Quantum Entry Point

A quantum model in Qmod is compiled into a quantum program - a concrete executable description. You can execute a quantum program any number of times on quantum hardware or simulators, and specify execution preferences (such as number of shots). When executing a parametric quantum program you must assign values to its parameters. Both compilation and execution start from a user-defined quantum function called 'main', which is the quantum program entry point. Function main specifies the inputs and outputs of the quantum program, that is, its interface with the external classical execution logic.

Model Outputs

Function main can declare quantum output arguments. Upon invocation, the quantum program executes the specified number-of-shot times, and its outputs are measured each time. When using the sample operation, each output variable is measured in the computational (z) basis. Their names and values, along with the respective counts, are available in the returned result. The values are interpreted according to their specified types (see Quantum Types).

Function main cannot declare quantum arguments other than using the output modifier, since the classical execution logic cannot pass quantum states as arguments.

For example, consider the following model:

NativePython

qfunc main(output a: qbit, output b: qnum<2, UNSIGNED, 2>) {
  allocate(1, a);
  H(a);
  allocate(2, b);
  control (a) {
    apply_to_all(lambda(target) {
      X(target);
    }, b);
  }
}

from classiq import qfunc, Output, QBit, QNum, UNSIGNED, allocate, H, control, X


@qfunc
def main(a: Output[QBit], b: Output[QNum[2, UNSIGNED, 2]]) -> None:
    allocate(1, a)
    H(a)
    allocate(2, b)
    control(a, lambda: apply_to_all(lambda target: X(target), b))

This model can be synthesized and executed in the SDK with the following code -

model = create_model(main)
qprog = synthesize(model)
job = execute(qprog)
result = job.result()[0]
print(result.value.parsed_counts)

The printout will show the counts of the measured values of a and b thus -

[{'a': 1.0, 'b': 0.75}: 503, {'a': 0.0, 'b': 0.0}: 497]

Similarly, the results of the execution job in the Classiq web application show the values as tooltip on the histogram bars.

Model execution parameters

Function main can declare classical parameters of scalar types (integers and reals) and arrays thereof with explicitly specified lengths. These are called execution parameters. They are assigned by the external classical execution logic using arguments to the execution operations sample and estimate. Execution parameters are left in their symbolic form in the quantum program, so that passing different sets of parameter values does not require re-synthesis of the model.

Execution parameters can figure in restricted contexts only as rotation angles in gate-level functions and as the exponent value of the power operation. They cannot be used in classical control flow statements (repeat and if) or in subscript/slice expressions.

Below is a full SDK example of a very simple model with an execution parameter. The quantum program is invoked in a loop using the method sample of ExecutionSession.

from math import pi
from classiq import qfunc, Output, QBit, CReal, allocate, RX, create_model, synthesize
from classiq.execution import ExecutionSession
from classiq.interface.executor.execution_preferences import ExecutionPreferences


@qfunc
def main(angle: CReal, res: Output[QBit]) -> None:
    allocate(1, res)
    RX(angle, res)


model = create_model(main)
qprog = synthesize(model)

es = ExecutionSession(qprog, ExecutionPreferences(num_shots=10000))
for i in range(4):
    result = es.sample({"angle": i*pi/4})
    print(result.parsed_counts)

Running this script will print, for example -

[{'res': 0.0}: 10000]
[{'res': 0.0}: 8560, {'res': 1.0}: 1440]
[{'res': 0.0}: 5019, {'res': 1.0}: 4981]
[{'res': 1.0}: 8514, {'res': 0.0}: 1486]