TimeLike Glossary |
Top Previous |
|
This glossary is intended to help current and prospective TimeLike users gain a better understanding of how TimeLike works, and how it can be used. We have tried to provide definitions for all unfamiliar or specially defined terms likely to be encountered` in the course of using the TimeLike System Editor and/or working with TimeLike at the source code level; if you find any omissions, please email us at support@timelikesystems.com. For the convenience of those less familiar with object-oriented programming (OOP) and C++, we have included some of that terminology as well; for a more thorough treatment see The C++ Programming Language by Bjarne Stroustrup. Among the terms defined are a number of C++ classes, functions, and operators; all C++ source code appears in a fixed font to distinguish it from ordinary text. Class names and operators are indicated by parenthetical comments, while function names are indicated by empty parentheses. Only the most commonly encountered classes and functions are documented here; for complete source code documentation see the TimeLike Programmer’s Manual. All terms appearing in italics in definitions are themselves defined in the glossary.
Note: Many of the definitions in this glossary have been copied verbatim from the glossary originally prepared for an earlier version of the software. Some of these definitions still require updating to reflect recent changes to the software.
A system which contains no subsystems. Can be implemented in Matlab (M-System), Python (Py-System) or C++(C++-System).
A class from which another more specialized class is derived, which defines a portion of the interface and/or implementation of the derived class. See also derived class. An operator such as + or = which takes two arguments, which appear on either side of the operator. Apart from the syntax, a binary operator is exactly equivalent to a function taking two arguments. A binary operator can be defined either as a global function, or as a member function of its first (left side) argument.
(C++ intrinsic type) bool is a C++ intrinsic type used to represent Boolean (true/false) values.
One of the most widely used object-oriented programming languages (OOPL).
The C++ Programming Language, Bjarne Stroustrup, Addison-Wesley, ISBN 0-201-88954-4 and 0-201-70073-5. http://www.research.att.com/~bs/3rd.html
Generally has the .h file (or .h and .cpp file) where the system designer defines the behavior by overriding one or more of the four respondTo- methods, possibly adding logic to the constructor and/or destructor, and possibly adding additional data members and/or member functions. For system classes with simple behaviors the designer may choose to keep complete implementation may be in .h file, but for more complex behaviors it is good practice to put the logic in a separate .cpp file, to permit separate compilation.
One of the key concepts in object-oriented programming, a class is the most general kind of user-defined type, defining both the state information used by objects of the class (data members) and their behavior (member functions). Classes may be related to one another via inheritance relationships, wherein base classes define portions of the interface and/or implementation of derived classes.
Complex is a C++ class, part of the TimeLike support libraries, use to represent complex numbers. The memory layout exactly matches the way complex numbers are stored in Fortran, so that complex numbers and arrays can be passed between Fortran and C++. In the near future all occurrences of Complex in TimeLike will be replaced with complex<float>, which shares the same memory layout, and is part of the new ANSI standard for C++.
A system which contains one or more subsystems. Unlike an atomic system class, a composite system would normally not contain any user-defined code. Instead, the behavior of a composite system is defined by the behaviors of its subsystems, their parameter settings, and the way they are connected.
A “connection” in TimeLike refers to a relationship established between an object used to refer some other object, e.g. a subsystem input, and another object which serves as the referent for the first object, e.g. a subsystem output. Connections are represented graphically in the GUI as arrows from the referent objects to the reference objects. A C++ keyword related to the notion of read-only access. An object or reference may be designated as const, to make it clear that it should not be modified within the given context. A member function of a class may also be designated const, to indicate that calling that function will not result in a change in the object’s state, and therefore that that function may be called for a const object. A special member function of a class, used to create and initialize objects of that class.
A C++ source code file. .cpp file (for a system class) A C++ source code file sometimes used to store the functions defining the behavior of a system class, named the same as the system class, but with the extension ".cpp". A .cpp file is used only in cases where the .h file has been modified by hand, and only when the system class designer chooses to do so, in order to permit separate compilation. An object (or variable) that constitutes part of the state information for a given class. A class may have any number of data members, each with its own unique name, and each may be of any legal C++ type. Ordinary (non-static) data members are used to store data related to one specific object of the class, and are accessed using a ‘dotted" notation, as shown: myObject.myDataMember. Static data members are used to store data shared among all objects of the class, and are accessed using the class name followed by a double colon, as follows: MyClass::myStaticDataMember. A copy operation performed when the user saves a system or a hierarchy into another directory. In this case all local reusable components, all referenced images and input files are copied over to the target directory.
