QuSim

QuSim is a quantum circuit simulator I built from scratch to understand quantum computing at a level textbooks don’t reach: every phase convention, every measurement collapse, every decoherence channel, implemented and inspectable.

backends

Rather than one monolithic simulator, QuSim exposes four backends behind a single Backend interface, each with different tradeoffs:

• Statevector - exact evolution of pure states in a 2ⁿ-dimensional Hilbert space.
• Density matrix - mixed states and open-system dynamics, the substrate for realistic noise.
• Stabilizer (Gottesman–Knill) - polynomial-time simulation of Clifford circuits, scaling to hundreds of qubits.
• Tensor-network - memory-efficient simulation of large, lightly-entangled systems.

noise and diagnostics

Gates are unitary matrices; noise is modeled with Kraus operators applied to the density matrix, so depolarizing, amplitude-damping, and dephasing channels are first-class. The simulator tracks von Neumann entropy and mutual information in real time, and runs automatic failure diagnostics that flag when a circuit has become noise-dominated rather than silently returning garbage.

# A Bell state, then measure the entanglement it creates
circuit = Circuit(2)
circuit.h(0)        # Hadamard on qubit 0
circuit.cx(0, 1)    # CNOT: 0 -> 1

result = simulator.run(circuit)          # (|00⟩ + |11⟩) / √2
print(result.von_neumann_entropy())      # ~1.0 bit - maximally entangled

error correction

A dedicated sandbox implements syndrome extraction and decoding, so you can watch a logical qubit survive physical errors - the payoff of everything above, made visible.

performance

Through sparse matrix representations and Numba JIT compilation (a 10–50× speedup on gate application), the statevector backend handles circuits up to ~25 qubits on a laptop - enough for most educational and many research scenarios - while the stabilizer backend goes much further for Clifford circuits.

why I built it

I wanted to understand quantum computing more deeply than reading allowed. Building a simulator forced me to confront every detail I would otherwise have hand-waved. The tool I built for myself turned into something genuinely useful for exploring quantum algorithms and noise.