GPAClib

What is GPAClib

GPAClib is a header-only library implemented in C++ for defining and simulating analog circuits (GPAC = General Purpose Analog Computer). The library comes with a parser that allows to define circuit in a format that is specifed below.

Prerequisites

In order to use GPAClib, you need to have:

• a compiler with support for C++14
• Boost library (more specifically: System, Program Options, Iostreams, Odeint and Spirit)
• OpenMP

If you want to compile the example code provided, you will need CMake version >= 3.1.

How to use

Since GPAClib is header-only, the usage is very simple: one just has to include the main header file GPAC.hpp. The header file GPACparser.hpp provides the tool to load a circuit from a specification file. A simple example of usage of GPAClib (loading a circuit and simulating it) is given in file GPACsim.cpp, which can be compiled by entering the following command while being in the main directory:

mkdir -p build
cd build
cmake ..
make

It creates a program called GPACsim that takes a specification file name as argument and simulates the corresponding circuit. Execute GPACsim --help for more information about the options.

You can generate the documentation of GPAClib using Doxygen:

cd doc
doxygen

The documentation is generated in subdirectories html (open index.html with your favorite web browser) and latex.

Circuit specification format

It is possible to comment a circuit specification file by inserting a # symbol: all remaining symbols on the line are then ignored.

The first way of defining a circuit is by specifying its gates:

Circuit <circuit_name>:
    <gate_name>: ....
    <gate_name>: ....
    <gate_name>: ....
;

Circuit names cannot contain spaces and should not be named as builtin circuits provided in the following. Moreover, t is a reserved name and circuit names should not be confusable with floating point numbers.

The gate specification is the following:

• for constant gate: <value>
• for product gates: <operand1> * <operand2>
• for addition gate: <operand1> + <operand2>
• for integration gates: int <integrated> d(<variable>) | <initial_value>
• for a copy of a previously defined circuit: <identifier>

Operands are names of gates defined (before or after) in the same circuit or t. The last type allows to import a copy of a previously defined circuit and to use the gate as the output of this circuit. Gate names cannot be empty, cannot start with underscore and t is reserved. A value is either an integer or a floating point number, it can be negative. The last gate entered is the output gate.

The second way to define a circuit is by combination of previously defined circuits:

Circuit <circuit_name> = <expression>;

where expressions are defined with the following grammar:

<op> ::= + | * | @ | - | /
<expr> ::= <value> | <identifier> | <identifier>[<integer>] | (<expr> <op> <expr>) 
	     | (<expr> <op> <expr>)[<integer>] | (int <expr> d(<expr>) | <value>) 
		 | max(<expr>,<expr>) | select(<value>,<value>,<value>,<value>)
		 | deriv(<expr>) | deriv(<expr>, <integer>) 

The @ operator corresponds to composition of circuits. An identifier is the name of a previously defined circuit, or the name of a builtin circuits, or t. An integer is non-negative and a value is a floating point number or an integer (no sign restriction). The [] operator is for iterating an expression or a circuit, e.g. C[5] represents circuit C iterated 5 times. select(a,b,x,y) corresponds to a circuit computing a function that has value x for t <= a and y for t >= b. Warning: always leave a space between the - operator and values.

List of builtin circuits:

• Exp: exponential
• Sin, Cos, Tan
• Tanh
• Sqrt
• Arctan
• Inverse: function 1/(1+t)
• Id: identity (same as t)
• L2: approximation of a "toggle" (close to 0 for t < 1/4 and close to 1 for t > 3/4)
• Abs: approximation of the absolute value function
• Round: approximation of rounding function, contracts any real which is not an half integer to the closest integer
• Mod10: approximation of the mod 10 function

Examples of circuit files are given in the circuits folder.