Program Synthesis is Possible

May 9, 2018

Program synthesis is not only a hip session title at programming languages conferences. It’s also a broadly applicable technique that people from many walks of computer-science life can use. But it can seem like magic: automatically generating programs from specifications sounds like it might require a PhD in formal methods. Wisconsin’s Aws Albarghouthi wrote a wonderful primer on synthesis last year that helps demystify the basic techniques with example code. Here, we’ll expand on Aws’s primer and build a tiny but complete-ish synthesis engine from scratch.

You can follow along with my Python code or start from an empty buffer.

Z3 is Amazing

We won’t quite start from scratch—we’ll start with Z3 and its Python bindings. Z3 is a satisfiability modulo theories (SMT) solver, which is like a SAT solver with theories that let you express constraints involving integers, bit vectors, floating point numbers, and what have you. We’ll use Z3’s Python bindings. On a Mac, you can install everything from Homebrew:

$ brew install z3 --with-python

Let’s try it out:

import z3

To use Z3, we’ll write a logical formula over some variables and then solve them to get a model, which is a valuation of the variables that makes the formula true. Here’s one formula, for example:

formula = (z3.Int('x') / 7 == 6)

The z3.Int call introduces a Z3 variable. Running this line of Python doesn’t actually do any division or equality checking; instead, the Z3 library overloads Python’s / and == operators on its variables to produce a proposition. So formula here is a logical proposition of one free integer variable, $x$, that says that $x \div 7 = 6$.

Let’s solve formula. We’ll use a little function called solve to invoke Z3:

def solve(phi):
    s = z3.Solver()
    return s.model()

Z3’s solver interface is much more powerful than what we’re doing here, but this is all we’ll need to get the model for a single problem:


On my machine, I get:

[x = 43]

which is admittedly a little disappointing, but at least it’s true: using integer division, $43 \div 7 = 6$.

Z3 also has a theory of bit vectors, as opposed to unbounded integers, which supports shifting and whatnot:

y = z3.BitVec('y', 8)
print(solve(y << 3 == 40))

There are even logical quantifiers:

z = z3.Int('z')
n = z3.Int('n')
print(solve(z3.ForAll([z], z * n == z)))

Truly, Z3 is amazing. But we haven’t quite synthesized a program yet.


In the Sketch spirit, we’ll start by synthesizing holes to make programs equivalent. Here’s the scenario: you have a slow version of a program you’re happy with; that’s your specification. You can sort of imagine how to write a faster version, but a few of the hard parts elude you. The synthesis engine’s job will be fill in those details so that the two programs are equivalent on every input.

Take Aws’s little example: you have the “slow” expression x * 2, and you know that there’s a “faster” version to be had that can be written x << ?? for some value of ??. Let’s ask Z3 what to write there:

x = z3.BitVec('x', 8)
slow_expr = x * 2
h = z3.BitVec('h', 8)  # The hole, a.k.a. ??
fast_expr = x << h
goal = z3.ForAll([x], slow_expr == fast_expr)

Nice! We get the model [h = 1], which tells us that the two programs produce the same result for every byte x when we left-shift by 1. That’s (a very simple case of) synthesis: we’ve generated a (subexpression of a) program that meets our specification.

Without a proper programming language, however, it doesn’t feel much like generating programs. We’ll fix that next.

A Tiny Language

Let’s conjure a programming language. We’ll need a parser; I choose Lark. Here’s my Lark grammar for a little language of arithmetic expressions, which I ripped off from the Lark examples and which I offer to you now for no charge:

?start: sum

?sum: term
  | sum "+" term        -> add
  | sum "-" term        -> sub

?term: item
  | term "*"  item      -> mul
  | term "/"  item      -> div
  | term ">>" item      -> shr
  | term "<<" item      -> shl

?item: NUMBER           -> num
  | "-" item            -> neg
  | CNAME               -> var
  | "(" start ")"

%import common.NUMBER
%import common.WS
%import common.CNAME
%ignore WS

You can write arithmetic and shift operations on literal numbers and variables. And there are parentheses! Lark parsers are easy to use:

import lark
parser = lark.Lark(GRAMMAR)
tree = parser.parse("(5 * (3 << x)) + y - 1")

As for any good language, you’ll want an interpreter. Here’s one that processes Lark parse trees and takes a function in as an argument to look up variables by their names:

def interp(tree, lookup):
    op =
    if op in ('add', 'sub', 'mul', 'div', 'shl', 'shr'):
        lhs = interp(tree.children[0], lookup)
        rhs = interp(tree.children[1], lookup)
        if op == 'add':
            return lhs + rhs
        elif op == 'sub':
            return lhs - rhs
        elif op == 'mul':
            return lhs * rhs
        elif op == 'div':
            return lhs / rhs
        elif op == 'shl':
            return lhs << rhs
        elif op == 'shr':
            return lhs >> rhs
    elif op == 'neg':
        sub = interp(tree.children[0], lookup)
        return -sub
    elif op == 'num':
        return int(tree.children[0])
    elif op == 'var':
        return lookup(tree.children[0])

As everybody already knows from their time in CS 6110, your interpreter is just an embodiment of your language’s big-step operational semantics. It works:

env = {'x': 2, 'y': -17}
answer = interp(tree, lambda v: env[v])

Nifty, but there’s no magic here yet. Let’s add the magic.

