This repository is DM_Simulator based implementation of QC-Application Oriented Benchmarks framework. Aim of this repository is to benchmark the performance of algorithms mentioned in the Application oriented Benchmark Suite with the QSIM (Simulator) which works on the basis of dm_simulator.

References :

To view that repository please go through the following link : QC-Application-Oriented-Benchmark-Suite

To Know more about this benchmarking framework please go through the following papers :

Application-Oriented Performance Benchmarks for Quantum Computing

Optimization Applications as Quantum Performance Benchmarks

Quantum Algorithm Exploration using Application-Oriented Performance Benchmarks

All the algorithms in the framework are benchmarked in qiskit, cirq, bracket environments.

Instructions for setting up the repository and further information regarding the benchmarks are already provided in this Repository

For DM_Simulator (QSIM):

For Using DM_Simulator kindly clone the following repository :

https://github.com/indian-institute-of-science-qc/qiskit-aakash/tree/terra_upgrade

Note : Use the hash id -> bea98fbff86c234c0b1990add17493a1e86917cb while cloning the above repository.

In this repository we have changed the QC-Application-Oriented-Benchmark-Suite framework and made it into a single .ipynb file (for each algorithm) so that there will be no need dependencies.

All the algorithms of Benchmarking Suite are categorized into three types Tutorial, Sub-routine and Functional. So the same flow is maintained in this repository.

The status of every algorithm in each category will be maintained in their corresponding directory.

Installation Procedure :

1. Create an environment in Anaconda.

conda create -y -n QSIM python=3.8

2. Activate the environment

conda activate QSIM

(Optional) To deactivate the environment run the following command in terminal.

conda deactivate

3. Clone the repository :

git clone -b terra_upgrade https://github.com/indian-institute-of-science-qc/qiskit-aakash.git

4. Go to the specified hash-id:

cd QSIM
git checkout -b dmsim bea98fbff86c234c0b1990add17493a1e86917cb
git log -n1

5. Installation of requirements : Install ipython-genutils :

python -m pip install ipython-genutils

or

conda install -c conda-forge ipython_genutils

Install rustup :

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
python -m pip install setuptools-rust

Install requirements : If you want to run tests or linting checks, install the developer requirements.

pip install -r requirements-dev.txt

Install Qiskit-Aakash :

pip install .

NOTE :

As you were installing a specific version of Qiskit-Aakash repository for dm_simulator please note that it has the following issues.

1. Reverse Order of Probabilities :

Density Matrix simulator (on the basis of Basic-Aer) represents the Measurements in the reverse order when compared with that of QASM-Simulator.

• To resolve this we are using a built-in functionality called reverse_bits().

• This function places measurement gates in reverse order of which QASM Simulator uses.

• Use this fucnction while creating the Quantum Circuit.

For example:

qc = BersteinVazirani(num_qubits, s_int, method).reverse_bits()

2. Partial and Ensemble probabilities :

In this repository we mostly measure our resulting probabilities in two methods.

a. Measurement with Ensemble Probability

b. Measurement with Partial Probability

• While creating Quantum Circuit, if we use basis as "Ensemble" then we get our measurements as ensemble_probability in our result object. But if we haven't mentioned any basis specifically, dm_simulator by default gives partial_probability in our result object.

3. Issues with Ensemble Probability Measurement :

Ensemble probability measures every qubit in the quantum circuit and returns measurements accordingly. But in our benchmarking suite most of the algorithms use n-1 measurements for n number of qubits. This is because the remaining one qubit is considered as auxillary which isn't meant for measurement. (Check the algorithms for better clarification).

• Due to this reason we are not using Ensemble-measurement for our Benchmarking. Benchmarking with Ensemble measurement is implemented in Deutsch-Jozsa Algorithm (optional).

4. Limitations of Partial probability measurement :

Atleast 2 measurements are needed to be performed (i.e., the Quantum circuit should have atleast 2 Measurement gates at the output) to calculate partial_probability values. Anything less-than 2 doesn't work, it simply throws an error. This version of DM_Simulator doesn't have the implementation to get the partial_probability for single-qubit measurement.

Information about benchmarking limitations on native computers will be available as README of each directory of Algorithm.