| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
This is a catch-all section for documented functions, methods and relations that haven’t been categorized yet into any of the previous sections. They are in random order and many of them will never be part of the official PowerLoom interface. So beware!
Return the element of array at position [row, column].
Return the element of array at position [row, column].
Set the element of array at position [row, column] to value and return the result.
Set the element of array at position [row, column] to value and return the result.
Add a query and score pair to the master list of testing examples
Add a query and score pair to the master list of training examples
Return an iterator generating all asserted members of the collection self.
If direct? is true, do not consider sub-collections. This is an iterator reimplementation
of asserted-collection-members (which see).
Allocate an iterator that generates all propositions that support that member is in collection.
Finding one such support is enough for the original member-of-collection? semantics (which see), however, we
generalized it to allow us to handle all-proofs situations where we want to find all supports. This function
generates the iterator which in each iteration produces the same member value but a potentially different
supporting proposition and truth value. We set it up this way to be in mostly equivalent to asserted-collection-members.
Return a set of all of the types that are asserted to be satisfied by self.
Return a set of instances that belong to the class type.
Return a list of lists of descriptions that are provably co-extensional.
Return a set of relations that immediately specialize relation. If removeEquivalents? (recommended), don’t include any relations equivalent to relation.
Return a set of relations that immediately subsume relation. If removeEquivalents? (recommended), don’t include any relations equivalent to relation.
Return a set of most specific types that are satisfied by self.
Return a list of all relations equivalent to relation. If reflexive?, include relation in the list.
Return a list of all definite (TRUE or FALSE) propositions attached to self.
This is a generalization of all-facts-of (which see). With n = 0
and only one instance this command behaves just like all-facts-of.
Otherwise, returns a cons list of all definite (TRUE or FALSE) propositions
that reference any of the instances listed in instanceRefs, plus if n >= 1
all propositions that reference any instances that are arguments of
propositions collected in the previous step, plus if n >= 2... and so on.
That is, if we only consider binary propositions, this can be viewed as
growing a graph with instances as its nodes and predicates as its arcs
starting from the set of seed instanceRefs to depth n-1.
Caution: with a fully connected KB and large enough n this could return
the whole knowledge base.
The returned propositions include those asserted to be true or false by
default, but it does not include propositions that are found to be
true only by running the query engine. Facts inferred to be true by
the forward chainer will be included. Hence, the returned list of
facts may be longer in a context where the forward chainer has been
run then in one where it has not (see run-forward-rules).
Iterate over all conceived propositions visible from module that have an inconsistent truth value. If local?, only return inconsistent propositions conceived locally in module.
Iterate over all instances (or individuals) visible from module. Only instances that haven’t been deleted will be considered. If local?, only return instances created locally in module.
Iterate over all conceived propositions visible from module that had their truth value locally modified (initialized, asserted, retracted). Note, that this will always include all locally conceived propositions.
Iterate over all named descriptions visible from module. If local?, return only named descriptions interned in module. If module is null, return all named descriptions interned everywhere.
Iterate over all named instances (or individuals) visible from module. Only instances that haven’t been deleted will be considered. If local?, only return instances created locally in module.
Iterate over all named terms visible from module. A term can be an instance (or individual) as well as a description. Only terms that haven’t been deleted will be considered. If local?, only return terms created locally in module.
Iterate over all conceived propositions visible from module. Only propositions that haven’t been deleted will be considered. If local?, only return propositions conceived locally in module.
Return a set of values that satisfy the relation relation (a surrogate) applied to nMinusOneArguments plus that last value.
Return a list of sentences describing facts about instanceRef.
Return a set of the most specific types for fillers of the slot relation applied to self.
Return a set of values for the slot relation (a surrogate) applied to self (an object).
Return a set of all (named) relations that specialize relation.
Return a set of all relations that subsume relation.
Return a set of all of the types that are satisfied by self, using only assertions and upward taxonomic reasoning.
Return a list of all terms visible from module. A term can be an instance (or individual) as well as a description. Only terms that haven’t been deleted will be considered. If local?, only return terms created locally in module.
Return a set of all of the types that are satisfied by self.
Iterate over all unnamed terms visible from module. A term can be an instance (or individual) as well as a description. Only terms that haven’t been deleted will be considered. If local?, only return terms created locally in module.
Allocates array space for a neural network with given number of input, hidden and output units.
Allocates array space for a neural network with given number of input, hidden and output units.
Allocates array space for a neural network with given number of input, hidden and output units.
Allocates array space for a neural network with given number of input, hidden and output units.
Similar to allocate-transitive-closure-iterator (which see),
but return a SUPPORTED-CLOSURE-ITERATOR instead.
Return an iterator that generates the transitive closure of applying iterators generated by allocateAdjacencyIterator to startNode. If filterFunction is non-null, that function is applied as a filter to each node generated (nodes filtered out still generate descendants, but they don’t get returned).
Execute a yes/no query composed of input-variables
inputVariables and body queryBody. Before executing, bind variables
to inputBindings (in sequence).
(apply-ask inputVariables queryBody inputBindings)
Apply (inherit) the description description
to members of the vector vector. Return TRUE if no clash was detected.
Constraint propagation happens only if it is enabled prior to calling
apply-kappa?.
Execute a query composed of io-variables variables
and body queryBody. Before executing, bind variables to inputBindings
(in sequence). If one variable is left unbound, returns a cons list of
bindings of that variable. If two or more are unbound, returns
a cons list of cons lists of bindings. Setting the option :singletons?
to FALSE always returns a list of lists. Example call:
(apply-retrieve variables queryBody inputBindings)
Return TRUE if the arguments in proposition subProposition unify with those in referenceProposition. If one of the propositions is a HOLDS, the other one has to be also for unification to succeed.
Similar to ask (which see), but return the highest partial match score
for the supplied proposition instead of a truth value. If the option
:MAXIMIZE-SCORE? is set to FALSE, return after the first partial match score
has been generated.
List version of all-asserted-collection-members for backwards compatibility.
This is called lots of places, including MEMBER-OF-specialist. If self is a native collection,
a skolem collection, or a named description, return a list of asserted members of self, otherwise,
return NULL. If direct?, don’t look down isa hierarchies.
Return true if self has been asserted to be false.
Return true if self has been asserted to be true.
Recurse through :multiple-proofs antecedents guided by the current rule combination scheme.
Propagate the error between proofs conclusion and the desired target value
through proofs network and its antecedents, and adjust weights to decrease the error. Multiple
iterations through forward/backward-propagate-cached-network-proof with updated top-level
errors will train the involved networks to minimize the error as much as possible.
Given a properly forward activated network self for the current set of inputs, and a training error between the current output and the goal value, backpropagate the error and update selfs vector of input errors. Reads output, hidden activations and weights and updates hidden errors and input errors.
Given a properly forward activated network self for the current set of inputs, and a training error between the current output and the goal value, backpropagate the error and update selfs vector of input errors. Reads output, hidden activations and weights and updates hidden errors and input errors.
Given a properly forward activated network self for the current set of inputs, and a training error between the current output and the goal value, backpropagate the error and update selfs vector of input errors. Reads output, hidden activations and weights and updates hidden errors and input errors.