Default value (for input or parameter) A setting expression provided by the designer of a system class to be used to set the value of an input to that class in case that input is not connected and no other setting expression has been provided. The expression is evaluated in the scope of the containing class, so the only way to ensure that it will be valid is to use either no symbols at all, e.g. a numerical consant, or only symbols defined in the user-provided C++ code. Alternately, the designer may choose to use symbols in the default expression, and rely on the automatic error-checking mechanisms to inform the user about the undefined symbols, which he may then choose to make parameters of the containing system class. A class that is derived from another less specialized class, its base class, which defines a portion of the interface and/or implementation of the derived class. See also derived class. A special member function for a class, called when an object of that class is to be removed from memory, and typically used to perform any necessary cleanup, such as freeing dynamically allocated memory.
double is a C++ intrinsic type used to represent real-valued numbers, similar to float, but with greater precision and range. Events are used in modeling discrete transitions in a system’s state caused by its internal time dependence. Generally each system which undergoes such transitions schedule events for itself, and only for itself, using the method scheduleEvent(). Each call to scheduleEvent() results in a delayed call-back to the virtual method respondToScheduledEvent(), unless it is cancelled before, using cancelEvent(). Each event has a character string descriptor, used to distinguish between different types of state transitions, and a void* pointer which can be used to attach information pertaining to the event. For example in modeling a packet-based communications system the pointer might be used to attach the contents of a message packet. A data structure containing all pending events for all systems within a given virtual universe, in order of their scheduled execution times. The simulation progresses, and virtual time advances, as events are pulled off the front of the queue and executed; this happens within advanceTime(). Events can be placed on the queue using scheduleEvent(), and removed from the queue using cancelEvent().
An object used to identify a specific event, which can be used for the purpose of canceling that event. Each call to scheduleEvent() returns an EventId, and cancelEvent() takes an EventId as an argument. An input to a system or system class; the adjective "external" is sometimes used to make it clear we are referring to an input of the system, rather than an input of one of its subsystems. An output of a system or system class; the adjective "external" is sometimes used to make it clear we are referring to an output of the system, rather than an output of one of its subsystems. A loop that repeats the same sequence of comands until a specified condition is met. Typically a for-loop has a loop variable, and integer variable which starts at zero and increments with each iteration of the loop, and exit condition is met when that counter reaches a specified number.
float is a C++ intrinsic type used to represent real-valued numbers.
Grid is a C++ template class, part of the TimeLike support libraries, used to define a two-dimensional rectangular grid over a specific rectangular region on a plane. A Gird<float> can be used to define a real-valued function over such a region, e.g. light intensity, while a Grid<float> can be used to define a complex-valued function, such as a complex field. Grid is used extensively in WaveTrain, both internally, and at the user interface level, where parameters of type Grid can be used to specify things like reflectance maps and detector responsivities.
GridGeometry is a C++ class, part of the TimeLike support libraries, used to define the geometry of a two-dimensional rectangular grid over a specific rectangular region on a plane. GridGeometry is used to specify the geometry for instances of the template class Grid, which is used extensively in WaveTrain, to describe things like optical wavefronts, reflectance maps and detector responsivities. See header file. A C++ header file which contains the source code definition for a system class, named the same as the system class, but with the extension ".cpp". The .h file may be entirely automatically generated – generally the case for composite system classes – or it may be modified by hand – generally the case for atomic system classes. When a .h file has been hand-modified, and is later automatically regenerated, e.g. when the user modifys the interface in the visual editor then does a "save", the obsolete hand-modified version is automatically appended to the new generated version, for use in cutting and pasting. A hard real-time application is one in which strict and precise time constraints apply, and failure to meet those constraints would result in unacceptable consequences. A typical example would be an automated control loop for a weapons system.
