Language overview

Basic features

Feature	Example
Single-line comment	// This is a single line comment
Multi-line or intervening comment	/* This is a multiline comment, when more text is needed */
Convert integer to string	int2string(n)
Convert real to string	real2string(pi)
Convert boolean to string	bool2string(true)
Convert integer to real	int2real(n)
Convert real to integer	real2int(pi)
Terminate statements with semi-colon ;	printLn("Hello, world");

Printing and debugging

Feature	Example
Print something (real/integer/string) to debug	debug(something)
Printing and adding a new line	printLn(text);
Printing without adding a new line	print(text);
Print only an empty line	print("\n");

• All types must be converted to string before printing, unless using debug.
• For reals, only a few decimals will be retained when printed.

Functions

Feature	Example
Creating a function without a return	function flip(datapoint: Bool, probability: Real) { observe datapoint ~ Bernoulli(probability); };
Creating a function with a return	function flip(datapoint: Bool, probability: Real) => Real { observe datapoint ~ Bernoulli(probability); return probability; }
Conditional (if/else statement)	if (condition) { execute_when_true } else { execute_when_false }
Loop (for statement)	for i in 1 to n { ... }

• A function needs to specify its input parameters and their types.
• A function can have zero input parameters.
• A function needs to specify its return parameters and their types, if there is a return.
• A function may return nothing via return; or have no return statement at all.
• The return statement may appear within a for or an if.
• The only effects that a function can have are changing the likelihood, printing things, and returning an output.
• Looping allows for limited side-effects.

Working with scalar data

Feature	Example
The basic datatypes are integer, real, boolean, and string	42 (integer), 3.14 (real), true or false (boolean), "hello" (string)
Arithmetic on numbers	2.0 + 3.0, 2.0 - 3.0, 2.0 * 3.0, 2.0 / 3.0
Comparisons (return booleans)	2.0 < 3.0, 2.0 <= 3.0, 2.0 == 3.0, 2.0 >= 3.0, 2.0 > 3.0
Boolean logic, AND (true if both are true)	a && b
Boolean logic, OR (true if one is true)	a || b
Boolean logic, NOT (negates)	!a
Explicit numeric conversion before comparing	Real(n) > 5.0

Working with sequences and matrices

Feature	Example
Joining with paste0 (no separator)	paste0(["espresso", "macchiato"]) returns "espressomacchiato"
Joining with paste (defining separator)	paste(["espresso", "macchiato"], " ") returns "espresso macchiato"
Creating a new sequence	let testSeq = [1, 2]; will create a sequence of integers with length 2
Getting the i-th element of the sequence (sequences are 1-indexed)	testSeq[1]
Getting the length of a sequence	length(testSeq)
Applying a function to each element of a sequence	let strSeq = apply(bool2string, boolSeq);
Creating a new matrix	let testMat = mtxCreate(2, 2, [1, 2, 3, 4]); creates a matrix with dimensions 2x2

• The basic datatypes are integer, real, boolean, and string. Each of them can be converted to a sequence by [].
• It is not possible to change individual elements. For example: let testSeq[1] = x; // compiler error!

Probabilistic programming

Feature	Example
A model function is needed by every program. It specifies the probabilistic model.	model function coinModel(coinflips: Bool[]) => Real[] { // specify prior // condition the likelihood on data // return posterior };
Use the assume keyword to sample form a prior distribution (here using the Beta distribution)	assume p ~ Beta(a, b);
Use the observe keyword to condition the likelihood on observed data (here using the Beta distribution)	observe data ~ Beta(a, b);
To manipulate the likelihood directly use weight	weight(lik);
or logWeight on the logarithmic scale	logWeight(lik);
The posterior is the returned value of the model function over one or more parameters	return p;