# AMPL¶

class ampl::AMPL

An AMPL translator.

An object of this class can be used to do the following tasks:

AMPL stores one or more problems which may consume substantial amount of memory. The AMPL object has a destructor which automaticallly closes the underlying AMPL interpreter and deletes all C++ entitites linked to it. Some entities (notably all instances of DataFrame and DataFrame::Slice) are not owned by the AMPL object and are not therefore deleted.

The initialization of the C++ representation of the AMPL entities (any class derived from ampl::Entity) is lazy and consists of two steps. When a function listing available Elements is called, only a shallow list with names, types and declarations of the Elements is populated. The same happens when a reference to a C++ representation of one entity is obtained (through a call to any other single entity factory function). When accessing any instance of an entity (through the methods ‘get’ of any class), the (memory hungry) list of instances for that entity is created.

Consistency is maintained automatically. Any command issued to the translator through eval and similar functions invalidates all entities, and any further access to any entity will require communication between the native translator and the C++ code. Conversely, any operation called through the C++ equivalent function, like fixing variables or solving the model will invalidate only the entities involved. A list of dependencies between entities is automatically updated.

Error handling is two-faced:

• Errors coming from the underlying AMPL translator (e.g. syntax errors and warnings obtained calling the eval method) are handled by the ErrorHandler which can be set and get via getErrorHandler() and setErrorHandler().
• Generic errors coming from misusing the API, which are detected in C++, are thrown as exceptions.
The default implementation of the error handler throws exceptions on errors and prints to console on warnings.

The output of every user interaction with the underlying translator is handled implementing the abstract class ampl::OutputHandler. The (only) method is called at each block of output from the translator. The current output handler can be accessed and set via AMPL::getOutputHandler() and AMPL::setOutputHandler().

Public Functions

AMPL()

Default constructor: creates a new AMPL instance with the default environment.

Exceptions
• std::runtime_error -

If no valid AMPL license has been found or if the translator cannot be started for any other reason.

AMPL(const Environment &e)

Constructor: creates a new AMPL instance with the specified environment. This allows the user to specify the location of the AMPL binaries to be used and to modify the environment variables in which the AMPL interpreter will run

Exceptions
• std::runtime_error -

If no valid AMPL license has been found or if the translator cannot be started for any other reason.

~AMPL()

Default destructor Releases all the resources related to the AMPL instance (most notably kills the underlying interpreter.

DataFrame getData(StringArgs statements) const

Get the data corresponding to the display statements.

The statements can be AMPL expressions, or entities. It captures the equivalent of the command:

where ds1, ..., dsn are the displayStatements with which the function is called.

As only one DataFrame is returned, the operation will fail if the results of the display statements cannot be indexed over the same set. As a result, any attempt to get data from more than one set, or to get data for multiple parameters with a different number of indexing sets will fail.

Return
DataFrame capturing the output of the display command in tabular form.
Exceptions
• AMPLException -

if the AMPL visualization command does not succeed for one of the reasons listed above.

Parameters
• statements -

The display statements to be fetched.

Entity getEntity(fmt::CStringRef name) const

Get entity corresponding to the specified name (looks for it in all types of entities)

Return
The AMPL entity with the specified name
Parameters
• name -

Name of the entity

Exceptions
• std::out_of_range -

exception if the specified entity does not exist

Variable getVariable(fmt::CStringRef name) const

Get the variable with the corresponding name.

Parameters
• name -

Name of the variable to be found

Exceptions
• std::out_of_range -

exception if the specified variable does not exist

Constraint getConstraint(fmt::CStringRef name) const

Get the constraint with the corresponding name.

Parameters
• name -

Name of the constraint to be found

Exceptions
• std::out_of_range -

exception if the specified constraint does not exist

Objective getObjective(fmt::CStringRef name) const

Get the objective with the corresponding name.

Parameters
• name -

Name of the objective to be found

Exceptions
• An -

std::out_of_range exception if the specified objective does not exist

Set getSet(fmt::CStringRef name) const

Get the set with the corresponding name.

Parameters
• name -

Name of the set to be found

Exceptions
• std::out_of_range -

exception if the specified set does not exist

Parameter getParameter(fmt::CStringRef name) const

Get the parameter with the corresponding name.

Parameters
• name -

Name of the parameter to be found

Exceptions
• std::out_of_range -

exception if the specified parameter does not exist

void eval(fmt::CStringRef amplstatements)

Parses AMPL code and evaluates it as a possibly empty sequence of AMPL declarations and statements.