A source code file, typically with the extension ".h", intended to be included into other source code files, both .h and .cpp. One common use is define constants used in multiple places; another is to define the interface for one or more classes and/or global functions, but often not the actual implementation, which instead is put into a separate .cpp file, so that it may be compiled separately. .html file (for a system class) A hypertext help file for the system class, named the same as the system class, but with the extension ".cpp". A simple version is generated automatically, listing inputs, outputs, parameters, and subsystems, and so forth; additional text can be added by hand as desired. See header file. A C++ precompiler directive, used to indicated that the source code from a specified header file should be included into that of the present file, at the site of the include statement, prior to compilation. This mechanism is used to introduce definitions for classes, functions, and/or constants to be used later in the present file. One of the key concepts of object-oriented programming (OOP), inheritance is the mechanism by which one class, a derived class, can derive part of its interface and/or implementation from another, its base class. The derived class inherits all the data members and member functions of its base class, and may add its own. Also, a base class may define virtual methods, used to provide a common external interface to all derived classes, while permitting their specific behaviors to differ. A part of the external interface of a system or system class, used to define how a system’s environment can act upon the system. An input is a proxy object, an object which "stands for" some other object, typically an output from some other system. An input provides only read-only access to the referent object. This relationship is established by connecting the input to the output, using either an API function "connect" or calling Inputs' method "connect" if writing C++ code. Each input must be connected to exactly one output, unless a default value has been provided, in which case it need not be connected. A system may have any number of inputs, and each has a name (a character string, which must be a legal C++ variable name) and a type (any C++ type, including user-defined types). A system is automatically notified each time any of its inputs may be about to undergo a discrete state transition, such as a change of value, via a call to the virtual method respondToInputWarning(), and receives a second notice after the transition has occurred via a call to the virtual method respondToChangedInputs(). Also, whenever a system tries to access the value of one of its inputs, the system with the connected output is notified via a call to the virtual method respondToOutputRequest(). int is a C++ intrinsic type used to represent integer-valued numbers. Generally, the term "instance" refers to a specific object of the class, resident in memory. Sometimes the term is also used to refer to a specific variable of the class type, such as a data member of a class, even when that same variable may represent multiple actual objects. For each (non-static) data member of a class, there are as many actual objects as there were instances of the class. We sometimes refer to the latter case as a usage instance, to distinguish it from the first. A Library in TimeLike is a collection of Reusable Components which are, essentially, black boxes that users may assemble together to create large models. A library may contain Atomic Reusable Components (aka Simple Systems) and Composite Reusable Components (aka Composite Systems).
A directory used to store TimeLike systems intended for reuse. TimeLike defines two locations for libraries: •%TIMELIKE_DIR%\libraries - this is where all the libraries provided as a part of installation or provided as separate products by TimeLike Systems are stored. %TIMELIKE_DIR% is an Environment Variable that points to the location where TimeLike was installed. This environment variable is created by the installer and is typically set to "C:\Program Files\TimeLike Systems\TimeLike".
•%TIMELIKE_USER_DIR%\libraries - this is where all libraries created by the user are stored. %TIMELIKE_USER_DIR% is an Environment Variable that points to the location where all user-related TimeLike-related files are stored. This environment variable is created by the installer and at the time of installation is set to %USERPROFILE%\timelike.
any other directory where timelike systems may be stored are Work Directories.
The strategy of waiting until the results of a computation are needed before performing the computation, as opposed to, for example, computing it as soon as new inputs value become available. This strategy is especially useful when the inputs may change more often than the results are needed, and each computation costs appreciable cpu-time. Legal variable names in C++ may be up to 255 characters long, consisting of letters, numerals, and underscores, and must begin with a letter or underscore. Variable names cannot be the same as C++ keywords, or the same as other variable names in the namespace. C++ is case-sensitive. An integer variable, used in conjunction with a for-loop, which starts at zero and increments with each iteration of the loop, and exit condition is met when that counter reaches a specified number. Loop variables are a special type of run variable, used to define a set of nested for-loops within a run set, while the actual simulation runs would take place inside the innermost loop. To add a for-loop in the Run Set Edit window, one simply adds a new run variable, say iloop, and sets its type to int and its value to $loop(niter), where niter specifies the number of loop iterations. In the generated source code this would appear as follows, where we have set niter equal to 3: for (iloop=0; iloop<3; iloop++){…} Generally the reason one would wish to set up a for-loop would be to change the value of one or more system parameters with each iteration of the loop. This is done by setting the value of the parameter to an expression of the form [iloop]:{v1,v2,v3…}, where v1,v2,v3… are the desired values for the parameter, one for each loop iteration, or of the form [iloop]: expression, where expression is any legal C++ expression, generally involving the loop variable. If multiple loop variables are used, the for-loops are nested, with the loop corresponding to the first loop variable outermost. To make a parameter dependent on two or more loop variables, one simply lists them inside the square brackets, separated by commas, as follows: [iloop,jloop,kloop] The total number of runs in the run set is given by the product of the numbers of iterations for all loop variables.
