Indirect-Control VQE and ZNE Error Mitigation

Installation

• Python Version: 3.11
• To install dependencies, run:

  pip install -r requirements.txt

Usage

To run the program, use:

python3 main.py <config.yml>

Configuration Details

The program uses a YAML configuration file to define its parameters. Below is a detailed description of the configuration categories and their parameters:

Section	Key	Type	Description
Run Configuration	run	String	Defines the algorithm to run. Options: 'vqe', 'redundant', 'zne'.
System Configuration	nqubits	Integer	Specifies the number of qubits in the quantum system.
	state	String	Defines the state representation. Options: 'dmatrix' (density matrix) or 'statevector'.
Observable (Target Hamiltonian)	def	String	Defines the type of Hamiltonian. Options: 'custom', 'ising', 'heisenberg'. Warning: Coefficients are overwritten based on the selected type. For 'ising', cn values are set to [0.5], bn values are set to [1], and r is set to 1. For 'heisenberg', only cn is used.
	coefficients	Object	Contains coefficients for the Hamiltonian terms:
		cn	List of Floats
		bn	List of Floats
		r	Float
Output Configuration	file_name_prefix	String	Specifies the prefix for the output file name.
	draw	Object	Configures drawing options for output files:
		status	Boolean
		fig_dpi	Integer
		type	String
VQE Configuration	iteration	Integer	Specifies the number of iterations for the VQE algorithm.
	optimization	Object	Configures optimization settings:
		status	Boolean
		algorithm	String
		constraint	Boolean
	ansatz	Object	Configures the ansatz circuit:
		type	String
		layer	Integer
		gateset	Integer
		ugate	Object
			coefficients
			cn
			bn
			r
			time
			min
			max
		noise	Object
			status
			value
	init_param	String	Defines the initialization method for parameters. Options: 'random', etc. Example: "random".
Redundant Circuit Configuration	identity_factors	List of Lists	Specifies the identity scaling factors for gates. Example: [[1, 0, 0, 1], [2, 3, 2, 1], ...].
	Warning	String	Warning: Identity scaling for the U gate is supported only if the ansatz type is 'xy-iss'. For other types, the identity scaling factor for the U gate must be set to 0.
Zero Noise Extrapolation (ZNE)	method	String	Defines the method for zero-noise extrapolation. Options: 'linear', 'polynomial', 'richardson', 'richardson-mul'.
	degree	Integer	Sets the degree for polynomial or Richardson methods. Used for regression.
	sampling	String	Specifies the sampling method for extrapolation. Options: 'default', 'default-N', 'random-N', where N is an integer.
	data_points	List of Lists	Provides the data points for extrapolation. Each entry contains several values, such as time and noise levels. Example: [[12, 1, 0, 3, -3.3480294367352315], [20, 7, 10, 3, -0.05316450776222178], ...].

Warnings and Important Notes

1. Observable (Target Hamiltonian) Coefficients Overwritten:

   • For the ising Hamiltonian (observable.def: "ising"), the coefficients are automatically overwritten as follows:
     • cn: [0.5]
     • bn: [1.0]
     • r: 1
   • For the heisenberg Hamiltonian (observable.def: "heisenberg"), only the cn coefficient is used, and bn and r are ignored.

2. Ansatz Type for Zero Noise Extrapolation (ZNE):

   • When using Zero Noise Extrapolation (ZNE) with redundant circuits, the ansatz type must be set to 'xy-iss' (vqe.ansatz.type: "xy-iss"). This is required for identity-scaling the circuit gates (U, Y, and CZ).
   • Warning: The identity-scaling for the U gate is only supported if the ansatz type is 'xy_model-xz-z'. For other ansatz types, the identity scaling for the U gate must be set to 0.

3. VQE Ansatz Coefficients Overwritten:

   • When using the VQE algorithm with certain ansatz types ('xy-iss'), the coefficients are automatically overwritten as follows:
     • cn: [0.5]
     • bn: [0]
     • r: 0
   • For ansatz types such as 'ising' or 'heisenberg', the coefficients are predefined according to the selected type.

4. Initialization of Parameters:

   • The initial parameters for the ansatz are set to random by default (vqe.ansatz.init_param: "random"). If you need to modify the initialization method, ensure to update this value accordingly.

5. Sampling Method for Zero Noise Extrapolation:

   • The sampling method should be selected appropriately:
     • 'default' - All points are sampled.
     • 'default-N' - The first N points are sampled.
     • 'random-N' - N points are sampled randomly.
   • Ensure that the method aligns with the desired sampling strategy for extrapolation.

Testing

For testing, use pytest.
To run the tests, execute:

pytest test/.

Linting and Formatting

• Use flake8 for linting.
• Use black and isort for formatting the code.

  # Run linting
  flake8 .
  
  # Run formatting
  black .
  isort .