Given a properly forward activated network self for the current set of inputs, and a training error between the current output and the goal value, backpropagate the error and update selfs vector of input errors. Reads output, hidden activations and weights and updates hidden errors and input errors.
Run backward propagation on the current input and target batch and store back-propagated errors in the input error batch.
Run forward propagation on the current input batch and store outputs in the output batch.
Return true if input batch arrays have been filled to capacity.
Compute the same partial match score as the call to compute-partial-truth
that generated proof (which is assumed to have been compacted with a call to
compact-partial-proof-to-network-proof. The score will only be identical of course, if
the various networks and their weights have not yet been updated during learning.
Queue and process operations for update-network-weights for proof.
Queue and process operations for update-network-weights for proof.
Run weight updates for the current input and target batches.
Return TRUE if self is the undefined individual BOTTOM.
Build a neural network for the proposition prop and link them. This builds a two-layer perceptron network whose input nodes are activated by the truth of props arguments and whose output node computes the truth of prop.
Build a neural network for the proposition prop. This builds a two-layer perceptron network whose input nodes are activated by the truth of props arguments and whose output node computes the truth of prop.
Build a neural network for the proposition prop. This builds a two-layer perceptron network whose input nodes are activated by the truth of props arguments and whose output node computes the truth of prop.
Build a neural network for the proposition prop. This builds a two-layer perceptron network whose input nodes are activated by the truth of props arguments and whose output node computes the truth of prop.
Build a neural network for the proposition prop. This builds a two-layer perceptron network whose input nodes are activated by the truth of props arguments and whose output node computes the truth of prop.
Return a list of all definite (TRUE or FALSE) propositions that reference the instance instanceRef.
Callable version of ask (which see). Accepts queries
specified by a query iterator, or specified as a CONS-list of arguments as they
would be supplied to ask. Raises LOGIC-EXCEPTIONs in case of illegal
queries and logical expressions.
Callable version of the defconcept command (which see).
Expects the same arguments as defconcept but supplied as a list.
Callable version of the deffunction command (which see).
Expects the same arguments as deffunction but supplied as a list.
Callable version of the defobject command (which see).
Expects the same arguments as defobject but supplied as a list.
Callable version of the defproposition command (which see).
Expects the same arguments as defproposition but supplied as a list.
Callable version of the defrelation command (which see).
Expects the same arguments as defrelation but supplied as a list.
Callable version of list-undefined-relations (which see).
OBSOLETE: use call-run-forward-rules instead.
Callable version of retrieve (which see). Accepts queries
specified by a query iterator, or specified as a CONS-list of arguments as they
would be supplied to retrieve. Raises LOGIC-EXCEPTIONs in case of illegal
queries and logical expressions.
Callable version of retrieve-partial (which see). Accepts queries
specified by a query iterator, or specified as a CONS-list of arguments as they
would be supplied to retrieve-partial. Raises LOGIC-EXCEPTIONs in case of
illegal queries and logical expressions.
Callable implementation for run-forward-rules (which see).
Set the inference level of module to the level specified by levelKeyword. If module is NULL and we are inside a query, set the level of the current query iterator. Otherwise, set the level globally.
Return TRUE if x is an explicitly marked as ignored or a vector relation that is not also marked as a truth value relation.
Return TRUE if a query is computing Chameleon partial matches.
Return TRUE if x is an explicitly marked primitive value relation.
Return TRUE if x is an explicitly marked truth value relation or otherwise not known to be ignored.
Return TRUE if x is an explicitly marked vector relation.
Return TRUE if objectRef denotes a class.
Clear all currently batched inputs (with keys) and associated target values.
Return the logic instance referred to by self.
Return the logic instance referred to by self, or self if it is a literal (e.g., string or number) that can’t be coerced.
Return a vector containing the elements in self. Coerce each element of self to be a logic object or literal.
Return TRUE if objectRef denotes a relation or a class.
Combine partial match scores from alternative :multiple-proofs scores
according to the current *rule-combination* strategy.
Convert proof into a compacted network tree form that only contains
:AND, :OR, :MULTI and :PRIMITIVE nodes (some of which such as instance-of might contain
further antecedents if they were computed by specialists, for example).
Return the simplest inversion of prop, trying to reuse existing negations or negated arguments of prop if possible.
tree is a term expression (a string or an s-expression), or is a class reference (a symbol or surrogate). Return a (possibly newly-conceived) term representing the internalized representation of that term.
Return the logical conjunction of truth values tv1 and tv2.
Return a CONS tree representation of the proof self. Each proof step is represented as a CONS tree of the form (<proposition> (<key> <value>...) <antecedent>...) where each <antecedent> is a CONS tree representing a subproof. The consification follows the original proof structure literally, i.e., no uninteresting nodes such as patterns or AND-introductions are suppressed.
Generate all solutions for the query self, and collect them into a cons list
of result tuples. If :SINGLETONS? TRUE, collect a list of atoms rather than a
list of lists for tuples of arity=1.
Collect all solutions of self into a cons list and return the result.
Collect queryIterators current solutions into a list of CONS tuples. Only collect output values that have been bound as a result of the query, that is exclude any bindings provided as initial values. If truthValues?, also return a list of truth values for each solution (NIL otherwise). If justifications? and justifications are currently being recorded, also return a list of justifications for each solution (NIL otherwise). Despite the name, this will also work for regular, non-cached queries.
Collect the current solutions of self into a cons list
of result tuples. If :SINGLETONS? TRUE, collect a list of atoms rather than a
list of lists for tuples of arity=1.
Return a CONS tree representation of the proof self. Each proof step is represented as a CONS tree of the form (<proposition> (<key> <value>...) <antecedent>...) where each <antecedent> is a CONS tree representing a subproof. style indicates what nodes in the proof tree should be suppressed. :RAW preserves the original structure literally, :VERBOSE keeps AND- introductions but suppresses all auxiliary (non-logical) nodes such as pattern nodes, and :BRIEF additionally suppresses AND-introduction nodes.
Return true if self is a constant such as TRUE or FALSE.
Return TRUE if objectRef denotes a literal or scalar.
Return a copy of the proof starting at self. Allocates all new justification objects, but structure-shares other information such as propositions and substitutions.
Copy the float values in sequence to buffer between start and end. Missing values will be set to 0.0, extra values will be ignored.
Create a logic object with name name and return it. If type is also supplied, assert that the object belongs to that type.
Create a two-dimensional array with nof-rows rows and nof-columns columns, and initialize it in row-major-order from values. Missing values will be padded with NULL, extraneous values will be ignored.
Create a two-dimensional array with nof-rows rows and nof-columns columns, and initialize it in row-major-order from values. Missing values will be padded with NULL, extraneous values will be ignored.
Return a vector containing values, in order.
Return a new marker storage object, used to
remember with objects have been marked. If supportRecall?
is set, then the iterator recall-marked-objects can be invoked
on the new marker storage object.
Return a vector containing values, in order.
Return the number of currently batched inputs.
Return the current inference level that is active in the current query, the current module, or, otherwise, globally.
