INDIAN INSTITUTE OF TECHNOLOGY ROORKEE

Quantum Circuit Design for Quantum

Computing : Variational Quantum Algorithms

Presented by :
Tanmay Bakshi
21122048
B.Tech. (Computer Science and Engineering) : III-YR
Contents

1. Prologue : Leap from Classical to Quantum Computation

2. Problem Statement & Current State of the Art

3. Principle : The Variational Theorem

4. Rigorous Treatment and Optimality of Approach

5. Process Flow-Chart

6. Challenges and Limitations

2
Prologue
• What forms the basis of Classical Computation as we know today - The Church-Turing Thesis
Idea : "Any Real-World process can be simulated by a Turing Machine to any desired precision. "

• The Extended Church-Turing Thesis says that on simulating reality on a digital computer,
there's at most a polynomial blowup in computational resources.

• Feynman's Dilemma : How do we simulate quantum mechanics on a computer ?

Naïve approach – Keep track of every particle's wave-function. (several variables that scale
exponentially!) and use approximation methods.
OR
• In Feynman's Words, "Nature isn't classical. If you want to make a simulation of nature, you'd
better make it quantum mechanical.“

• Hence, with mathematical rigor and physical theoretics a new class of computers were
designed, with a new unit of information – called QUBITS.
A Qubit is a two-state device, much like classical switches – but quantum mechanics allows a
qubit to exist in superposition of the two states.
A pure qubit state is a coherent superposition of the basis states. A single qubit () can be
described by a linear combination of computational basis states in 2-D Hilbert Space –

3
Problem Statement & Current State of
Art
• Current state-of-the-art technology is faraway from realizing a Universal Quantum
Computer.
• The most promising developments have focused on a particular model of computation –
The Quantum Circuit Model.

• In a quantum circuit, wires represent qubits (vectors) and gates represent unitary matrix
operations acting on these qubits, followed by standard basis measurements.
• Current state of quantum computing is referred to as the Noisy Intermediate-Scale
Quantum (NISQ) era. These processors, containing physical implementations of qubits are
highly sensitive to noise and prone to quantum decoherence.

• Problem : How to make best use of today’s NISQ devices to achieve quantum
advantage.

Variational Quantum Algorithms (VQAs) have emerged as the leading strategy to obtain
quantum advantage on NISQ devices. These algorithms use optimization/learning based
approaches.
Effectively, VQAs are the quantum analog of ML-based methods, and leverage the classical
prowess of computing for optimization.
4
Principle : Variational Theorem
• Just like how a car's state can be described using physical properties like speed or acceleration, the
state of a quantum system/circuit can be described using observables – essentially matrices acting as
linear operators.
• A common goal of variational algorithms is to find the quantum state with the lowest or highest
eigenvalue of a certain observable.
• The observable corresponding to energy is called the Hamiltonian, denoted by .

Its Spectral Decomposition is given by (where : i th Eigenvector and : Corresponding Eigenvalue)

Where N : Dimensionality of the State Space.

The Expected Energy for a system with state is :

Then, for a parametrized state of a quantum system, the optimal approximation of the ground state (i.e.,
the eigenstate with the minimum eigenvalue ) is the one that minimizes the expectation value of the
Hamiltonian :

5
 Idea/Approach Details
The implementation of Variational Theorem is broken down into modules that can be optimized separately and combined together. The
following are some particular details of the implementation of the Variational Approach :

(1) INITIALIZE PROBLEM :

Start the quantum computer in some default state ( transform it
Into some desired state ( that is non-parameterised via gate operations. This
Newly obtained state is called the REFERENCE STATE. This transformation can be
Represented as -
=

(2) PREPARE ANSATZ : (Use to define and search the solution space)
These are parametrized circuits to be prepared at the start of every iteration.
We define a unitary transform, (
Any particular combination of the Variational Form and the Reference State is referred to as an ansatz, such that :
=
=
=

(3) EVALUATE COST FUNCTION : (Describes the specific problem with a set of parameters). The problem is encoded as the minimization
objective of a cost function (Operators ( O(n) number of operators, all of which can be measured in O(1) time in ideal scenarios).

(4) OPTIMIZERS : The quantum circuit is used to compute the gradients, (use parameter shift rule to make computation of gradients more
efficient) and the gradients are send to classical computers, that perform gradient descent update on the parameters
Hence, we hope of obtaining an approximation of optimal state that minimizes .

6
Process flow chart

INITIALIZE PROBLEM
Prepare Default State Transform to Reference State

PREPARE ANSATZ (Parametrized Circuit)

Heuristic Ansatz (N-local circuits) Problem-Specific Ansatz (Pauli Operators)
Adjust Ansatz
Parameters
and feed-back
to the circuit
to obtain new
EVALUATE COST FUCTION
results.
Sampler and Estimator Primitives to measure Error Mitigation to deal with Noise Continue till
Optimizer
Condition not
satisfied.
UPDATE PARAMETERS USING CLASSICAL OPTIMIZERS
Local/Global Optimizers Gradient-based/Gradient-free Optimizers

7
Challenge and Limitations
• PHYSICAL LIMITATIONS :
Since we are using a noisy quantum device as a black-box, quantum noise can make the retrieved values
non-deterministic, leading to random fluctuations which, in turn, will harm – or completely prevent the
convergence of certain optimizers to a proposed solution.

• DEPTH AND CIRCUIT SIZE :

Our search space depends on the chosen variational form. A long circuit and more number of qubits could
ensure larger search space, but would lead to increasing decoherence, gate errors, readout errors etc.

• HYBRID COMPUTING OVERHEADS :

Variational quantum algorithms often involve a combination of quantum and classical computing. As the
quantum part becomes more complex, the classical part must work harder to adapt and analyze the
quantum outputs, potentially negating any quantum speedup.

• LACK OF EFFECTIVE ERROR CORRECTION TECHNIQUES :

Despite overcoming other overheads, a fundamental problem in realizing theoretical quantum computation
on actual NISQ devices is the lack of effective error correction techniques, which become exponentially
more relevant as the circuit depth increases.

8
Applications

• Variational Quantum Algorithms are currently the forerunners in the race

of providing "Actual" Quantum advantage through their compatibility with
NISQ devices and integrity with classical computing resources.
• Some areas of Applications are -

1. Chemistry & Material Science :

o Predicting Molecular structure and dynamics (NP-hard protein folding
problems).
o Efficient materials simulation of strongly correlated materials .

2. Optimization Problems :
o The idea is to encode the optimization problem as a Hamiltonian
(QUBO), use the parameter-shift rule to compute gradients and predict
ground-state observables, encoding the optimal solution.
o Can efficiently compute Combinatorial Optimization problems like
the Vehicle Routing Problem.
9
References

• Quantum theory, the Church-Turing principle, and the Universal

Quantum Computer – David Deutsch
• Introduction to Quantum Information Science (Fall 2018, UT Austin) -
Scott Aaronson
• Variational Quantum Algorithms – M. Cerezo et. Al (
https://arxiv.org/pdf/2012.09265.pdf)
• Quantum Theory, the Church-Turing Principle and the Universal
Quantum Computer – David Deutsch(1985) (
https://www.cs.princeton.edu/courses/archive/fall04/cos576/papers/deuts
ch85.pdf
)
• IBM Quantum Computing Platform : Variational Algorithm Design (
https://learning.quantum-computing.ibm.com/course/variational-algorithm
-design/
)

10