Quantum Programming with Qiskit

You've learned the theory: qubits, gates, and circuits. Now let's write code. Qiskit is IBM's free, open-source Python toolkit for quantum programming. In this guide you'll install it, build your first quantum circuit, create an entangled pair of qubits, and learn how to read the results.

What is Qiskit?

Qiskit (pronounced "kiss-kit") is a Python library for describing quantum circuits and running them. The same code can run on a simulator on your laptop or on a real quantum computer in IBM's cloud.

How do you install Qiskit?

You need Python 3.9 or newer. Create a virtual environment so Qiskit doesn't clash with other projects, then install it with pip:

python3 -m venv qenv
source qenv/bin/activate      # on Windows: qenv\Scripts\activate
pip install qiskit

The four building blocks you'll use first

Almost every beginner program uses these four operations. Pick one to see what it does:

qc.h(0)   # |0⟩ → equal mix of |0⟩ and |1⟩

qc.x(0)   # |0⟩ → |1⟩

qc.cx(0, 1)   # control = qubit 0, target = qubit 1

qc.measure_all()   # adds classical bits named "meas"

Your first quantum circuit: a coin flip

A single qubit in superposition behaves like a perfectly fair coin. Let's build it and "flip" it 1,000 times:

from qiskit import QuantumCircuit
from qiskit.primitives import StatevectorSampler

qc = QuantumCircuit(1)      # a circuit with 1 qubit, starting in state 0
qc.h(0)                     # Hadamard: put qubit 0 into superposition
qc.measure_all()            # measure it

sampler = StatevectorSampler()                  # a simulator that runs on your computer
result = sampler.run([qc], shots=1000).result()
counts = result[0].data.meas.get_counts()
print(counts)

• QuantumCircuit(1) — creates a circuit with one qubit. Every qubit starts as 0.
• qc.h(0) — applies the Hadamard gate to qubit 0.
• StatevectorSampler() — Qiskit's built-in simulator for running circuits and sampling results.
• shots=1000 — run the circuit 1,000 times. Each run gives one random 0 or 1.
• data.meas.get_counts() — tally the results. meas is the name measure_all() gives the classical bits.

You'll get roughly 500 zeros and 500 ones, for example {'0': 509, '1': 491}. The exact numbers change every run, just like real coin flips. Add print(qc.draw()) to see a text diagram of your circuit.

Let's see it in action: entangling two qubits

Now for something no classical coin can do. We'll create a Bell state: two qubits that always agree when measured, even though each one on its own is completely random. This is entanglement.

from qiskit import QuantumCircuit
from qiskit.primitives import StatevectorSampler

qc = QuantumCircuit(2)
qc.h(0)          # qubit 0 into superposition
qc.cx(0, 1)      # entangle: qubit 1 copies qubit 0's fate
qc.measure_all()

counts = StatevectorSampler().run([qc], shots=1000).result()[0].data.meas.get_counts()
print(counts)

Why do Qiskit results look backwards?

Run this, which flips only qubit 0:

qc = QuantumCircuit(2)
qc.x(0)          # flip qubit 0 only
qc.measure_all()
print(StatevectorSampler().run([qc], shots=10).result()[0].data.meas.get_counts())

{'01': 10}

You might expect '10'. Qiskit writes results little-endian: qubit 0 is the rightmost character. Read result strings from right to left to match qubit numbers. It's a convention, and once you know it you'll never be confused by it again.

Scaling up: a three-qubit GHZ state

Chain more CNOTs and the entanglement spreads. All three qubits now share one fate:

qc = QuantumCircuit(3)
qc.h(0)
qc.cx(0, 1)
qc.cx(1, 2)
qc.measure_all()

Measuring gives only 000 or 111, about half the time each. States like this are used in quantum error correction and to benchmark real quantum hardware.

How do you run on a real quantum computer?

IBM offers free monthly time on real quantum processors through the IBM Quantum Platform. The steps are:

1. Create an IBM Quantum account and get an API key.
2. Install the runtime client: pip install qiskit-ibm-runtime.
3. Transpile your circuit: rewrite it using only the gates and qubit connections the chosen processor actually supports.
4. Submit it with the runtime Sampler and wait in the queue for results.

Other quantum programming frameworks

The concepts (qubits, gates, circuits, and measurement) are the same everywhere, so skills transfer easily between them.

Quick recap

What's next?

Try implementing Deutsch's algorithm in Qiskit, the simplest algorithm with a real quantum advantage. Then see where this is all heading in quantum machine learning.