TimeLike provides a robust mechanism for incorporating routines written in Matlab's m-file language as atomic systems. Such systems are called m-systems.
A function associated with a particular class that defines part of the interface and behavior of that class. Ordinary (non-static) member functions define operations that effect one specific object of the class, and are invoked using a ‘dotted" notation, as shown: myObject.myMemberFunction(). Static member functions define operations not necessarily related to any specific object of the class, such as operations that involve static data members of the class, and are invoked using the class name followed by a double colon, as follows: MyClass::myStaticMemberFunction() A synonym for member function. Similar to an ordinary input, except that it allows multiple outputs connected to it. A typical example of multi-input is an input of Sum system in BasicMath Component Library. Similar to an ordinary output, except it can be connected to a multi-input. Similar to an ordinary input, except that it provides full read-write access to the referent object. A mutableInput cannot be connected to an ordinary output; instead they can only be connected to a mutableOutput or an object, both of which support read-write access. See also Input, Input (the C++ class). Similar to an ordinary output, except it permits full read-write access to its referencors. A mutableOutput may be connected to ordinary inputs, mutableInputs, or both. See also Output, Output (the C++ class).
Is an arbitrary text that a user may add to a block diagram to annotate the system he is editing. The user may change the font, color and other visual properties of a note. now() is a member function of class System which returns the current virtual time. It takes no arguments. One of the key concepts of object-oriented programming (OOP), an OOP object is analogous to a real world object, in that objects can be created and destroyed, and each object has its own state information, which can evolve over time, in accordance with specified behavioral rules. In C++ each object is an instance of some class, and it is the class that defines the nature of the state information, in the form of data members, and behavior, in the form of member functions. Part of the TimeLike kernel library, Object is a template class which defines a simple kind of proxy object, a wrapper for an object of arbitrary type which can serve as the referent for any number of system inputs. Any action which would modify the state/value of the underlying object is automatically trapped, and any systems with attached inputs are called with respondToInputWarning(). A programming paradigm based on the concepts of objects and classes, as opposed to procedures. Many developers find that object-oriented programming, or OOP, conforms much more closely to the way they naturally think about problem solving and design.
Object-oriented software composition Object-oriented software composition, component-based software, is an approach for extending traditional object-oriented programming techniques for very large projects, and/or projects which involve the combined use of many separately designed and developed software components. The essential idea is to adopt an architecture and methodology which allows the programmer to treat software components more or less like “black boxes”, which he can connect together to accomplish the desired purpose without detailed knowledge of their inner workings. An acronym for object-oriented programming. A part of the external interface of a system or system class, used to define how a system can act upon its environment. An output is essentially a wrapper for an underlying object, which can be of any C++ type, where the output can be used as the referent for one or more inputs of that same type. This relationship is established by connecting the input to the output either via the GUI or by using API directly. A system may have any number of outputs, and each has a name (a character string, which must be a legal C++ variable name) and a type (any C++ type, including user-defined types). A system is automatically notified each time any of its outputs is about to be accessed through a connected input, via a call to the virtual method respondToOutputRequest. Also, whenever a system attempts to modify an output, e.g. in an assignment statement, all systems with connected inputs are automatically warned, via calls to the virtual method respondToInputWarning.
