The Yale Undergraduate Quantum Computing Club wants to deliver food via algorithms.
A few weeks after returning from a quantum computing hackathon at MIT, Brennan Lagasse ’26 opened an email inviting him to another in Abu Dhabi. “I thought it was a scam email, as one usually would think, where you win a free trip to the UAE,” he remembers. “Turns out, someone there liked our project.”
At the MIT iQuHack, Lagasse and his teammates from the Yale Undergraduate Quantum Computing Club had placed third in the 2024 Challenge sponsored by IBM, one of the leading developers of quantum computing. Their task there was to demonstrate quantum utility—the goal of making quantum computers as efficient as classical ones in solving complex problems.
Lagasse’s proof of concept was what won him the invitation. But when he arrived at NYU’s campus in Abu Dhabi, he discovered his task would include an additional goal. At the 12th annual International Hackathon for Social Good, all projects were required to “align with the UN Sustainable Development Goals.” It was a vague stipulation that conveyed a simple statement of intent: quantum computing, a technology with as many skeptics as believers, must not only function efficiently—it must also be humanitarian.
Lagasse’s team developed an app for drug delivery in the Middle East. “A lot of people in my team had significant problems getting medications in their home countries,” he explains. His teammates, who include Jordanian, Nigerian, and Palestinian students, often required relatives traveling abroad to buy medications that are hard to obtain domestically.
“The idea was to create a network of travelers who would pick up things when they’re in one country, then bring them to other places,” Lagasse says. Once near their home country, strangers could help deliver the medicine by car. “Then you coordinate all the logistics.”
That’s the hard part. Calculating the best route for just one driver is a herculean task. It requires accounting for the location of customers, the layout of the streets, average traffic time, and gas costs. Coordinating an entire fleet of drivers is harder. So Lagasse’s team turned to computational math: an algorithm could determine the cheapest and quickest way to deliver medicine.
This calculation is known as the Vehicle Routing Problem. It’s the mathematical puzzle that delivery companies like Amazon and FedEx have sought for decades to solve—a puzzle that, with every piece toward completion, delivers more food to the needy, transports more people between stops, and saves more money for cities and corporations to the benefit of citizens and consumers. It’s a problem of great humanitarian consequence. And it seemed quantum computing might have the solution.
—
The Vehicle Routing Problem is a relatively new iteration of an age-old problem. Its most direct ancestor is a formulation developed in the 1850s by the British and Irish mathematicians Thomas Kirkman and Sir William Rowan Hamilton. Their formulation became known as the Traveling Salesman Problem.
The gist of it is simple: you’re a salesman who needs to travel to some collection of cities in as few steps as possible. You can visit the cities in any order, and you must find the shortest route connecting them all. That route—which, depending on your goal, could also be the quickest or cheapest—is the “exact solution” to your Traveling Salesman Problem.
The math used to find that route isn’t simple. Computer scientists say the Traveling Salesman Problem becomes exponentially harder to solve as it grows in size. You could easily find the best route to a scenario with only a couple of cities. But try counting them in the dozens. There are more routes connecting the 48 state capitals in continental America than there are grains of sand on Earth or stars in the observable universe. You could try using the best supercomputers we have, but by the time you finish charting every route, the sun will have died—and even that would be ancient history.
The computer’s limitation isn’t that it can’t find exact solutions to large problems, but that it takes too long. That becomes all the more apparent when solving Vehicle Routing Problems, which are Traveling Salesman Problems made more complicated by routing multiple salesmen at once. And when you begin accounting for traffic and weather conditions, driver availability and salaries, towing capacities, order of deliveries, delivery time windows, fuel consumption, and emissions, the math quickly spirals even further out of control.
Computer scientists rely on heuristics. They’re mathematical rules of thumb that ease the calculating burden on computers. Ben Foxman, a second-year Ph.D. student at Yale studying computational complexity, offers an example of a clever Amazon delivery driver. “Suppose there’s one neighborhood where all the houses are arranged in a big circle, like a cul-de-sac,” he suggests. The driver needs to visit each of them. “We intuitively know,” he says, “that if you have a bunch of houses in a circle, the fastest way to deliver the packages is to go one by one. You go around the circle once and deliver packages along the way.”
