UT Austin: Qiskit Fall Fest 2021
Hackathons are an avenue where aspiring creationists, tech enthusiasts, and designers come together to collaborate and test their limits to solve challenging problems. Not only is it an opportunity to bring a game-changing idea from concept to reality, but hackathons evoke an addictive drive to learn, implement, and apply new concepts to coding paradigms to break through the barriers of the unknown; but what if that unknown we asked students to tackle was a completely new way of thinking about our world — and in just one weekend?
That unknown is quantum computing. The UT Quantum Collective asked students to change their classical lens of thinking for a weekend and learn a challenging topic in quantum computing as part of the Qiskit Fall Fest hosted by IBM. This was a month-long event with universities all around the world hosting their own local challenges where winners from these events won a chance to compete at Qiskit Hackathon Global 2021; for UT Austin, this marked the first-ever quantum hackathon for Longhorns.
Quantum Computing has been gaining traction for its potential applications in various industries ranging from simulating complex chemical reactions, portfolio optimization in finance, and machine learning with a quantum mechanical twist. The University of Texas at Austin is positioning itself to become one of the leading institutes for quantum technologies bringing talent, research avenues, and collaborative initiatives across various academic units with the Quantum Information Center. One of the biggest challenges of the field remains: how do you teach a multidisciplinary field that spans the fundamentals of physics, computer science, and mathematics to students?
In our educationally themed hackathon, we asked students to take on the challenge of learning a topic in quantum computing as part of our initiative to break the learning barrier and equalize opportunities for beginners and seasoned experts alike. Before our kick-off ceremony, we spent a week hosting workshops to get students familiar with Qiskit and the fundamentals of Quantum Computing.
From October 22nd to October 24th, students from a variety of backgrounds got together virtually in teams, formed ideas, and developed their projects, and produced presentations summarizing their work. Aside from hacking, we had a packed schedule with notable speakers from Amazon, IBM, Strangeworks, Rice, Baylor, and UT Austin giving their seminars to students about their research, the industry, and job opportunities in Quantum. We concluded our event with a closing ceremony and guest appearance from Dr. Scott Aaronson, who did a Q&A session to answer all burning questions participants had about quantum computing.
We were blown away at the quality of projects and volume of insight our hackers were able to take away in just a weekend; most importantly, we valued how willing experienced team members were to educate their peers and pass down their knowledge to make such astounding projects in just two days. The collaboration we saw at this event strengthened our organization’s mission to be a space for students of all backgrounds to learn, teach, and research topics in quantum computing.
Choosing our winners was a difficult feat, but keep reading to see the results of the top three winning teams from UT Austin’s Qiskit Fall Fest. Everyone’s contribution was a building block for this fantastic event, so we compiled all projects to showcase in our Github here.
If you would like to explore the winners of all the fall fest locations, check out the fall fest website here: Qiskit Fall Fest — Start (hypeinnovation.com)
1st Place: Random Number Generation
Our winning project explored the topic of random number generation to show the true randomness quantum computers are capable of without external sampling in classical random generation algorithms that are deterministic.
The winning team first demonstrates pseudorandomness. Pseudorandomness relies on the idea that classical computers are deterministic, which means that there exists a set of instructions that determine a computer’s actions. In the classical realm, you can get a random string of bits by iterating through this list. One starts from the seed or a certain index in the list. Generating a new list starting from the same seed gives the exact same sequence of bits, but by reseeding, we can get a new random string of bits. While it is possible to generate truly random numbers using classical computers, they rely on a sample of an external random physical phenomenon. This requires extra hardware to measure the phenomena as well as more time to measure the data.
In the quantum world, however, one can rely on the qubit, the most basic, yet important unit of quantum computing. By putting a qubit into superposition, one doesn’t know its final state until measurement. What’s unique about the qubit, however, is that one has the power to influence the probabilities of the qubit’s measured state. This time, we don’t have to rely on seeds, so repeating the process multiple times will give completely different random bits. The only issue now, however, is that noise or imperfect state preparation can compromise this random generation.
To prove the validity of random number generation with quantum computers, the winning team used the CHSH game where the probability of success breaks the classical bound of 75% and is instead 85% to generate a random bit string, appending a 1 or 0 to their string. From here, the game can be used to generate random bits; Alice’s resulting bit is appended to an output string (with losing rounds discarded entirely).
But the team would like to do more with their experimentation! They have considered redesigning a system where the quantum bits output is recycled to generate more quantum random bits. They have also considered involving more players with this implementation.
2nd Place: The Coin Flip Problem: Twirling a Noisy Coin
Our second-place project took an interesting “twist” on the coin flip problem: instead of just testing the outcomes of a fair and biased coin flip using one qubit, they also checked the resilience against noise in quantum hardware and tested how well error-correcting algorithms would work to reduce the noise found in a system.
For the coin flip problem, the quantum advantage is that only one qubit is needed to model the outcomes whereas 2^n bits are needed to model the outcome distribution of a coin flip classically. To model a fair quantum coin, you first initialize your qubit in a superposition of the 0 and 1 state with an equally likely probability of collapsing to either state and measure your outcome, where an outcome of 0 indicates heads and 1 is tails. An unfair coin would result from an unequal superposition. For the algorithm, you initialize a qubit to the 0 state and rotate the qubit in the opposite direction for heads or tails. An unbiased coin will generally yield 0, whereas a biased one will eventually move toward the 1 state.
The noise introduced in the coin flip problem is known as amplitude damping, where the qubit has a small chance to fall back to the 0 state. The team used an environment qubit in their circuit to model this noise given the gate and decoherence times and estimated the errors in the state their qubit had decayed to for some of the subsets of the Pauli twirling set.
3rd Place: Using Grover’s Algorithm to Find Collisions in Hash Functions
Our third-place team explored an application of Grover’s algorithm in cryptography to find collisions for hash functions.
Grover’s algorithm is used to expedite the time it takes to solve unstructured problems with traditional classic searching algorithms with an oracle function and amplification. Through amplification, the algorithm will increase the probability amplitude for the correct entry while simultaneously decreasing the amplitudes of all other entries with multiple iterations.
Hash functions allow us to return a unique value for a given input. This is a useful tool in many implementations that require input simplification and verifying uniqueness; for instance, SHA-256 is a popular encryption protocol that returns a unique value by hashing the contents in a file, allowing us to verify the data’s integrity.
In the context of Grover’s algorithm, a hash value can be analyzed to find all possible inputs. A collision would occur if amplification led to multiple states having equal nonzero probability on a histogram. Though their hash function was a simple one, the team’s project showed the potential advantages quantum computers might have for collision detection once quantum hardware evolves to handle more advanced hashing algorithms.