Part of the TimeLike kernel library, Output is a C++ template class used to represent system outputs. Parameters are variables used to fine-tune the behavior of each instance of a particular system class, and/or to define its initial conditions; at the source code level, parameters appear as constructor arguments. A system class may have any number of parameters, and each has a name (a character string, which must be a legal C++ variable name) and a type (any C++ type, including user-defined types). When an instance of a system class appears as a subsystem within another system class, its parameter values are specified by setting expressions defined in terms of the parameters of the containing system class. When a system class serves as the basis of a run set, its parameter values are specified by setting expressions defined in terms of the run variables defined for the run set. Typically the run variables would include one or more loop variables, used to set up multiple simulation runs, using different parameter values. A data member of class System, a list of all events the system currently has pending, i.e. scheduled, but not yet executed. For a system that schedules events, this list constitutes an important part of the system’s state, and in some cases it may be necessary or useful to make use of the pending events list within the system’s respondTo- methods. A simple object that records the address of some other object, of some specified type. Pointers can be used to access the pointed-to object by means of the dereference operator (*) and the member dereference operator (->). An object used to represent another object, so that actions performed on the proxy object are actually forwarded to the object it represents. A proxy object is similar to a reference, but unlike a reference, it is a first class object, defined by a class, with its own data members and member functions. A typical application is to trap attempts to access the represented object, and then perform some appropriate action. Ideally, one should be able to substitute a proxy object for the represented object in any valid expression and get exactly the same result, but unfortunately in C++ this is not possible, because of subtleties related to the rules for automatic type conversion and the member access (.) operator. To get around this, the designers of proxy classes typically define a member function or operator which returns a reference to the underlying object, which can then be used in any expression. In TimeLike, the proxy classes Input, Output, and SaveVariable provide both, via the dereference operator (*), implicit or explicit type conversion, and the member functions value() (const access) and object() (non-const access, Output and SaveVariable only).
TimeLike provides a robust mechanism for incorporating routines written in Python programming language as atomic systems. Such systems are called py-systems. In C++ a reference is similar to a pointer, except that, unlike pointers, references are used syntactically just like the represented object, and are not themselves considered first class objects, in that once a reference has been defined no further action can be performed on it, since all actions are forwarded to the referred-to object.
RectangularRegion is a C++ class, part of the TimeLike support libraries, used to define a two-dimensional rectangular region on a plane. RectangularRegion is closely related to the class GridGeometry, used to specify the geometry for instances of the template class Grid, which is used extensively in WaveTrain, to describe things like optical wavefronts, reflectance maps and detector responsivities. A set of four virtual methods of the class System, used to define the behaviors of all system classes. The four methods – respondToChangedInputs(), respondToInputWarning(), respondToOutputRequest(), respondToScheduledEvent() – define the system’s responses to changes to its inputs, requests for its outputs, and scheduled events. One of the four respondTo- methods used to define the behavior of a system class, respondToChangedInputs() is called whenever one or more of the system’s inputs have changed, after they have assumed their new values, but before virtual time is allowed to advance. It has no arguments. Note that "Inputs" is plural, reflecting the fact that the values of two or more inputs may change at the same virtual time; to find out which ones have changed, the programmer can use hasChanged(), a member function of the class Input. One of the four respondTo- methods used to define the behavior of a system class, respondToOutputRequest() is called whenever any one of the system’s outputs is about to be accessed. There is one argument, which identifies the output. This gives the programmer the option of adopting a strategy of lazy evaluation, i.e. computing the value of the output only when it is requested. One reason one might choose that strategy is efficiency, especially in the case of outputs whose values may change much more often than they are accessed, and where each output evaluation takes a significant amount of CPU time. Another reason would be if the output models a quantity that may change continuously in time, such as the instantaneous position of a damped harmonic oscillator. By waiting for the output request, we can compute the position for the precisely the right instant in virtual time. One of the four respondTo- methods used to define the behavior of a system class, respondToScheduledEvent() is called whenever an event scheduled by the system reaches the top of the event queue. See simulation run. A text file used by the TimeLike visual editor to store the definition of a run set, named the same as the system class, but with the extension ".cpp". .run files are user-readable, containing the same information displayed in the edit run set window, and can be useful as a form of documentation. A set of simulation runs, all based on the same system class, and generally executed as part of the same main program, but differing in the values used for one or more system parameters. A