Set the debug ID of self to id. If id is NULL, simply use the current ID counter and increment it appropriately.
Return true if self is default false.
Return true if self is default true.
Return TRUE if self is a default truth value.
Defines name as an arithmetic comparision operation using the
test test-name. It will take two wrapped number parameters and
return a wrapped number. The code will use the appropriate test
for the specific subtype of wrapped number actually passed in,
and return the appropriate subtype of wrapped number based on the
normal arithmetic contagion rules.
For example, if both input parameters are wrapped integers then the output will be a wrapped integer. If the inputs are a wrapped integer and a wrapped float then the output will be a wrapped float, etc.
Defines name as an arithmetic comparision operation using the
test test-name. It will take two wrapped number parameters and
return a boolean. The code will use the appropriate test for
the specific subtype of wrapped number actually passed in.
Defines name to be a constraint computation which uses
constraint-test to determine if a fully bound set of variables
satisfies the constraint. The forms in position-computations
are used to compute the value for each of the positions. All such
computations must set the variable value to be the result
computed for the missing position. Setting value to null for
any such computation means that that particular argument cannot
be computed from the others. The input variables in var-list
will be bound to the N arguments to the constraint.
The generated function will return a Stella Object and take as
inputs the values of the N arguments to the constraint. A value
of null means that the value is not available. If all
arguments are not null, then the return value will be a Stella
wrapped boolean indicating whether the constraint is satisified
or not.
If more than one input value is null, then this constraint
code will not be called.
Unlink the network self from its associated proposition and mark it as deleted.
Unlink the network self from its associated proposition and mark it as deleted.
Unlink the network self from its associated proposition and mark it as deleted.
Unlink the network self from its associated proposition and mark it as deleted.
Eliminate all neural networks and remove any connections to propositions and training examples.
Return trun if self has been deleted.
Return trun if self has been deleted.
Return trun if self has been deleted.
Return trun if self has been deleted.
Return a copy of self where all logic objects are
replaced by their generated parse-tree version. This is useful to
convert the result of a retrieval query into a regular parse tree.
Prints a description of self to stream stream. mode
can be :terse, :verbose, or :source. Used by describe.
Return the name of the description self.
Return the name of the description self, if it has one.
Return TRUE if self is a description.
Destroy all propositions that reference self,
and mark it as deleted?, thereby making it invisible within class
extensions.
Destroy self which can be a term or a proposition. Destroy all
propositions that reference self and mark it as deleted? (thereby
making it invisible within class extensions).
Retract and destroy the proposition proposition. Recursively destroy all propositions that reference proposition. Also, destroy all satellite propositions of proposition.
Destroy all propositions that reference self,
and mark it as deleted?, thereby making it invisible within class
extensions. Unlink descriptions from native relations.
Return direct super classes/slots of self.
Return true if the STELLA feature is currently disabled.
Return the logical disjunction of truth values tv1 and tv2.
Return TRUE if d1 and d2 belong to disjoint partitions.
Function version of clear-instances that evaluates its argument.
Save module to the persistent store store which can either be an output stream or a persistent OBJECT-STORE.
Return TRUE if self has zero entries.
Return TRUE if self has length 0.
Return true if the STELLA feature is currently enabled.
Ensure the chameleon.plm ontology file has been loaded (assumes it exists in the current load path).
Return the estimated length of the sequences in self, which could be too large if some of the members have been deleted.
Evaluate self against its arguments, possibly resulting in the setting or changing of its truth value.
Return :POSTED if proposition is on the evaluation queue for *context*, :EVALUATED if has been evaluated, or NULL if it has never been evaluated.
Record the evaluation state of proposition.
Execute queued ops in queue until there is nothing more to do.
Print an explanation for prop if there is one. This will only happen for forward-chained propositions.
Programmer’s interface to WHY function.
Programmer’s interface to the WHYNOT function.
Return TRUE if self represents some form of falsehood.
Return true if self is false (or default-false if we are considering default assertions).
Fetch a duplicate of prop if defined by store. EXPERIMENTAL! The exact semantics of this still needs to be worked out. This needs to be appropriately specialized on actual OBJECT-STORE implementations.
Fetch the instance identified by name (a string or symbol) from store and return it as an appropriate logic object. This needs to be appropriately specialized on actual OBJECT-STORE implementations.
Fetch the relation identified by name (a string or symbol) from store and return it as a named description. This needs to be appropriately specialized on actual OBJECT-STORE implementations.
Fill the two-dimensional array self in row-major-order from values. Missing values will retain their old values, extraneous values will be ignored.
Fill the two-dimensional array self in row-major-order from values. Missing values will retain their old values, extraneous values will be ignored.
Finalize all currently unfinalized objects.
The user-level entry point for this is (process-definitions).
Classify self and return three values, its direct
supers, direct subs, and a list of equivalent descriptions.
Setting supersOnly? may speed up the computation (perhaps by a lot).
If description is nameless and has no dependent propositions, then
it is automatically removed from the hierarchy after classification.
Classify self and return a list of most specific named descriptions among all descriptions that it satisfies.
Return the nearest instance with name instanceRef visible from the current module. instanceRef can be a string, symbol, or surrogate. If instanceRef is a surrogate, the search originates in the module the surrogate was interned in.
Search for a rule named ruleName. Like get-rule,
but find-rule implicity quotes its input argument.
Locally forward-propagate the network associated with just based on
previously cached positive-scores of antecedents.
Compute the same partial match score as the call to compute-partial-truth
that generated proof (which is assumed to have been compacted with a call to
compact-partial-proof-to-network-proof. The score will only be identical of course, if
the various networks and their weights have not yet been updated during learning.
Activates the current inputs of the network self to compute its output.
Sets selfs output slot and returns the computed value. Reads input activations and
weights and updates hidden and output activations.
Activates the current inputs of the network self to compute its output.
Sets selfs output slot and returns the computed value. Reads input activations and
weights and updates hidden and output activations.
Activates the current inputs of the network self to compute its output.
Sets selfs output slot and returns the computed value. Reads input activations and
weights and updates hidden and output activations.
Activates the current inputs of the network self to compute its output.
Sets selfs output slot and returns the computed value. Reads input activations and
weights and updates hidden and output activations.
Activates the current inputs of the network self to compute its output.
Sets selfs output slot and returns the computed value. Reads input activations and
weights and updates hidden and output activations.
Return TRUE if relationRef references a function.
Return an s-expression representing the source expression for self.
Method to generate a specialized term for self. This is designed to allow for extension of the term generation code to cover other types of objects for the logic. This particular method will signal an error unless there is a surrogate-value-inverse link set.
Variant of create-cached-network that takes a training example,
runs its cons query, and stores a compacted version of the associated proof tree
as the examples cached solution which will also be returned. If a cached and
up-to-date solution already exists, it will be returned instead.
Return the namespace module for Chameleon relations
Return the nearest class with name instanceRef visible from the current module. instanceRef can be a string, symbol, or surrogate. If instanceRef is a surrogate, the search originates in the module the surrogate was interned in.
Return propositions forward justifications.
