1. Why Quantum Simulation Is Hard
A quantum state of n qubits is described by a vector of 2ⁿ complex amplitudes.
Example:
10 qubits → 1,024 amplitudes
30 qubits → 1 billion amplitudes
50 qubits → ~1 quadrillion amplitudes
60+ qubits → practically impossible on a single classical computer
This exponential growth means:
Memory and compute explode quickly
Simulation time increases dramatically
Only small to medium circuits (typically up to 30–40 qubits) can be simulated exactly on a laptop
To go beyond this, specialized algorithms or HPC clusters are used.
2. Two Main Approaches to Classical Quantum Simulation
A. State Vector Simulation (Exact, but Exponential)
You store the full quantum state (2ⁿ amplitudes) in memory.
Key ideas:
Represent the quantum state as a complex vector.
Apply gates as unitary matrices acting on the vector.
Each gate modifies all relevant amplitudes.
Pros
Exact simulation
Fast for small qubit counts
Easy to implement and widely supported
Cons
Requires 16 × 2ⁿ bytes (complex64) or more
~28 qubits fits in 1 GB RAM
~34 qubits → ~128 GB RAM
Quickly becomes impossible for large n
Tools
Qiskit Aer (IBM)
QuTiP
Cirq Simulator
NVIDIA cuQuantum (GPU-accelerated)
Simulatrices in PennyLane
B. Tensor Network Simulation (Approximate or Structured Circuits)
Instead of storing the full 2ⁿ vector, you break the circuit into tensors and contract them efficiently.
Key ideas:
Use graph-like structures (MPS, PEPS, MERA, etc.)
Contract tensors in optimal order
Exploit low entanglement in many practical circuits
Pros
Can simulate 50–100+ qubits in special cases
Extremely efficient for circuits with limited entanglement
Used in physics and chemistry simulations (DMRG, TEBD)
Cons
Struggles with highly entangled circuits
Approximation errors may accumulate
Tools
ITensor
Quimb
TensorNetwork (Google)
Cotengra / opt-einsum
3. Hybrid Methods
Combining approaches allows simulation of larger or more complex systems:
Examples
Schrödinger–Feynman method: splits qubits into two halves
Path integral methods: sum over all histories (good for shallow circuits)
Stabilizer formalism (Gottesman–Knill): efficient for Clifford circuits
Density matrix simulation: for noise modeling, but even more expensive
Hybrid techniques often mix:
State vector methods
Tensor networks
Classical sampling
Some simulators automatically choose the best approach for each gate or region of the circuit.
4. Classical Hardware Used for Quantum Simulation
Because simulation requires massive compute resources, specialized architectures are helpful.
CPUs
Good for large memory
Used in supercomputing clusters
Can simulate 40–45 qubits with distributed memory
GPUs
Excellent for linear algebra
NVIDIA cuQuantum accelerates quantum simulation by huge factors
Can simulate up to 30–40 qubits efficiently on a multi-GPU setup
Supercomputers
Top HPC machines can simulate 50–60 qubits using:
distributed tensor networks
thousands of CPU cores
petabytes of memory
Examples:
Google’s 53-qubit Sycamore simulation
IBM’s HPC experiments
HPC clusters using cuQuantum DGX nodes
5. Noisy Quantum Simulation
To model real hardware, you include:
decoherence
depolarizing noise
T1/T2 relaxation
readout errors
crosstalk
Approaches:
density matrix representation (4ⁿ complexity) → expensive
Monte Carlo trajectory simulation (stochastic)
Approximate noise models (Pauli channels)
Tools:
Qiskit NoiseModel, Cirq Noise, PennyLane QNode with noise.
6. Practical Steps to Simulate a Quantum Circuit
Here’s a typical workflow:
Describe the circuit
Create qubits and apply gates.
Choose the simulation model
state vector (exact)
tensor network (efficient for structured circuits)
stabilizer (fast for Clifford circuits)
Run the circuit
Use a simulator backend (e.g., Qiskit Aer).
Measure the output
Classical sampling of the quantum state.
Analyze results
Compute expectation values, probability distributions, or final bitstrings.
7. What Classical Simulation Is Used For
Even with limitations, classical quantum simulation is essential for:
✔ Algorithm development (e.g., QAOA, VQE, QFT)
✔ Circuit optimization
✔ Error correction research
✔ Hardware testing and benchmarking
✔ Education and training
✔ AI + quantum hybrid workflows
It allows developers to validate quantum programs without needing access to real quantum devices.
8. The Limits of Classical Simulation
Even the best algorithms and supercomputers cannot simulate universal quantum circuits beyond ~60–70 qubits with high depth.
This is why:
quantum supremacy experiments target >50 qubits
real quantum hardware is essential for exploring deep, highly entangled computation
simulation helps research, but cannot replace real quantum machines
Summary
Simulating quantum computers on classical machines relies on:
✔ State vector simulation
Fast and exact, but scales as 2ⁿ.
✔ Tensor network simulation
More scalable for low entanglement circuits.
✔ Hybrid and specialized methods
Handle large circuits with structure or approximations.
✔ HPC + GPU acceleration
Pushes simulation limits to 50+ qubits.
✔ Noise simulation
Models realistic hardware behavior.
Despite classical limits, simulation is vital for building the quantum future.
Learn Quantum Computing Training in Hyderabad
Read More
Running Quantum Circuits on IBM Quantum Computers
The Differences Between Qiskit, Cirq, and Braket
Introduction to Cirq: Google’s Quantum Programming Framework
Getting Started with Qiskit: Your First Quantum Program
Visit Our Quality Thought Training Institute
Subscribe by Email
Follow Updates Articles from This Blog via Email
No Comments