/// # Sample
/// Simulation of a simple Ising model evolution
/// on a 2D grid with first-order Trotterization.
///
/// # Description
/// This sample demonstrates simulation of an Ising model Hamiltonian
/// on N1xN2 2D grid using a first-order Trotter-Suzuki approximation.
/// This sample can be easily simulated classically with 3x3 grid and
/// about 1000 shots. This sample is suitable for Base Profile.
/// For the purpose of simplicity this sample intentionally doesn't
/// post-process results or perform eigenvalue estimation.
operation Main() : Result[] {
// Dimensions of a 2D grid is N1 x N2
let N1 : Int = 3;
let N2 : Int = 3;
// Total evolution time
let evolutionTime : Double = 4.0;
// Number of steps
let numberOfSteps : Int = 5;
// Coefficient for 2-qubit interactions between neighboring qubits
let J : Double = 1.0;
// Coefficient for external field interaction for individual qubits
let g : Double = 1.4;
// Also try simulating with different strength of external field:
// let g = 0.2;
// let g = 1.0;
// let g = 1.4;
// let g = 2.0;
// Model evolution
IsingModel2DEvolution(N1, N2, J, g, evolutionTime, numberOfSteps)
}
/// # Summary
/// Simulate simple Ising model evolution
///
/// # Description
/// Simulates state |𝜓⟩ evolution to find |𝜓(t)⟩=U(t)|𝜓(0)⟩.
/// |𝜓(0)⟩ is taken to be |0...0⟩.
/// U(t)=e⁻ⁱᴴᵗ, where H is an Ising model Hamiltonian H = -J·Σ'ᵢⱼZᵢZⱼ + g·ΣᵢXᵢ
/// Here Σ' is taken over all pairs of neighboring qubits <i,j>.
/// Simulation is done by performing K steps assuming U(t)≈(U(t/K))ᴷ.
operation IsingModel2DEvolution(
N1 : Int,
N2 : Int,
J : Double,
g : Double,
evolutionTime : Double,
numberOfSteps : Int
) : Result[] {
// Allocate qubit grid and structure it as a 2D array.
use qubits = Qubit[N1 * N2];
let qubitsAs2D = Std.Arrays.Chunks(N2, qubits);
// Compute the time step
let dt : Double = evolutionTime / Std.Convert.IntAsDouble(numberOfSteps);
let theta_x = - g * dt;
let theta_zz = J * dt;
// Perform K steps
for i in 1..numberOfSteps {
// Single-qubit interaction with external field
for q in qubits {
Rx(2.0 * theta_x, q);
}
// All Rzz gates applied in the following two loops commute so they can be
// applied in any order. To reduce the depth of the algorithm, Rzz gates
// between horizontal "even" pairs of qubits are applied first - pairs
// that start at even indices. Then Rzz gates between "odd" pairs are
// applied. That way all Rzz between horizontal "even" pairs can potentially
// be done in parallel. Same is true about horizontal "odd" pairs,
// vertical "even" pairs and vertical "odd" pairs.
// Horizontal two-qubit interactions
for row in 0..N1-1 {
// Horizontal interactions between "even" pairs
for col in 0..2..N2-2 {
Rzz(2.0 * theta_zz, qubitsAs2D[row][col], qubitsAs2D[row][col + 1]);
}
// Horizontal interactions between "odd" pairs
for col in 1..2..N2-2 {
Rzz(2.0 * theta_zz, qubitsAs2D[row][col], qubitsAs2D[row][col + 1]);
}
}
// Vertical two-qubit interactions
for col in 0..N2-1 {
// Vertical interactions between "even" pairs
for row in 0..2..N1-2 {
Rzz(2.0 * theta_zz, qubitsAs2D[row][col], qubitsAs2D[row + 1][col]);
}
// Vertical interactions between "odd" pairs
for row in 1..2..N1-2 {
Rzz(2.0 * theta_zz, qubitsAs2D[row][col], qubitsAs2D[row + 1][col]);
}
}
}
MResetEachZ(qubits)
}microsoft/qdk
Publicmirrored fromhttps://github.com/microsoft/qdkAvailable
samples/algorithms/Ising/Simple2dIsingOrder1.qs
112lines · modepreview