Return the nearest instance with name instanceRef visible from the current module. instanceRef can be a string, symbol, or surrogate. If instanceRef is a surrogate, the search originates in the module the surrogate was interned in.
Return the neural network associated with an :AND or :OR justification. Raise an error if the associated proposition is not linked to a neural network.
Return a module named moduleRef.
Return the proposition linked to self.
Return the proposition linked to self.
Return the proposition linked to self.
Return the proposition linked to self.
Return one of propositions forward justifications supported in the current context.
Return the neural network associated with prop. If error?, raise an exception if it cannot be found, otherwise, simply return NULL.
Return the nearest relation with name instanceRef visible from the current module. instanceRef can be a string, symbol, or surrogate. If instanceRef is a surrogate, the search originates in the module the surrogate was interned in.
Used to convert a computation to reference so-called
template slots rather than own slots: If instanceRef denotes a class,
return a prototype of that class. Otherwise, return instanceRef.
Return a maximum value for the number of fillers of relation relation (a surrogate) applied to the instance self (an object).
Return a minimum value for the number of fillers of relation relation (a surrogate) applied to the instance self (an object).
Return a single value for the slot relation (a surrogate) applied to self (an object).
Return a most specific type for fillers of the slot relation (a surrogate) applied to self. If there is more than one, pick one.
Return all of propositions forward justifications supported in the current context.
Access the TensorFlow callback code registered under name.
Generate a single argument spec for arg that can be used for set-vector-input-values.
arg can either be a proposition or justification.
Generate a single argument spec for arg that can be used for set-vector-input-values.
arg can either be a proposition or justification.
Generate a single argument spec for arg that can be used for set-vector-input-values.
arg can either be a proposition or justification.
Generate a single argument spec for arg that can be used for set-vector-input-values.
arg can either be a proposition or justification.
Return the set of clash exceptions for incoherentProp that are visible from the current context. These are recorded most-recent first in order of occurrance.
Returns the current WHY justification. May also throw one of the following subtypes of EXPLAIN-EXCEPTION: EXPLAIN-NO-QUERY-EXCEPTION EXPLAIN-NO-SOLUTION-EXCEPTION EXPLAIN-NO-MORE-SOLUTIONS-EXCEPTION EXPLAIN-NOT-ENABLED-EXCEPTION EXPLAIN-NO-SUCH-LABEL-EXCEPTION EXPLAIN-QUERY-TRUE-EXCEPTION
Programmer’s interface to WHYNOT function. Derive justifications why query failed, or, if label was supplied as non-NULL, lookup its justification relative to mapping and return the result.
Return TRUE if proposition has any viable forward justifications.
Return TRUE if proposition has any forward justifications supported by the current context.
Return TRUE if self has at least one vector input argument.
Return TRUE if self has at least one vector input argument.
Return TRUE if we have at least one object store opened/linked into PowerLoom, which changes some index access routines to be considerate of that.
Memoizable helper function for truth-value-argument-index and friends.
Helper function for print-concept-outline
Helper function for print-concept-outline
Return TRUE if the partial truth value of arg will be ignored for selfs inputs. This top-level method only looks at arg and ignores self.
Change the current logic dialect to dialect.
Currently supported dialects are KIF, STELLA, and PREFIX-STELLA.
The STELLA dialects are not recommended for the construction of knowledge
bases, they are mainly used internally by PowerLoom.
Return TRUE if self represents INCONSISTENT.
Return true if self is inconsistent (true and false).
Initialize the weights of the neural network self - eiher randomly or from a saved state.
Insert value identified by key into self. If a solution with that key already exists, destructively modify it with the slot values of value. This is necessary to preserve the order of solutions in self.
Return the logical negation of self.
Return true if self is the JIT cache of some world state.
Return TRUE if self is a known truth value, that is either TRUE or FALSE, but not UNKNOWN, INCONSISTENT, etc.
Return the number of entries in self.
Link the network self to its associated proposition prop.
Link the network self to its associated proposition prop.
Link the network self to its associated proposition prop.
Link the network self to its associated proposition prop.
Return a list containing two lists, a list of currently enabled PowerLoom features, and a list of all available PowerLoom features.
Just like QUERY-ITERATOR.consify but return a LIST instead.
Loads all PowerLoom files specified on the command line.
If directories are listed, all PowerLoom files in those directories are loaded.
Since when this is called we might still have unprocessed command line args,
this only looks at files which are to the right of the last argument that
starts with a - character.
Analyze a command tree in a load file and determine when it should be processed.
Load all PowerLoom files (*.plm) in directory in alphabetic sort order.
Read logic commands from stream and evaluate them.
Does the actual work for load (which see for a description of options).
Read logic commands from stream and evaluate them in module.
This is a backwards-compatibility wrapper. Note that module will not override
any in-module declarations in stream.
Return true if proposition’s truth-value has been locally modified in module (initialized, asserted, retracted). This will always succeed for locally conceived propositions regardless of whether their truth value is defined or not.
Return TRUE if the class self or one of its supers supports indices that record extensions referenced by the logic system. Also return true for literal classes.
A sorting predicate for objects o1 and o2 that can appear in logical
forms. Performs a combined numeric and lexocographic sort that accounts
for lists, collections and propositions. Numbers precede all other values,
null follows all other values.
Return TRUE if self is a logic module, implying
that relations defined within it define a knowledge base. A module
is a logic module iff it inherits the module PL-KERNEL.
Lookup the solution identified by key in self and return its value, or NULL if no such solution exists.
Returns the native funtion code for native-name if it exists and the underlying programming languages supports such lookups. It is looked up using the signature of a computation function supported by the computation specialist.
Returns the native funtion code for native-name if it exists and the underlying programming languages supports such lookups. Uses the signature of a specialist function.
Lookup the neural network for prop in the global networks list.
Find or create a cached query and reinitialize as necessary.
variables are the list of IO variables, some of which will be bound by inputBindings.
variables can be NULL in case a query iterator was passed in for query.
query must be a query body cons tree, a query iterator found by find-cached-query,
or NULL. Called by apply-cached-retrieve and apply-cached-ask.
Merge two truth values that are describing the "same" logical proposition. This handles subordination of default to strict values, known over unknown and potential inconsistent values.
In particular, this can be used to check for negated values by asking for the truth of a proposition and its negation, inverting the negation and then using merge to come up with an answer.
Return true if self is the meta-JIT cache of some world state.
Return TRUE if proposition is tied only to at least one meta object and no other objects except literals or function terms that are itself meta objects.
Return TRUE if self is the description of a named class or relation.
True if normalization is governed by natural deduction semantics.
Return TRUE if self has at least 1 entry.
Return TRUE if self has length > 0.
Normalize and provide defaults for options supplied
to train-chameleon-neural-networks.
Notify proof that one of its prerequisites had its computation completed.
Return the nth solution in self, or NULL if it is empty.
Return the key associated with the n-th set of inputs in the input batch.
Return the output value for the n-th set of inputs in the input batch.
Return the 0-based n-th proposition input of self (ignores bias unit).
Return the 0-based n-th proposition input of self (ignores bias unit).
Return the 0-based n-th proposition input of self (ignores bias unit).