From Interpreter to Constraint Generator

The key ingredient we’ll need is a translation from our source programs into Z3 constraint systems. Instead of computing actual numbers, we want to produce equivalent formulas. For this, Z3’s operator overloading is the raddest thing:

formula = interp(tree, lambda v: z3.BitVec(v, 8))

Incredibly, we get to reuse our interpreter as a constraint generator by just swapping out the variable-lookup function. Every Python + becomes a plus-constraint-generator, etc. In general, we’d want to convince ourselves of the adequacy of our translation, but reusing our interpreter code makes this particularly easy to believe. This similarity between interpreters and synthesizers is a big deal: it’s an insight that Emina Torlak’s Rosette exploits with great aplomb.

Finishing Synthesis

With formulas in hand, we’re almost there. Remember that we want to synthesize values for holes to make two programs equivalent, so we’ll need two Z3 expressions that share variables. I wrapped up an enhanced version of the constraint generator above in a function that also produces the variables involved:

expr1, vars1 = z3_expr(tree1)
expr2, vars2 = z3_expr(tree2, vars1)

And here’s my hack for allowing holes without changing the grammar: any variable name that starts with an “H” is a hole. So we can filter out the plain, non-hole variables:

plain_vars = {k: v for k, v in vars1.items()
              if not k.startswith('h')}

All we need now is a quantifier over equality:

goal = z3.ForAll(
    list(plain_vars.values()),  # For every valuation of variables...
    expr1 == expr2,  # ...the two expressions produce equal results.

Running solve(goal) gets a valuation for each hole. In my complete example, I’ve added some scaffolding to load programs from files and to pretty-print the expression with the holes substituted for their values. It expects two programs, the spec and the hole-ridden sketch, on two lines:

$ cat sketches/s2.txt
x * 10
x << h1 + x << h2

It absolutely works:

$ python3 < sketches/s2.txt
x * 10
(x << 3) + (x << 1)

Better Holes with Conditions

Our example so far can only synthesize constants, which is nice but unsatisfying. What if we want to synthesize a shifty equivalent to x * 9, for example? We might think of a sketch like x << ?? + ??, but there is no pair of literal numbers we can put into those holes to make it equivalent. How can we synthesize a wider variety of expressions, like x?

We can get this to work without fundamentally changing our synthesis strategy. We will, however, need to add conditions to our language. We’ll need to extend the parser with a ternary operator:

?start: sum
  | sum "?" sum ":" sum -> if

And I’ll add a very suspicious-looking case to our interpreter:

elif op == 'if':
    cond = interp(tree.children[0], lookup)
    true = interp(tree.children[1], lookup)
    false = interp(tree.children[2], lookup)
    return (cond != 0) * true + (cond == 0) * false

These funky multiplications are just a terrible alternative to Python’s built-in condition expression. I’m using this here instead of a straightforward true if cond else false because this works in both interpreter mode and in Z3 constraint generation mode and behaves the same way. I apologize for the illegible code but not for the convenience of a single implementation.

The trick is to use conditional holes to switch between expression forms. Here’s an implementation of the sketch we want above:

$ cat sketches/s4.txt
x * 9
x << (hb1 ? x : hn1) + (hb2 ? x : hn2)

Each ternary operator here represents a ?? hole in the sketch we wanted to write, x << ?? + ??. By choosing the values of hb1 and hb2, the synthesizer can choose whether to use a literal or a variable in that place. In a proper synthesis system, we’d hide these conditions from the surface syntax—the constraint generator would insert a condition for every ??. By conditionally switching between a wider variety of syntax forms—even using nested conditions for nested expressions—the tool can synthesize complex program fragments in each hole.

Keep Synthesizing

It may be a toy language, but we’ve built a synthesizer! Program synthesis is a powerful idea that can come in handy in far-flung domains of computer science. To learn more about the hard parts that come next, I recommend James Bornholt’s overview. And you must check out Rosette, a tool that gives you the scaffolding to write synthesis-based tools without interacting with an SMT solver directly as we did here.