns:
- true iff the two invariants represent mutually exclusive mathematical formulas
-
find
public static @Nullable LinearTernaryFloat find(PptSlice ppt)
-
findAll
public static List<LinearTernaryFloat> findAll(VarInfo vi)
-
mergeFormulasOk
public boolean mergeFormulasOk()
Description copied from class:Invariant
Returns whether or not it is possible to merge invariants of the same class but with different formulas when combining invariants from lower ppts to build invariants at upper program points. Invariants that have this characteristic (eg, bound, oneof) should override this function. Note that invariants that can do this, normally need special merge code as well (to merge the different formulas into a single formula at the upper point.- Overrides:
mergeFormulasOk
in classInvariant
- Returns:
- true if invariants with different formulas can be merged
-
merge
public @Nullable Invariant merge(List<Invariant> invs, PptSlice parent_ppt)
Merge the invariants in invs to form a new invariant. Each must be a LinearTernaryFloat invariant. The work is done by the LinearTernary core- Overrides:
merge
in classInvariant
- Parameters:
invs
- list of invariants to merge. They should all be permuted to match the variable order in parent_ppt.parent_ppt
- slice that will contain the new invariant- Returns:
- the merged invariant or null if the invariants didn't represent the same invariant
-
-