Return the 0-based n-th proposition input of self (ignores bias unit).
Return the 0-based n-th proposition input of self (ignores bias unit).
Return the 0-based n-th proposition input error of self (ignores bias unit).
Return the 0-based n-th proposition input error of self (ignores bias unit).
Return the 0-based n-th proposition input error of self (ignores bias unit).
Return the 0-based n-th proposition input error of self (ignores bias unit).
Return the 0-based n-th proposition input error of self (ignores bias unit).
Return error of the k-th input in the n-th set of inputs in the input batch. k ignores the bias unit.
Return the number of arguments of prop whose partial truth value will be ignored for selfs inputs. This top-level method only looks at prop and ignores self.
Return the number of input values expected by self (ignores bias unit).
Return the number of input values expected by self (ignores bias unit).
Return the number of input values expected by self (ignores bias unit).
Return the number of input values expected by self (ignores bias unit).
Return the number of input values expected by self (ignores bias unit).
Return the number of arguments of prop whose partial truth value will be considered for selfs inputs. This top-level method only looks at prop and ignores self.
Return the total number of input positions to store all elements of all vector arguments of prop. prop can be NULL in which case the linked proposition will be used.
Return the total number of argument specs generated by vector arguments of prop.
This is only different than number-of-vector-arguments if at least one of props vector argument
relations has arity > 1. prop can be NULL in which case the linked proposition will be used.
Return the number of arguments of prop that yield one or more vectors for selfs inputs. prop can be NULL in which case the linked proposition will be used.
Return the number of arguments of prop that yield one or more vectors for selfs inputs. prop can be NULL in which case the linked proposition will be used.
Return the name symbol for the logic object self.
Return the name string for the logic object self.
Return the surrogate naming the object self, which may be a Stella class that is used in PowerLoom as well as a more normal powerloom object.
Return the name of the logic object self to name.
This is simply a wrapper to parse-options (which see).
We keep this alias in case we ever want to add any logic-specific extensions.
Remove and return the first solution of self or NULL if the table is empty.
Push self onto the evaluation queue (unless it’s already there).
Run the PowerLoom listener. Read logic commands from the
standard input, evaluate them, and print their results. Exit if the user
entered bye, exit, halt, quit, or stop.
Exit hook to stop the PowerLoom GUI if it is running.
Returns information about the current PowerLoom implementation. Useful when reporting problems.
Pretty-print the logical form form to stream according
to the current setting of *logic-dialect*.
Print the array self to stream.
Print the array self to stream.
Print the extension sizes of concepts visible in module. If module is NULL the current module is used. Do not report extensions with size less than sizeCutoff (default is 10).
Like all-facts-of, but sorts and prints each fact
on a separate line on the standard output stream.
Print stack of goals. Assumes that query has been interrupted with a full stack of control frames.
Print the logical form form to stream according to
the current setting of *logic-dialect*. Pretty-printing is controlled
by the current setting of *prettyPrintLogicalForms?*.
Produce a stringified version of a logical representation of self and write it to the stream stream. Use the dialect dialect, or use the current dialect if dialect is NULL.
Print the logical form form to stream according to
the current setting of *logic-dialect*. Pretty-printing is explicitly
forced to be turned off.
Print a WHYNOT justification to stream according to maxDepth and style. Print a summary only if summary? is TRUE.
Return true if the truth-value of self satisfies minimalStrength.
Push input values onto the input batch array and associate them with key (which can be NULL). Associating a key lets us easily map inputs/outputs to some processing object of interest (e.g., a justification).
Push a target value onto the target batch array which is assumed to correspond to the input values at the same respective position in the batch. Inputs and targets are associated by position in the batch, they can be pushed independently, as long as they are fully synchronized when processing of the batch starts.
Push vectorSpecs onto the vector argument batch array which is assumed to correspond to the input values at the same respective position in the batch. Truth-valued and vector-valued inputs are associated by position in the batch, they can be pushed independently, as long as they are fully synchronized when processing of the batch starts. If self has no vector-valued argument, the associated batch array can be left undefined.
Queue inputs in nets input batch. Execute the current batch if we are full.
Generate a random integer in the interval [0..n-1]. n must be <= 2^15.
Randomize the weights of the neural network self.
Randomize the weights of the neural network self.
Randomize the weights of the neural network self.
Randomize the weights of the neural network self.
Undo all training and randomize weights in all neural networks.
Return an iterator that generates all marked objects recorded in self.
Return TRUE if every query records justifications to enable the explanation of concluded results.
Creates a registration entry for name as a computation which executes code. Essentially just builds the Stella meta-information tructure needed to funcall name as a computation function by the computation specialist. The function definition in code needs to accept ARITY Stella OBJECTs as arguments and return a Stella OBJECT suitable for PowerLoom use. (These are generally LOGIC-OBJECTs and the literal wrappers FLOAT-WRAPPER, INTEGER-WRAPPER and STRING-WRAPPER.)
registers a computation function stella-name based on the native-name for the particular programming language in question. Use of this command makes the resulting code or knowledge bases non-portable to other target languages.
Register fn as a logic-object print function for dialect.
Each function should have the signature ((self OBJECT) (stream OUTPUT-STREAM)).
Any return values will be ignored.
Register the network self on the global networks list (assumes self has been linked).
Register store as an active object store.
Creates a registration entry for name as a specialist which executes code. Essentially just builds the Stella meta-information tructure needed to funcall name as a specialist. The function definition in code needs to accept a CONTROL-FRAME and KEYWORD as arguments and return a KEYWORD. Side effects on elements of the proposition in the control frame can be used to bind and thus return values.
registers a specialist function stella-name based on the native-name for the particular programming language in question. Use of this command makes the resulting code or knowledge bases non-portable to other target languages.
Special-purpose callback support that registers code under the logic symbol with name, which by convention we make the qualified method name of the method we are using this for. This is a special-purpose hack which eventually we might want to generalize so others can use it too.
Given a relation object, return it’s name.
Return TRUE if objectRef denotes a relation or a class.
Remove the solution identified by key from self. To preserve the solution ordering chain, the solution is marked as deleted and will be completely removed upon the next iteration through self.
Destructively remove all deleted members of self.
Resets the collection of clash exceptions
Zero out all caches managed by the query optimizer, so that it will reoptimize subgoal queries upon next invocation.
Retract all definite (TRUE or FALSE) propositions attached to self.
Partial-match version of retrieve (which see) that generates scored
partial solutions based on the current partial match strategy. By supplying
BEST instead of ALL, or by adding the option :SORT-BY :SCORE, the generated
solutions will be sorted so that solutions with higher scores come first.
Use the :MATCH-MODE option to override the global default setting established
by set-partial-match-mode, e.g., use :MATCH-MODE :NN to use the neural
net partial match mode. The :MINIMUM-SCORE option can be used to only
retrieve solutions that have at least the specified minimum match score.
By default, retrieve-partial does not maximize the match scores of its returned
bindings. To only get maximal scores use :MAXIMIZE-SCORE? TRUE (this is not
yet implemented - you can use ask-partial to maximize scores for individual
solutions by hand).
Retrieve a subset of current training examples defined via cham/training-example.
