How do I do printf debugging in Q# in a convenient way?

Question

When initially writing an operator in Q#, I often want to see intermediate values of registers during the computation. This allows me to check that I haven't made mistakes. For example, if I was writing an addition circuit I would input a computational basis state and print out the computational basis states of qubits at particular key points.

I am not aware of a way to do this conveniently. If I print out the qubits like Message($"{qubit}") then I get their IDs instead of their values. That makes sense. I have to do a measurement to access their value. But if I do a measurement, then Q# will e.g. not automatically generate an adjoint operation and this tends to cause compilation failures. Also, I don't actually want to perform a measurement (which may have side effects) I just want to peek at the simulator state.

(I originally thought I could package the concept of "peeking" at a value into an operation that did a hidden measurement, which would have resolved the issue. But Q# doesn't allow operations with an adjoint to have a return type.)

Is there some built-in way to get at the computational basis value of some qubits, and print it to the console during simulation under the Toffoli simulator?

Answer 1

For Toffoli simulator in particular, DumpRegister will provide this information. For example, the following code

operation XorTest() : Bool {
    using ((a, b) = (Qubit[2], Qubit[2])) {
        // initialize: a = 1, b = 2
        ApplyPauli([PauliI, PauliX], a);
        ApplyPauli([PauliX, PauliI], b);
        // check initialization
        Message("a = ");
        DumpRegister((), a);
        Message("b = ");
        DumpRegister((), b);
        // calculate a ⊕ b and write it to b
        CNOT(a[0], b[0]);
        CNOT(a[1], b[1]);
        // check result: a ⊕ b = 3
        Message("a xor b = ");
        DumpRegister((), b);
    }
    return true;
}

will print the following result (and throw an exception in the end because the qubits are released not in zero state):

a =
State:
0:      False
1:      True
b =
State:
2:      True
3:      False
a xor b =
State:
2:      True
3:      True

The numbers before values are qubit ids.