State Preparing with Genetic Algorithm

About this project

This project was built by members of Team #5 at the NTU-Qiskit Hackathon Taiwan 2023 and major maintenanced by Yu-Cheng Chung. This project focuses on performing Quantum State Preparation by utilizing a modification of a genetic algorithm structure.

Algorithm

graph 
    A(Initialize Population)
    F(End)
    B[[Evaluate Genes]]
    C[[Generate Parents]]
    D[Crossover & Mutation]
    E[iter += 1]
    
    if2{have gene fidelity<br/> > threshold}
    if1{iter < maxiter}
    
    A --> B --> if2
    C --> D --> A

    if1--> |no| F
    if2-->|no| F
    if2-->|yes| E--> if1
    if1-->|yes| C

Loading

graph LR
    subgraph Evaluate Genes
        subA[/gene/]
        subB[optimize]
        subC[/depth/]
        subD[/fidelity/]
        subA---subB-->subC
        subB-->subD
    end

Loading

graph LR
    subgraph Generate Parents
        sub(entry point)
        subC{have more than two<br/>fidelity >= filter}
        subD[get 10 or less<br/>smallest depth]
        subE[filter -= 0.01]
        subG[/smallest genes/]
        subH(return)
        sub-->subC
        subC --> |no|subE -->subC
        subC --> |yes|subD -->subG -->subH
    end

Loading

Gene

Generate Gene using the matrix look like below,

$$ \begin{bmatrix} G_{00} & G_{01} & \ldots & G_{0m} \\ G_{10} & G_{11} & \ldots & G_{1m} \\ \vdots & \vdots & \ddots & \vdots \\ G_{n0} & G_{n1} & \ldots & G_{nm} \\ \end{bmatrix} $$

Here, $G_{ij}$ is the j-th gate of the i-th qubit. Each $G_{ij}$ is represented as a tuple (gate, control), where:

• gate is the index of the gate in gene_gates.

• control is the index of the control qubit (if the gate has no control, the control value is ignored).

Last Version

• Use basis gate in real computer to generate the circuit: CZ, ID, RZ, SX, X.

• Use empty gate to reduce depth of the circuit.

This Version

• use high level to generate the circuit: RX, RY, RZ, CX.

• Add more rule when generate:

  • do not allow same rotate gate continuously.

  • do not allow more than three rotation gate in a rotation chain continnously.

Results

w-state (5 qubits)

• circuit

• probability distribution

• depth v.s. iter

gaussian state (5 qubits)

• circuit

• probability distribution

• depth v.s. iter

Environment setup

This project was built using Python 3.8.5. The required packages can be installed using the following command:

No-GPU

pip install -r requirement/requirement.txt

NVIDIA-GPU

For cuda11.

pip install -r requirement/requirement-cu11.txt

For cuda12.

pip install -r requirement/requirement-cu12.txt