Run forward inference rules in the module defined by the :module option (which defaults to the current module). If forward inferencing is already up-to-date in the designated module, no additional inferencing will occur, unless the :force? option is specified as TRUE, in which case all forward rules are run or rerun (for backwards compatibility, this command also supports the old <module> :force arguments specified with a non-standard keyword notation). If :local? is TRUE (the default), only propositions asserted locally in the module will be posted for forward inference, otherwise, all true propositions visible in the module will be posted.
Calling run-forward-rules with :disable-jit? TRUE will disable future just-in-time
forward inference which will preserve cached forward inference in case of monotonic
assertions that do not clip (implicitly retract) any existing values. Subsequent calls
can then trigger new inferences based on these updates only. If the JIT stays enabled,
even monotonic updates will throw away the JIT cache. Also, once any retractions have
been performed, all cached inferences will be thrown away and forward inference has to
be rerun from scratch or just-in-time forward inference needs to be enabled again.
Run the PowerLoom test suite. Currently this simply runs all demos and echos commands and their results to standard output. The output can then be diffed with previously validated runs to find deviations.
Try to prove whether instanceOrTuple satisfies the definition of the relation relationRef and return the result truth value of the query. instanceOrTuple can be a single object, the name or surrogate of an object, or a collection (a list or vector) of objects. relationRef can be a relation, description, surrogate or relation name.
Save all neural networks to file (if file is non-NULL).
If networks are saved periodically (see set-save-network-cycle) this file
name will be used to perform periodic saves.
Helping function for specialists. Return the appropriate keyword indicating success or failure of a proof.
Helping function for specialists testing the validity of a
fully bound inference frame. Based on the test result success?
and reversePolarity?*, set the truth value of frame and return
an appropriate keyword. The keyword will be either :final-success
:terminal-failure if terminal? is true. Otherwise it will be
:final-success or :failure.
Select a subset of currently defined training examples. Currently the selection is purely based on module and/or number. Results will be shuffled if :shuffle? is TRUE (default).
Set number of cycles between which error rates are saved to
the file established by the last call to save-all-neural-networks appended
with extension .err. A number <= 0 (or NULL) turns off periodic saving.
Set the inference level of module to the level specified
by levelKeyword. If module is NULL, set the level globally.
Set the current truth-value inputs of the network self to float values in sequence. Missing inputs will be set to 0.0, extra values will be ignored.
Set the current truth-value inputs of the network self to float values in sequence. Missing inputs will be set to 0.0, extra values will be ignored.
Set the current truth-value inputs of the network self to float values in sequence. Missing inputs will be set to 0.0, extra values will be ignored.
Set the current truth-value inputs of the network self to float values in sequence. Missing inputs will be set to 0.0, extra values will be ignored.
Set the current truth-value inputs of the network self to float values in sequence. Missing inputs will be set to 0.0, extra values will be ignored.
Record membership of object in the marker storage object self.
Sets the number of nearest neighbors to predict from.
Sets the number of training examples for each case in the training set.
Enable the PowerLoom environment feature feature.
Set number of cycles between which networks are saved to the
file established by the last call to save-all-neural-networks. A number <= 0
or a NULL number turns off periodic saving.
Set the current vector inputs of the network self to the vectors described by vectorSpecs. Each vector spec describes a vector-generating proposition that produces one or more vectors. How those specs are translated into actual numeric vectors such as embeddings is specific to the particular neural network type.
Set the current vector inputs of the network self to the vectors described by vectorSpecs. Each vector spec describes a vector-generating proposition that produces one or more vectors. How those specs are translated into actual numeric vectors such as embeddings is specific to the particular neural network type.
Set the current vector inputs of the network self to the vectors described by vectorSpecs. Each vector spec describes a vector-generating proposition that produces one or more vectors. How those specs are translated into actual numeric vectors such as embeddings is specific to the particular neural network type.
Set the current vector inputs of the network self to the vectors described by vectorSpecs. Each vector spec describes a vector-generating proposition that produces one or more vectors. How those specs are translated into actual numeric vectors such as embeddings is specific to the particular neural network type.
Set the current vector inputs of the network self to the vectors described by vectorSpecs. Each vector spec describes a vector-generating proposition that produces one or more vectors. How those specs are translated into actual numeric vectors such as embeddings is specific to the particular neural network type.
Similar to copy but does not copy antecedent justifications.
Perform a stable, destructive sort of self according to
predicate, and return the result. If predicate has a < semantics, the
result will be in ascending order.
Try to prove if the description associated with subObject specializes the description for superObject and return the result truth value of the query.
Start the PowerLoom HTTP server at :port (defaults to 9090). Loads the required support systems in Lisp and Java if necessary (C++ is not yet supported).
Start the PowerLoom server at :port (defaults to 9090) and launches
the GUI which will communicate with the server at that port. If :host is specified,
the GUI will try to communicate with a server at host:port instead of the local
embedded server (note, you can always point the GUI manually to a different server
from its Connect to Server menu item). Loads the required support systems if necessary.
Embedded calls to the GUI are currently only supported in Java; however, when the GUI is
run in standalone mode, it can communicate with any PowerLoom installation that supports
an HTTP server (currently Lisp and Java).
Start the PowerLoom HTTP server at :port (defaults to 9090). Loads the required support systems in Lisp and Java if necessary (C++ is not yet supported).
Stop the PowerLoom HTTP server and free up any bound ports. This is a no-op if no server is running or the server is not supported.
Closes the PowerLoom GUI application if it is currently visible. This is a no-op if the GUI is not running or if it is not supported.
Stop the PowerLoom HTTP server and free up any bound ports. This is a no-op if no server is running or the server is not supported.
If tv2 has greater strength than tv1, adapt the strength of tv1 (not its value!) and return the result. Otherwise, return tv1 unmodified.
Return TRUE if self is a strict truth value.
Compare the two truth values and return true if tv1 is strictly
stronger than tv2. Stronger uses the following partial order:
INCONSISTENT > STRICT > DEFAULT > UNKNOWN.
Truth values at the same level are equal in strength.
Return TRUE if TensorFlow callbacks have been properly registered.
Convert self into an equivalent PowerLoom object that can be passed as an argument wherever an instance is expected.
Return TRUE if relation (a surrogate) is asserted to be closed or if the current module closes all relations.
Return TRUE if relation (a surrogate) is a function.
Return TRUE if object is stored (marked) in self.
Return TRUE if relation (a surrogate) is TRUE when applied to arguments.
Return TRUE if the proposition (<relation> <self> <filler>)
is true.
Return TRUE if the object self is stored (marked)
in the table pointed at by the special variable *specialMarkerTable*.
Designed for use by remove-if.
Return TRUE if subrelation specializes superrelation.
Return TRUE if self satisfies type.
Train rule neural networks based on :n-train (or all) training examples looked up in :module/:local?. Train for :epochs (defaults to 20) or until :error-cutoff is reached. Print every :print-cycle epochs or not at all. If :shuffle? (the default) randomly shuffle the selected training examples before every epoch. If :batch?, use batch training mechanism (which will fail if the current network implementation does not support it).