run set is typically used to perform a specific parameter study based on the given system model, e.g. how does such-and-such performance measure change when one turns some particular knob? For system models that include stochastic effects one of the parameters varied would often be a random number seed, for the purpose Monte Carlo averaging. Run sets can be created and modified using the run set edit window, where the user specifies which parameters are to be varied and what values they should be set to, what values the other parameters should be set to, the simulation stop time, what outputs to record, and so forth. The information pertaining to a run set is stored on disk in a .tlr file, kept in the same directory as the system on which the run set is based. See also simulation run, run set edit window, run variable, loop variable. One of the two main windows in the TimeLike visual editor, used to create and modify run sets. To see the window layout, click here. See also run set, simulation run, run variable and loop variable. A variable used within a run set for specifying parameter values and/or setting up for-loops over parameters. One application would be to define quantities to be used in multiple setting expressions; another would be to store expensive-to-compute results which can be reused in multiple simulation runs. There is one special type of run variable, called a loop variable, used for setting up for-loops over parameters. A method of class System used to schedule events, used to trigger discrete changes in a system’s state related to its own internal time dependence. ScheduleEvent takes three arguments: the delay (in virtual time) after which the event should occur, a character string descriptor used to distinguish between different types of events, and a void* pointer which can be used to attach information pertaining to the event. ScheduleEvent() returns an EventId, which can be used to cancel the event should that become necessary. A global function provided to TimeLike users which allows the generation of a deterministic sequence of random number seeds from a single random number seed. TimeLike is often used for Monte Carlo runs where it is necessary to change the random number seed for each sample. However, it would be inconvenient for the user to explicitly specify the seed to be used for every single case. seedSequence provides a mechanism to set the seed in a deterministic way depending on a single user-specified seed and iteration number. The form of seedSequence is: int seedSequence(const int seed, const int index) {return (seed + (2*index));} seedSequence is usually used in conjunction with a loop variable. Suppose we have loop variable irand. We can run a Monte Carlo study by setting the random number seed for some entity (such as the atmospheric path) to: [irand]:seedSequence(-123456789,irand) On the first iteration the seed -123456789 will be used, on the next -123456787, and so on. This sceme works well in seeding random number generators for which it is unlikely that two numbers, even if they are almost equal, are unlikely to be close to each other in a given random sequence. Some random number generators (in particular, the one in the Intel performance library) do not have this feature so we do not use it. One might ask why we don't use a single random number seed to seed all random number processes throughout the simulation. This has the undesirable feature that it would be difficult to hold one random process constant while allowing another to vary. Also, the actual sequences that were generated would depend on how many numbers were drawn and if we changed the number of numbers slightly we could get significantly different results.
A Python expression used to specify a value for a parameter, input, or output, which may be as simple as a numerical constant, e.g. "5.0", or as complicated as needed. A system class may specify default values for its own parameters, inputs, and outputs, and may set the values for its subsystems’ parameters and inputs, which overrides any defaults specified in the subsystem’s system classes. Also, run sets use setting expressions to specify the parameter values for the system model used. Setting expressions are evaluated in the scope where the value is needed, and must therefore be valid Python expressions within that scope. This means that the setting expressions for subsystem parameters and inputs and external outputs can involve only the parameters of the containing system, or symbols defined in the user-provided C++ code for the class. Default setting expressions for parameters and inputs are likewise evaluated within the containing class, not the class where they were defined. The only way to ensure that a default expression will be valid expressions is to use either no symbols at all, or only symbols defined in the user-provided C++ code. Alternately, one can use symbols in the default expression, and rely on the automatic error-checking mechanisms to inform the user about the undefined symbols, which he may then choose to make parameters of the containing system class. A copy operation performed on a usage instance of a class, that involves copying only the instance information, not the class itself. For example whenever the user drags-and-drops or copies and pastes a reusable component from a library to the system he is editing, a shallow copy is performed, creating a new usage instance of the existing reusable component. See also deep copy. A single execution of a simulation, generally occurring within the innermost for-loop of a run set main program. A simulation run generally consists of (1) the construction of a new virtual universe, (2) a call to the constructor of the system class on which the run set is based, which each parameter value set as specified in the run set, (3) set up of output recording as specified in the run set, and (4) a call to advanceTime, used to advance virtual time to the stop time specified either in the run set, or on the command line. Simulation runs can also be used in other ways, such as in