Page 44 - 23_EETE_03

44 EE|Times EUROPE

QUANTUM COMPUTING
‘Quantum Calculator’ Algorithm Tackles

Optimization Problems

By Stefano Lovati

ultiverse Computing has developed a “quantum calculator” Optimization problems can be classified
algorithm intended to transform a quantum computer into a according to the types of variables they deal
with:
mathematical tool for executing complex calculations that are • In continuous optimization, all the vari-
Mcurrently carried out by specialized software to solve problems ables in the problem can take continuous
like optimization. The calculations are on par with the results obtained by the values.
• In discrete optimization, all the variables
highest-performance conventional computers and will improve as quantum in the problem can take values only on a
computing capability grows, proving that quantum computers can be useful finite set.
now, according to Multiverse, a deep-tech company with headquarters in San • In mixed optimization, some of the
variables are continuous and others are
Sebastian, Spain, and subsidiaries in Toronto, Paris and Munich. discrete.
Additionally, an optimization problem is
Quantum computing promises extraordinary increases in computa- said to be a linear program or linear programming problem (LP) if both
tional speed and power relative to today’s computers, yielding benefits the function to be optimized and the function describing the constraint
in fields like pharmaceuticals, healthcare, manufacturing, cybersecurity are linear. Conversely, a nonlinear program or nonlinear programming
and finance. For example, quantum computing can accelerate financial problem (NLP) is an optimization problem in which one of the func-
portfolio management models, such as the Monte Carlo model, for tions is nonlinear.
calculating likely outcomes and attendant risks. The computers can A typical combinatorial optimization problem, important in theoreti-
carry out multiple sophisticated calculations simultaneously, which cal computer science and operations research, is the traveling salesman
is especially helpful for factorizations and could advance the develop- problem (TSP). The cities that the salesman might visit are the points
ment of decryption technology. in the problem statement, and the goal of the TSP is to minimize the
Quantum computers can run challenging simulations, as they travel time and distance to minimize costs.
can mimic highly complex systems at speeds well beyond the reach This seemingly simple example can be solved using integer linear
of traditional computers. programming, but it has been demonstrated that the exact solution of
Multiverse’s quantum Molecular simulations, which the TSP requires algorithms with exponential complexity. The TSP is
calculator algorithm are crucial in the creation therefore said to be an NP-hard problem (NP stands for nondeterminis-
of prescription drugs, might
tic polynomial time).
In the case of nonlinear programming, where either the
demonstrates how benefit from this capability. objective function or the constraint function is nonlinear (i.e.,
And quantum computers’
quantum computers promised optimization includes nonlinear operations), the complexity spirals and the
can efficiently results, with their ability to optimization problem grows even more challenging. As a result,
process enormous volumes of
approximate linear models are usually chosen, even when a non-
implement arbitrary complex data, could revolu- linear objective would be the more appropriate approach given the
tionize artificial intelligence
conditions.
multidimensional and machine learning.
Research collabora-
function calculus. tions and investments by THE ‘QUANTUM CALCULATOR’ ALGORITHM
Founded in 2019, Multiverse Computing has already provided software
large tech companies have for financial companies looking to gain an edge with quantum comput-
advanced the technology in recent years to the point where the first ing for portfolio optimization, risk analysis and market simulation. The
commercial quantum computers have been built and brought to company’s Singularity SDK product for portfolio optimization offers an
market, though these early models are constructed from sparse sets of interface that Multiverse says is as simple to use as a common spread-
noisy qubits. Future CPUs will likely be more powerful and noise- sheet and doesn’t require expert understanding of quantum computing,
resistant, if historical trends provide any indication. In the meantime, even though its performance is dependent on cutting-edge quantum
the task is to put today’s noisy intermediate-scale quantum (NISQ) techniques.
devices to productive use. In creating its quantum calculator algorithm, the company has
Optimization is an important area in which today’s NISQ devices can demonstrated how quantum computers with few qubits can already
start to add value. Hybrid quantum-classical solutions can deploy mod- implement arbitrary multidimensional function calculus in a remark-
ern quantum annealers to address challenging optimization problems ably efficient way. The basic building block of Multiverse’s approach is
in practical settings. a variational quantum algorithm to optimize functions with continuous
domains. This last point is important: Unlike the discrete optimization
THE OPTIMIZATION PROBLEM AND HOW TO SOLVE IT problems normally addressed by quantum computing, here, the objec-
An optimization problem entails maximizing or reducing a function tive function uses continuous variables.
in relation to a set that represents the range of options possible in a Continuous optimization is fundamental to many real-world issues
particular circumstance. The tool enables comparison of the various in mathematical science and engineering beyond financial modeling,
options to identify which may be the best. such as biomolecular design and fluid dynamics.

MARCH 2023 | www.eetimes.eu