That’s his rule of thumb for circles. There’s no need for further calculation. “That intuition allows you to develop a faster solution than just brute-force checking all possible paths,” he says. Heuristics cut down on computing time. Combine enough heuristics, and problems that would have taken untold millennia to solve become solvable in minutes.
Sometime in the ‘90s, heuristic methods began sparking small revolutions in delivery services. There were measurable changes in the waste disposal industry alone: communities in Hempstead, New York, were saving $200,000 a year on routes determined by computers. In Florida’s Metro-Dade County, disposal crews made ten to fifteen percent more stops in the same amount of time. For one company, that newfound efficiency led to four layoffs on its Ohio routes. In the U.K., one brewing group enjoyed not only fewer hours on the road but also fewer missed orders, shorter operation times, and better adherence to schedules. Similar trends emerged internationally in industries that deliver food, soft drinks, dairy, newspapers, and groceries. Within a decade, deliveries to many thousands of consumers and retail outlets had become demonstrably more affordable.
But the tradeoff with heuristics is that they don’t search for exact solutions. The solutions they find are often close to exact but can be off by margins of a percentage point or two. At scale, these small percentages can transform into immense losses. For large delivery companies like Amazon and FedEx, even a percentage of optimization could represent hundreds of thousands of dollars every day in fuel costs and labor hours.
The same principle holds for smaller hunger relief organizations. Money saved in transportation is money saved for food. In other words, if you could reduce the mathematical margin of error, you could accomplish something of great humanitarian significance.
—
Back from Abu Dhabi, Lagasse introduced the Vehicle Routing Problem to members of the Yale Undergraduate Quantum Computing club. He formed a research group. Could quantum technology find accurate solutions more quickly than classical computers?
The trick is to work with probabilities. Instead of storing information with bits of ones and zeros, the way MacBooks and Windows do, quantum computers use jumpy qubits that, until you measure them, can represent ones and zeros simultaneously. It’s a little like flipping a coin in space. The coin keeps spinning until you grab it. Then it chooses heads or tails.
In the spinning state, qubits can process more information than bits. A classical computer needs eight bits, for example, to plot out one integer between zero and 255. With eight qubits, a quantum computer can simultaneously representeveryinteger up to 255—even if the final measurement collapses everything down to one outcome. That explosive processing power is why drug developers have begun using these machines to simulate complex molecular structures. Investors and urban planners are also exploring near-future applications where quantum computers could model financial trends and predict traffic flow.
The World Food Program, the branch of the United Nations that delivers critical food aid to needy and conflict-ridden regions of the world, is also studying quantum applications. The WFP has a large fleet of vehicles that daily relies on solutions to the Vehicle Routing Problem. Its operations include 5,000 trucks, 20 ships, and around 80 aircraft. On average, it delivers 44 million rations per day. “When you are hungry, every minute counts,” says a video embedded on its homepage. It partnered with CERN, the particle physics laboratory in Geneva, in 2023, and has since continued to work with quantum research groups, including the one at Yale that Lagasse started.
—
Lagasse met with two of the World Food Program’s logistics representatives. At the time, the organization was scrambling to make up for the Trump administration’s drastic cuts to USAID. They were preparing for a loss of billions of dollars in food aid. “You could see that they were very concerned about things they would be able to do in the near-term future,” Lagasse says. The Program’s concern for the immediate future was a sharp departure from the Quantum Computing Club’s more speculative research. “Our project is definitely more long-term oriented,” he admits.
So is quantum computing as a field. No one can say for certain when quantum computers will finally achieve fault tolerance, meaning they can reliably perform computations without error. The technology is naturally prone to error. Qubits offer tremendous processing power, but their ability to represent ones and zeros simultaneously is inherently unstable. It makes them unusually sensitive to their environment. That’s why quantum computers are typically kept in large, isolated refrigerators at nearly absolute zero—a temperature colder than in deep space. Open the fridge, and even a stray Wi-Fi signal could knock a qubit out of its spinning state and flip it to a wrong zero or one. Cascading mistakes can transform a quick answer into nonsense. Qubits easily create math typos of the worst kind.