As a side effect, it invalidates all entities (as the passed statements can contain any arbitrary command); the lists of entities will be re-populated lazily (at first access)

The output of interpreting the statements is passed to the current OutputHandler (see getOutputHandler and setOutputHandler).

By default, errors are reported as exceptions and warnings are printed on stdout. This behavior can be changed reassigning an ErrorHandler using setErrorHandler.

Parameters
• amplstatements -

A collection of AMPL statements and declarations to be passed to the interpreter

Exceptions
• std::runtime_error -

if the input is not a complete AMPL statement (e.g. if it does not end with semicolon) or if the underlying interpreter is not running

void reset()

Clears all entities in the underlying AMPL interpreter, clears all maps and invalidates all entities.

void close()

Stops the underlying engine, and release all any further attempt to execute optimisation commands without restarting it will throw an exception.

bool isRunning() const

Returns true if the underlying engine is running.

bool isBusy() const

Returns true if the underlying engine is doing an async operation.

void solve()

Solve the current model.

Exceptions
• std::runtime_error -

If the underlying interpreter is not running

void readAsync(fmt::CStringRef filename, Runnable *cb)

Interprets the specified file asynchronously, interpreting it as a model or a script file.

As a side effect, it invalidates all entities (as the passed file can contain any arbitrary command); the lists of entities will be re-populated lazily (at first access)

Parameters
• filename -

Path to the file (Relative to the current working directory or absolute)

• cb -

Callback to be executed when the file has been interpreted

void readDataAsync(fmt::CStringRef filename, Runnable *cb)

Interprets the specified data file asynchronously.

When interpreting is over, the specified callback is called. The file is interpreted as data. As a side effect, it invalidates all entities (as the passed file can contain any arbitrary command); the lists of entities will be re-populated lazily (at first access)

Parameters
• filename -

Full path to the file

• cb -

Callback to be executed when the file has been interpreted

void evalAsync(fmt::CStringRef amplstatement, Runnable *cb)

Interpret the given AMPL statement Throws runtime_error if the underlying ampl interpreter is not running

void solveAsync(Runnable *cb)

Solve the current model asynchronously.

Parameters
• cb -

Object encapsulating the function to be called when solver is done

void interrupt()

Interrupt an underlying asynchronous operation (execution of AMPL code by the AMPL interpreter).

An asynchronous operation can be started via evalAsync(), solveAsync(), readAsync() and readDataAsync(). Does nothing if the engine and the solver are idle.

std::string cd() const

Get the current working directory from the underlying interpreter (see https://en.wikipedia.org/wiki/Working_directory ).

Return
Current working directory

std::string cd(fmt::CStringRef path)

Change or display the current working directory (see https://en.wikipedia.org/wiki/Working_directory ).

Return
Current working directory
Parameters
• path -

New working directory or null (to display the working directory)

void setOption(fmt::CStringRef name, fmt::CStringRef value)

Set an AMPL option to a specified value.

Parameters
• name -

Name of the option to be set (alphanumeric without spaces)

• value -

String representing the value the option must be set to

Exceptions
• std::invalid_argument -

if the option name is not valid

Optional<std::string> getOption(fmt::CStringRef name) const

Get the current value of the specified option.

If the option does not exist, the returned Optional object will not have value. Usage: if (auto value = ampl.getOption(“solver”)) value.get();

Return
Value of the option as a string.
Parameters
• name -

Option name (alphanumeric)

Exceptions
• std::invalid_argument -

if the option name is not valid

Optional<int> getIntOption(fmt::CStringRef name) const

Get the current value of the specified integer option.

If the option does not exist, the returned Optional object will not have value.

Return
Value of the option (integer)
Parameters
• name -

Option name (alphanumeric)

Exceptions
• std::invalid_argument -

if the option name is not valid

• std::invalid_argument -

If the option did not have a value which could be casted to integer

void setIntOption(fmt::CStringRef name, int value)

Set an AMPL option to a specified integer value.

Parameters
• name -

Name of the option to be set

• value -

The integer value the option must be set to

Exceptions
• std::invalid_argument -

if the option name is not valid

Optional<double> getDblOption(fmt::CStringRef name) const

Get the current value of the specified double option If the option does not exist, the returned Optional object will not have value.

Return
Value of the option (double)
Parameters
• name -

Option name

Exceptions
• std::invalid_argument -

if the option name is not valid

• std::invalid_argument -

If the option did not have a value which could be casted to double

void setDblOption(fmt::CStringRef name, double value)

Set an AMPL option to a specified double value.