conjunction with a nonlinear minimization algorithm for design optimization or fitting a model to data, or as part of an interactive training environment. A soft real-time application is one in which time constraints apply, but they are not terribly strict, somewhat imprecise, and failure to meet those constraints would not result in unacceptable consequences. A typical example would be an interactive training environment for operators of some complex system. A system which is used as part of a larger system or system class. Every subsystem is an instance of some system class, with its parameter values defined by setting expressions defined in terms of the parameters of the containing system class, and its inputs and outputs connected to other subsystems and/or external inputs and outputs as specified by the containing system class. In TimeLike a system is a C++ object, an instance of a system class, a C++ class derived from the base class System. The class System is designed to be a software representation of the abstract notion of a system, as the term is used in science and engineering, where a system can be just about anything, provided that it has well-defined inside – the system itself - and outside – its environment. Following the methods of systems engineering, any system can be described in terms of its inputs, outputs, internal state, and behavior, where its behavior consists of its response to changes in its inputs and its internal time dependence. The class System is designed for use as a base class for a wide variety of more specialized classes, each representing a specific type of system, with its own particular inputs, outputs, and behavior. These classes, called system classes, can be created and modified using the TimeLike visual editor. Some, called composite system classes, are made up of simpler systems, called subsystems, connected together. Others, called atomic system classes, contain no other systems. The behaviors of composite systems are defined by the behaviors of their subsystems; the behaviors of atomic systems are defined by source code. All system classes, both atomic and composite, are generally parameterized, meaning that their behaviors and/or initial conditions are defined in terms of a number of parameters, passed in as constructor arguments. A TimeLike system is simply one instance of some system class, possibly one of many. Any changes to a system class will affect all its instances, unless the user elects to rename the modified class, in which case he can choose which instances will be affected by the changes, and which will not. Part of the TimeLike kernel library, the class System, which serves as the base class for all TimeLike system classes,, is designed to be a software representation of the abstract notion of a system, as the term is used in science and engineering. System provides a customizable interface mechanism based on the use of inputs and outputs, instances of the template classes Input<T> and Output<T>, where inputs are used to define the means by which a system’s environment can act upon the system, and outputs are used to define the means by which a system can act upon its environment. System has four virtual methods, collectively called the respondTo- methods, which are used to define the behavior of each system class: respondToChangedInputs(), respondToInputWarning(), respondToOutputRequest(), respondToScheduledEvent(). The default behavior for each is to do nothing, and for composite system classes that generally suffices, while atomic system classes will generally override one or more of the four. For more details consult the TimeLike Programmer’s Manual. One of the two main window in the TimeLike visual editor, the system class edit window is used to edit or view system classes, represented graphically, as block diagrams. The user may open any number of edit windows, each of which shows one system class at a time. The user can cut and paste subsystems, inputs, and outputs between windows, and can rearrange subsystems, make and break connections, and so forth, all by direct manipulation. He can also navigate through a complete hierarchical system model, double-clicking on a subsystem to descend into the corresponding system class, double-clicking on whitespace to ascend to the level above. If two or more edit windows happen to be looking at the same system class, all actions performed in one will also be reflected in the other. There is an undo capability, and automated error-trapping. A status button at the bottom right turns red whenever any errors are detected; clicking on it brings up a detailed report. Templates are the C++ version of generic types, used to define classes and functions that can be used with typed objects, where the type matters, but where at least part of the desired behavior can be defined independent of the specific type. Typical examples include multi-object container classes, such as lists, and proxy objects. A template class or function is defined in terms of one or more template arguments, used to represent the unspecified type or types. An argument used in the definition of a template class or function to represent the unspecified type or types, and/or to convey other information needed at compile time, such as array dimensions. In the definition of a template class or function, the template arguments appear at the top, immediately following the keyword template, and enclosed in angle brackets: template <class T> Input {…}; In usage, the template arguments appear immediately after the class or function name, again enclosed in angle brackets. Input<float> u; Within the definition of the class or function, a template argument representing a class is used in place of the class name in declarations, type casts and so forth. To refer to an instance of a template class where a template argument is unspecified we typically write "T" in place of the template argument, as in "Input<T>". A class defined in terms of one or more template arguments. Template function A function defined in terms of one or more template arguments.
