YaoLang is a Julia compiler extension. It compiles a subset of Julia programs to quantum device. As a language aims to solve the two language problem, we want to provide our solution to the two language problem in quantum programming.

YaoLang extends the native Julia semantics via macros and interpret these extra semantics via custom interpreter based on Julia's own interpreter during Julia's own type inference stage then runs our own specific optimization passes after Julia compiler optimizes the classical parts.

The YaoLang project aims to:

  1. compiles native Julia program to quantum devices and quantum device simulators
  2. provide an infrastructure for quantum compilation related research.


Writing Hybrid Programs

One of the major goal of YaoLang is to represent hybrid programs, which means programs mixed with classical functions and quantum routines. This is something happens very frequently in practical quantum computation and all the actual program controls quantum devices can be seen as such a hybrid program.

In YaoLang, you can use ANY classical Julia program semantics, such as control flows, function calls, and even other Julia packages. It is fully compatible with native Julia code. The compiler will only check if the program is compatible with your target machine or not. Here is a QFT example written using classical control flow:

@device function qft(n::Int)
    1 => H
    for k in 2:n
        @ctrl k 1 => shift(2π / 2^k)

    if n > 1
        2:n => qft(n - 1)

We don't have a real quantum device that supports running YaoLang natively, but ideally we can. Given Julia itself is actually Just-Ahead-of-Time (JAOT) compiled, there will not be any latency issue when we actually controls the quantum device - YaoLang as a subset of JuliaLang is static itself. It requires one to write type-stable Julia program in most cases except for native Julia simulator backend.

QASM Support

You can call QASM code like other Julia FFIs - simple and elegant:

julia> circuit = qasm"""OPENQASM 2.0;
       include "";
       gate custom(lambda) a {
           u1(sin(lambda) + 1) a;
       // comment
       gate g a
           U(0,0,0) a;

       qreg q[4];
       creg c1[1];
       U(-1.0, pi/2+3, 3.0) q[2];
       CX q[1], q[2];
       custom(0.3) q[3];
       barrier q;
       h q[0];
       measure q[0] -> c1[0];
       if(c1==1) z q[2];
       u3(0.1 + 0.2, 0.2, 0.3) q[0];
##qasm#702 (generic routine with 1 methods)

julia> YaoLang.@echo circuit()
[ Info: executing 3 => Rz(-1.0)
[ Info: executing 3 => Ry(4.570796326794897)
[ Info: executing 3 => Rz(3.0)
[ Info: executing @ctrl 2 3 => X
[ Info: executing 4 => Rz(0)
[ Info: executing 4 => Ry(0)
[ Info: executing 4 => Rz(1.2955202066613396)
[ Info: executing @barrier 1:4
[ Info: executing 1 => H
[ Info: executing @measure 1
[ Info: executing 1 => Rz(0.30000000000000004)
[ Info: executing 1 => Ry(0.2)
[ Info: executing 1 => Rz(0.3)
(c1 = 0,)

this string literal @qasm_str (See string literal section of Julia documentation) will create a YaoLang routine for all the gate declaration and in the and creates a anoymous YaoLang routine for the toplevel QASM program.

Hybrid Program Optimization

YaoLang can optimize your hybrid program using both Julia and its custom compiler optimization pass. See optimization section for more details.