Parameters
• name -

Name of the option to be set

• value -

The double value the option must be set to

Exceptions
• std::invalid_argument -

if the option name is not valid

Optional<bool> getBoolOption(fmt::CStringRef name) const

Get the current value of the specified boolean option.

In AMPL, boolean options are represented as integer: 0 for false, 1 for true. If the option does not exist, the returned Optional object will not have value.

Return
Value of the option (boolean)
Parameters
• name -

Option name

Exceptions
• std::invalid_argument -

if the option name is not valid

• std::invalid_argument -

If the option did not have a value which could be casted to boolean

void setBoolOption(fmt::CStringRef name, bool value)

Set an AMPL option to a specified boolean value.

Note that in AMPL, boolean options are represented as integer: 0 for false, 1 for true

Parameters
• name -

Name of the option to be set

• value -

The boolean value the option must be set to

Exceptions
• std::invalid_argument -

if the option name is not valid

void read(fmt::CStringRef fileName)

Interprets the specified file (script or model or mixed) As a side effect, it invalidates all entities (as the passed file can contain any arbitrary command); the lists of entities will be re-populated lazily (at first access)

throws runtime_error In case the file does not exist.

Parameters
• fileName -

Full path to the file.

void readData(fmt::CStringRef fileName)

Interprets the specified file as an AMPL data file.

As a side effect, it invalidates all entities (as the passed file can contain any arbitrary command); the lists of entities will be re-populated lazily (at first access). After reading the file, the interpreter is put back to “model” mode.

Parameters
• fileName -

Full path to the file.

Exceptions
• std::runtime_error -

In case the file does not exist.

Variant getValue(fmt::CStringRef scalarExpression) const

Get a scalar value from the underlying AMPL interpreter, as a double or a string.

Return
A Variant which represent the value of the expression.
Parameters
• scalarExpression -

An AMPL expression which evaluates to a scalar value.

void setData(const DataFrame &df, fmt::CStringRef setName)

Assign the data in the dataframe to the AMPL entities with the names corresponding to the column names.

If setName is null, only the parameters value will be assigned.

Parameters
• df -

The dataframe containing the data to be assigned.

• setName -

The name of the set to which the indices values of the DataFrame are to be assigned.

Exceptions
• AMPLException -

If the data assignment procedure was not successful.

std::string toString() const

Get a string describing the object.

Returns the version of the API and either the version of the interpreter or the message “%AMPL is not running” if the interpreter is not running (e.g. due to unexpected internal error or to a call AMPL::close)

Return
A std::string that represents this object.

void readTable(fmt::CStringRef tableName)

Read the table corresponding to the specified name, equivalent to the AMPL statement:

Parameters
• tableName -

Name of the table to be read.

void writeTable(fmt::CStringRef tableName)

Write the table corresponding to the specified name, equivalent to the AMPL statement.

Parameters
• tableName -

Name of the table to be written.

void display(StringArgs amplExpressions)

Writes on the current OutputHandler the outcome of the AMPL statement.

where are the strings passed to the procedure.

e1...en


Parameters
• amplExpressions -

Expressions to be evaluated.

void display(EntityArgs entities)

Writes on the current OutputHandler the outcome of the AMPL statement.

where are the objects passed to the procedure.

e1...en

Parameters
• entities -

The entities to display

void show(EntityArgs entities)

Writes on the current OutputHandler the outcome of the AMPL statement.

where are the objects passed to the procedure.

e1...en

Parameters
• entities -

The entities to show

void expand(EntityArgs entities)

Writes on the current OutputHandler the outcome of the AMPL statement.

where are the objects passed to the procedure.

e1...en

Parameters
• entities -

The entities to expand

void setOutputHandler(OutputHandler *outputhandler)

Sets a new output handler.

Parameters
• outputhandler -

The function handling the AMPL output derived from interpreting user commands.

void setErrorHandler(ErrorHandler *errorhandler)

Sets a new error handler Note that the ownership of the error handling object remains of the caller.

Parameters
• errorhandler -

The object handling AMPL errors and warnings.

OutputHandler *getOutputHandler() const

Get the current output handler.

Return
A pointer to the current output handler.

ErrorHandler *getErrorHandler() const

Get the current error handler.

Return
A pointer to the current error handler.

EntityMap<Variable> getVariables() const

Get all the variables declared.

EntityMap<Constraint> getConstraints() const

Get all the constraints declared.

EntityMap<Objective> getObjectives() const

Get all the objectives declared.

EntityMap<Set> getSets() const

Get all the sets declared.

EntityMap<Parameter> getParameters() const

Get all the parameters declared.