A comprehensive user’s manual guide for all the basic activities involved in the using TimeLike at the source code level. Our intention is for the manual to eventually cover every detail of every feature, but as TimeLike is still in active development it will be some time before that is true. TimeLike stores the results of simulations in .h5 files - in the HDF5 format. Please see the www.hdfgroup.org for details on HDF5 format. A comprehensive user’s manual for all the basic activities involved in the use of TimeLike, from installing the software to building system models and performing parameter studies in the TimeLike visual editor, then looking at the simulation outputs in Matlab. It does not cover use of TimeLike at the source code level. Our intention is for the manual to eventually cover every detail of every feature, but as TimeLike is still in active development it will be some time before that is true. The TimeLike graphical user interface (GUI), called the "visual editor" because it is a visual programming environment, used to create and modify TimeLike system classes and run sets. Part of the TimeLike kernel library, the Universe is a C++ class designed to represent a virtual universe, within which we can construct a system model, then cause virtual time to advance, independent of any other virtual universes and system models within the same process. Often only a single virtual universe is in use at any one time – that’s the case in a typical run set – but some applications require multiple virtual universes, such as those that involve models within models. A variable representing an object of a given class, which does not necessarily refer to a single specific object in memory (i.e. an "instance" in the ordinary sense), because which object in memory it refers to depends upon the usage context. For example, if class A has a non-static data member b of class B, there would be one b for each instance of A – possibly many, or zero - so b would be considered a usage instance of B, but not an instance in the ordinary sense. Object-oriented programming languages such as C++ allow programmers to define their own data types, which can then be used just as the intrinsic types defined by the language are. The term user-defined type is almost synonymous with the termclass, but in C++ there is also a simpler form called a struct, which can be thought of as a class without any member functions and with all data members publicly accessible.
Vector<T> (C++ template class) Vector is a C++ template class, part of the TimeLike support libraries, used to describe an ordered set of similar objects. Vector is used extensively in WaveTrain, both internally and at the user interface level, where parameters of type Grid can be used to specify things like phase screen locations and strengths.. A member function of a class designed for use as a base class, used to define a common interface to all derived classes, while allowing the behaviors of derived classes to differ. The base class may define a default behavior (implementation of the function), which the designer of a derived class may choose to override. If the base class does not provide a default the function is called a "pure virtual function", and each derived class must provide its own implementation. See virtual member function. Virtual time plays the same role in a computer simulation that real time does in the real world; when a real world system has been correctly modeled, the behavior of the model unfolds as virtual time advances in a manner closely corresponding to the behavior of the actual system in real time. Unlike real time, virtual time is subject to our complete control, so we can ensure that all simulation events occur at exactly the correct times and in the correct sequence, deterministic and repeatable. Depending on the application, the model, and the computer system, virtual time may advance much more slowly than real time, or much more quickly. In some applications, such as interactive training environments, it is desirable for virtual time to advance at the same pace as real time; given sufficient computing power this is generally easy to do in an approximate sense, but impossible to do with exactitude for an arbitrary computer model. For this reason standard virtual time simulation techniques can be applied directly to soft real-time applications, but not necessarily for hard real-time applications. Simulation techniques can be useful in hard real-time applications, but it is then necessary to either restrict the kinds of software components used (e.g fixed time algorithms), or else adopt an architecture that partly decouples the simulation from the control loop proper. A C++ keyword used in place of the return type to indicate that a function does not have a return value, and also used in place of the referent type when declaring pointers to objects of unspecified type. Any directory used to store TimeLike system classes and run sets, and not specifically designated to be a library directory. The distinction between the two is that in a work directory it is assumed that the user is trying to use TimeLike as a tool, to solve a problem, whereas a library directory is used to extend the capabilities of TimeLike, by providing a set of reusable and mutually compatible components. In a work directory the user typically makes frequent changes to his systems, as he develops and refines his overall system model or models, so it would not make sense to use those same system classes as subsystems in unrelated models. In contrast, reusable systems in a library directory are intended for reuse and designed for it, and may be used as subsystems in any number of work directories. |