Translate the Loom file input to PowerLoom and write the translation to the file output. Note that this will only work for fairly vanilla Loom files that do not contain any Lisp-isms. It might require to clean the Loom file manually before this translation will work.
Return TRUE if self represents some form of truth.
Return true if self is true (or default-true if we are considering default assertions).
Return the 0-based input position of truth value argument arg. Ignores bias unit which
is a network-implementation-specific detail. Generates indices in the order expected by set-input-values.
If arg is not a truth value argument, returns -1.
Return TRUE if the partial truth value of arg will be considered for selfs inputs. This top-level method only looks at arg and ignores self.
Retract the truth, falsity or inconsistency of proposition. This is a
more general version of retract that also handles falsity. For example, if
we assert the proposition "(not (sad Fred))", and then execute the statement
"(unassert (sad Fred))", the truth value of the proposition "(sad Fred)"
will be set to UNKNOWN. If we had called retract in place of unassert,
the proposition "(sad Fred)" would remain set to FALSE. Note that for this
unassertion to succeed, the logic constant Fred and the relation sad must
already be defined.
Retract the truth, falsity or inconsistency of the proposition self.
Return TRUE if self represents UNKNOWN.
Return true if the truth of self is unknown.
Unlink the network self from its associated proposition.
Unlink the network self from its associated proposition.
Unlink the network self from its associated proposition.
Unlink the network self from its associated proposition.
Return TRUE if self is an unnamed description.
Unregister the network self on the global networks list.
Remove store from the list of active object stores.
Disable the PowerLoom environment feature feature.
Classify all named descriptions.
Classify all named instances.
Classify instances local to module and inherited by module. If local?, don’t classify inherited descriptions. If module is NULL, classify descriptions in all modules.
Classify named descriptions local to module and inherited by module. If local?, don’t classify inherited descriptions. If module is NULL, classify descriptions in all modules.
Given a properly forward activated and backpropagated network self for the current inputs and training error, update the network’s weights according to current gradients, learning rate and momentum terms to reduce the error for the given inputs. Reads output, hidden and input activations, hidden error, weights and weight deltas, and updates weights and weight deltas.
Given a properly forward activated and backpropagated network self for the current inputs and training error, update the network’s weights according to current gradients, learning rate and momentum terms to reduce the error for the given inputs. Reads output, hidden and input activations, hidden error, weights and weight deltas, and updates weights and weight deltas.
Given a properly forward activated and backpropagated network self for the current inputs and training error, update the network’s weights according to current gradients, learning rate and momentum terms to reduce the error for the given inputs. Reads output, hidden and input activations, hidden error, weights and weight deltas, and updates weights and weight deltas.
Given a properly forward activated and backpropagated network self for the current inputs and training error, update the network’s weights according to current gradients, learning rate and momentum terms to reduce the error for the given inputs. Reads output, hidden and input activations, hidden error, weights and weight deltas, and updates weights and weight deltas.
A module with store has had the truth value of proposition change according to update-mode. The default method does nothing.
Assert or retract a proposition that applies relation to arguments.
Return the 0-based input position of vector argument arg. Ignores bias unit which is a network-implementation-specific detail. If arg is not a vector argument, returns -1.
Return the 0-based input position of vector argument arg. Ignores bias unit which is a network-implementation-specific detail. If arg is not a vector argument, returns -1.
Return TRUE if arg yields one or more vectors for selfs inputs.
Return TRUE if arg yields one or more vectors for selfs inputs.
If tv2 has lesser strength than tv1, adapt the strength of tv1 (not its value!) and return the result. Otherwise, return tv1 unmodified.
Execute body within the module resulting from moduleForm.
*module* or NULL are acceptable moduleForms. It will locally rebind
*module* and *context* and shield the outer bindings from changes.
Used during classification. Execute body within the indicated classification session and inference world.
Execute body within the meta cache of the current module. Set appropriate special variables.
Access the first element of the CONS-list self (available to support access in C).
Return TRUE if the CONS-list self is empty (available to support access in C).
Return the length of the CONS-list self (available to support access in C).
Access the n-th element of the CONS-list self (available to support access in C).
Access the n-th rest (tail) of the CONS-list self (available to support access in C).
Access the rest (tail) of the CONS-list self (available to support access in C).
Set the first element of the CONS-list self to value (available to support access in C).
Set the n-th element of the CONS-list self to value (available to support access in C). The run-time complexity of this operation is O(n).
Set the n-th rest (tail) of the CONS-list self to rest (available to support access in C). If rest is NULL, it will point to the empty list. The run-time complexity of this operation is O(n).
Set the rest (tail) of the CONS-list self to rest (available to support access in C). If rest is NULL, it will point to the empty list.
Create a STELLA boolean wrapper for self.
Create a STELLA CONS-cell whose value points to first and tail points to rest. If rest is NULL, it will point to the empty list (available as a constructor in C).
Create a STELLA CONS-list of length with elements initialized to NULL (available as a constructor in C).
Create a STELLA string wrapper for self. This wraps self directly and does not allocate a PowerLoom copy first, hence, memory management of self needs to be carefully handled by the caller.
Create a STELLA float wrapper for self.
Create a STELLA integer wrapper for self.
Returns the Stella keyword name, creating it if necessary. name is treated literally and case-sensitively. This should generally not be necessary to do.
Create a STELLA long-integer wrapper for self.
Create a PowerLoom copy of the string content which uses PowerLoom memory management and garbage collection. This is primarily necessary in C environments when a string is passed to a PowerLoom PLI function which might then be pointed to by a PowerLoom result object such as a string wrapper. If such an argument string is later explicitly freed or overwritten, the result object would have a dangling reference or its content destroyed.
Create a STELLA string wrapper for self. This automatically copies self before the wrapper is allocated to avoid garbage collection issues in case self is freed.
Returns the Stella symbol name visible in module, creating it if necessary. name is treated literally and ALWAYS case-sensitively, even if module is case insensitive. This should generally not be necessary to do.
Return the module named name or raise a STELLA no-such-context-exception if it doesn’t exist.
Return the arity of the proposition currently being inferred by frame. This will be the actual number of arguments, even if the proposition is of variable arity.
Return a string representation of the argument binding pattern of frame.
The syntax of the pattern is the same as used for is-frame-binding-pattern (which see).
Return a frame variable binding stack state that can be used to unbind all frame variables to the state they were in when the binding stack state was recorded.
Return the iterator generating successive bindings for the proposition that is currently being inferred by frame.
Return the proposition currently being inferred by this control frame. This proposition will generally have bound and/or unbound pattern variables as its arguments.
Return inconsistent propositions visible in module.
Returns the Stella KEYWORD name if it exists. Case sensitive.
Return the name symbol of the logic object type of self. This type describes the basic
role this object plays in the logic system such as PROPOSITION, CONCEPT, THING, etc., as indicated by
its native object type, it does not have anything to do with type propositions asserted about self.
Return the qualified name of the logic type of self.
Return the unqualified name of the logic type of self.
Return the name, qualified as necessary, so that obj can be found from
module. If there is no name for the object return null.