Lagasse’s research sought to write more efficient algorithms. Generally, the simpler the formulas, the less chance qubits have of making mistakes—and the less processing power is required. Rohan Wassink ’26, a member of the research group, describes an inefficiency he noticed in their algorithms. “The formulas we had been using allowed a vehicle to be at more than one place at a given time,” he says, “which is not physical.” If a problem had one car and twenty depots, the algorithm ran twenty qubits to check each depot, “even when nineteen of them are going to be zero.” It was a waste of computing power. “My idea was instead to look at the number of vehicles,” Wassink says. By encoding the problem as a string of stops, where vehicles can only visit one place at a time, “you won’t run into the problem of having one vehicle be at multiple places or multiple vehicles be at the same place.”
The reality is that quantum computers aren’t ready to solve Vehicle Routing Problems at a commercially useful scale. Some doubt they’re even needed. A leading expert on heuristic approaches to the Vehicle Routing Problem, Professor Frederick Semet at France’s Centrale Lille, is among the skeptics. “In the current state of literature, there are some first approaches to solving the Vehicle Routing Problem with quantum computing,” he explains, “and the number of customers they consider is very, very low. I mean, it’s less than ten.” By comparison, classical computer scientists are today competing to solve challenges with 10,000 customers. “It’s just not in the same order of magnitude,” Semet says. “I’m not sure that we need quantum computing for the Vehicle Routing Problem.”
—
There are about a dozen quantum computers on the fourth floor of Dunham Laboratory on Hillhouse Avenue. The machines hang from the ceiling in cylindrical refrigerators half the width of a tube slide you might find at a playground. Inside are smaller tubes of the same shape, one inside the other like a metallic babushka doll. Together they help cool the cocooned computer. The coolest tubes are maintained at eleven millikelvin. That’s a fraction of a degree Fahrenheit away from the temperature at which all atomic movement—except on the quantum level—ceases.
The actual computer bears little resemblance to its classical counterparts. It’s a golden chandelier of wires and circular metal frames that looks more like a piece of space-race machinery. When it’s cupped by the refrigerating tubes, resuming operations, it lets out high-pitched chirps similar to the rhythmic creaking of a treadmill. Each computer costs a million dollars. On the fourth floor alone, the monthly electricity bill is $20,000.
The Yale Quantum Institute maintains the machines. The Institute was launched during Peter Salovey’s presidency as a hub where Yale’s quantum experts can share research. Its founding members include Michel Devoret, Professor Emeritus of Applied Physics and one of the three recipients of the 2025 Nobel Prize in Physics. It has received funding from the university, the state of Connecticut, and federal agencies. In 2018, it received $16 million from the U.S. Army Research Office. In September, it became one of fifteen finalists seeking the National Science Foundation’s competitive ten-year, $160-million grant.
Grand strategy helps explain the government’s investment in quantum computing. “Science and technology have been the source of American economic growth for decades,” says James Andrew Lewis, a former member of the Senior Executive Service under Clinton and Bush. “It’s part of a pattern we’ve had as a government.” Lewis is now a senior advisor on economic security and technology at the Center for Strategic and International Studies, in D.C. His writings there informed Obama’s first remarks on cybersecurity in 2009.
“Everyone loves the race analogy, that it’s a race with China,” Lewis says. “The Chinese believe it’s a race too.” Both superpowers want to develop the first fault-tolerant quantum computer. If that happens—if tomorrow either nation unveils a quantum computer of unseen capacity—“all standard encryption will suddenly be vulnerable.” Its processing power would overwhelm current cryptographic defenses. That means leaks in emails, banking information, and browsing history. It also means intercepted financial transactions and stolen government data. It would herald the quick and brutal death of modern cybersecurity.
Lagasse knows this. He knows that the antagonistic governments of the world are intercepting and saving each other’s coded messages for a time when the technology exists to decipher them. “There’s a scheme,” he says, “called save now, decrypt later.” It’s the scheme of pirates: hoarding treasure in chests as their locksmith works on the keys.
But without the promise of better quantum computers, his research is for naught. “You don’t have a long history of movies showing Arnold Schwarzenegger destroying the human race because he’s a quantum computer,” Lewis says. “So culturally, it’s in a different space and it doesn’t excite the same.” And in any case, it’s no Manhattan Project. “You might even be able to route cars around a little bit better.”
Ryne Hisada | Yale Daily News | Original Article ↗