Return the n-th argument of the proposition that is currently being inferred by frame. This will generally be a pattern variable that might be unbound or bound.
Return the binding of the n-th argument of the proposition that is currently being inferred by frame. This will be NULL if the argument is unbound, otherwise, the object bound to the variable.
Synonym for get-property-string (which see).
Return the value of the global STELLA configuration property as a boolean or defaultValue if it is undefined. Raises an error if the defined value is not a boolean.
Return the value of the global STELLA configuration property as a float or defaultValue if it is undefined. Raises an error if the defined value is not a float.
Return the value of the global STELLA configuration property as an integer or defaultValue if it is undefined. Raises an error if the defined value is not an integer.
Return the value of the global STELLA configuration property or defaultValue if it is undefined.
Return the value of the global STELLA configuration property as a string or defaultValue if it is undefined.
Return the short name of obj, if it has one. Otherwise return null.
Return the fully qualified name of skolem. Note that skolems are anonymous objects which cannot be referenced by their name in logic expressions, so this function is only useful in special circumstances such as inside specialist computations, etc.
Return the unqualified name of skolem. Note that skolems are anonymous objects which cannot be referenced by their name in logic expressions, so this function is only useful in special circumstances such as inside specialist computations, etc.
Return the name symbol of the STELLA class of which self is a direct instance.
Return the qualified name of the STELLA class of which self is a direct instance.
Return the unqualified name of the STELLA class of which self is a direct instance.
Returns the Stella SYMBOL name visible in module if it exists. name is ALWAYS treated literally and case sensitively.
Return propositions such that (relation arg) is true. The relation argument must be bound to a relation. arg may be set to NULL, which is interpreted as a wildcard.
Return relations that have been referenced but not defined in module.
Return the fully qualified name of variable.
Return the unqualified name of variable.
Return TRUE if property has a defined value.
Return TRUE if the arguments of frame are bound according to pattern.
Each character in pattern corresponds to an argument at the particular position and must be
either B (for bound) or _ for unbound. The function returns TRUE if frame has at least
as many arguments as pattern has characters and their bindings match pattern. If frame
has additional arguments, they will be ignored. If pattern has more positions than frame
has arguments, the function returns FALSE.
Return TRUE if object is an instance of the concept concept. This uses subsumption inference only.
Tests whether tv is a known truth value (i.e., true or false).
Test whether obj is of type SKOLEM (which subsumes pattern variables).
Return TRUE if a proposition (relation args) has
been asserted (or inferred by forward chaining).
Test whether obj is of type PATTERN-VARIABLE.
Access the first element of the list self (available to support access in C).
Access the n-th element of the list self (available to support access in C).
Access the rest (tail) of the list self (available to support access in C). NOTE: the rest of a STELLA list is a CONS-list not a LIST.
Read logic commands from the file named filename and evaluate them.
Binds the load command’s :module option to module. See load command help for more information.
Read logic commands from the native input stream stream and evaluate them.
Assumes stream is a line-buffered stream which is a safe compromise but does
not generate the best efficiency for block-buffered streams such as files.
Binds the load command’s :module option to module. See load command help for more information.
Read logic commands from the STELLA stream stream and evaluate them.
Binds the load command’s :module option to module. See load command help for more information.
Return TRUE if the iterator self has no more elements.
This is here primarily to support the is-... naming convention for tests.
C-callable iterator value-slot accessor.
Run the PowerLoom read/eval/print loop.
Print all currently defined STELLA configuration properties and values.
Register name as a function name in module which will invoke the
native code procedure described by function-reference. The name
is a fully-qualified name which will be interpreted by the normal
rules for reading names in PowerLoom. The function must conform
to the signature for computation functions used by the computation
specialist. Arity specifies the number of arguments the computation
accepts.
The exact form of function-reference depends on the underlying programming language. The following type mappings are used: C++: cpp_function_code (a pointer to the function code) Common Lisp: FUNCTION (result of #’ or (FUNCTION ...)) Java: java.lang.reflect.Method
Register object to prevent it from being garbage-collected by the PowerLoom GC. The result of registration is a handle for object which can later be used to unregister it. Currently it is assumed that calling this immediately after an object was returned by a PLI function is safe; however, we might have to handle registration in PLI functions before objects are returned. This is not thread safe and needs to be explicitly synchronized in a threaded environment.
Register name as a function name in module which will invoke the
native code procedure described by function-reference. The name
is a fully-qualified name which will be interpreted by the normal
rules for reading names in PowerLoom. The function must conform
to the signature for specialist functions.
The exact form of function-reference depends on the underlying programming language. The following type mappings are used: C++: Common Lisp: FUNCTION (result of #’ or (FUNCTION ...)) Java: java.lang.reflect.Method
Return TRUE if a proposition that prints as the string
relation-and-arguments is true in the module named module-name.
A module name of null or the empty string refers to the current module.
If no module can be found with the name module-name, then a Stella
no-such-context-exception is thrown.
Register name as a function name in the module named module-name. This function will the native code named native-name. The name is a fully-qualified name which will be interpreted by the normal rules for reading names in PowerLoom. The native-name will be processed in a manner that depends on the underlying programming language. The following type mappings are used: C++: Not available. Error signaled. Common Lisp: The native-name is read by READ-FROM-STRING and then the SYMBOL-FUNCTION is taken. Java: A fully package-qualified name is required. It is looked up using the Reflection tools. The function found must conform to the signature for computation functions. Arity specifies the number of arguments the computation accepts.
Register name as a function name in the module named module-name. This function will the native code named native-name. The name is a fully-qualified name which will be interpreted by the normal rules for reading names in PowerLoom. The native-name will be processed in a manner that depends on the underlying programming language. The following type mappings are used: C++: Not available. Error signaled. Common Lisp: The native-name is read by READ-FROM-STRING and then the SYMBOL-FUNCTION is taken. Java: A fully package-qualified name is required. It is looked up using the Reflection tools. The function found must conform to the signature for specialist functions.
Reset the frame variable binding stack state to state. This will set all frame variables to the state they were in when the binding stack state was recorded.
Set the iterator for generating successive bindings for the proposition that is currently being inferred by frame to iterator.
Set the binding of the n-th argument of the proposition that is currently
being inferred by frame to value. If the binding was successful, that is if the argument
was unbound or is already bound to binding, the function returns TRUE. Otherwise, the
argument will keep its current binding and FALSE will be returned.
Synonym for set-property-string (which see).
Set the global STELLA configuration property to a boolean value.
Set the global STELLA configuration property to a float value.
Set the global STELLA configuration property to an integer value.
Set the global STELLA configuration property to value.
Set the global STELLA configuration property to a string value.
Read the STELLA expression in module and return the result. Raises a reader exception in case of incorrect STELLA syntax.
Test if env has level set to level
Unregister the object whose registered handle is objectHandle to make it available for PowerLoom garbage collection. This also is not thread safe and needs to be explicitly synchronized in a threaded environment.
Convert self into a Stella CONS.
Convert self into a Stella LIST.
Bootstrap the PowerLoom built-in kernel KB.
| [ << ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
This document was generated by Hans Chalupsky on January 6, 2023 using texi2html 1.82.