Finding optimal control sequences with control theory

Control theory is about finding control sequences to control a quantum system in such a way that it performs operations optimally. The optimization can, for example, relate to time – operations should be executed as quickly as possible – or to robustness – operations should be as little susceptible as possible to fluctuations in the environmental parameters. “The nice thing is that the algorithm itself is platform-independent, but in the end you put it on a platform and try it out,” explains Santana. Santana models the optimal pulse sequences using theoretical models before testing them experimentally on the hardware. He is currently working closely with the Walther Meissner Institute (WMI). “The WMI now has a cloud connection, for which I am basically one of the beta testers,” says Santana, describing the process: ”I prepare my experiments and am on a call with the experimentalist. Then I send over my experiments, he makes sure that everything arrives correctly and then I get my results back.” The big challenge is then to interpret the results that don't quite match the theory. Drawing the right conclusions from this for the next steps is anything but trivial: “If building blocks in theory demand a certain feature, you have to think hard about how to do this on a hardware system. You have to think of something clever.” The computer scientist is particularly pleased that, as an MQV scientist, he has dedicated access to the systems in the ecosystem without having to put up with long waiting times. “That's just cool! You just run the experiments, get the results back and then you can get straight on with it,” he enthuses.

“Control theory is a necessary step in the development of almost all quantum computing platforms,” emphasizes the scientist. You can only get so far with the mere calibration of gates: “Then the system parameters change a little and the next day your gate no longer works well. You can combat many of these things with control theory.” Compared to error correction, another topic in the field of quantum information processing, the aim is to avoid errors in the system from the outset. “After that, you can add error correction. Or you can combine the two a bit, which is what we are also currently working on,” he explains.

Developing standardized pulse interfaces

In addition to designing optimal pulse sequences, Santana is also working with various members and partners of MQV on interfaces that make it possible to run pulse programs directly on different hardware. For quantum circuits, MQV has already been and is intensively researching a standardized interface that manufacturers of quantum hardware can implement. Pulses are one level deeper. In the case of superconducting hardware, for example, the pulses are microwaves; in the case of hardware based on neutral atoms, they are laser beams. “For us in control theory, it is important to bring such pulse programs to different platforms in order to run our experiments,” explains Santana. “For example, if I've come up with a really great theoretical gate, then I have to be able to run it on the platform somehow. The aim is to define a pulse interface, analogous to the case of quantum circuits, a standardized language, so to speak.”

The initial impetus to think about such a pulse interface came from Santana at a cross-sectional meeting in which the Munich Quantum Software Stack was presented. Attending various lectures and colloquia, listening to what is happening in areas outside his own research focus – the 33-year-old likes to take advantage of this opportunity whenever he can find the time. “I simply asked how it was with such a pulse interface,” he says looking back. Now it's a matter of first coming up with the interface and then implementing it on a test basis. Standardized interfaces at different levels are then also interesting for the industry, especially hardware suppliers. For Santana, there is also a personal reason: “As a software person, I simply like interfaces and standardized access. That's very important to me,” he says and laughs.

Driven by a thirst for knowledge

With his background as a software developer in industry, Santana also brings his own perspective to the table. Programming for science is often simply different, tailored to a specific experiment. In his training, however, the focus was on writing software that is extensible, maintainable and reusable. The transition from “writing perfect software” to “writing good software” was a major challenge for him, especially at the beginning. “I wanted to do everything perfectly, that's how I was trained, that's how it had to be,” he says. He first had to learn that less is often enough for an experiment. However, the fact that he can usually reuse and adapt the code he has developed for experiments is another benefit of the perfectionism he acquired in the software industry. “I used to develop software to sell it. Now I develop software to use it as a tool for scientific activities. And I enjoy that immensely. At heart, I simply love developing software,” the 33-year-old sums it up.

The fact that he now works in research is something that fills Santana with pride. “I never thought I would go into academia,” he says. “I used to have really big problems with school.” Back then, he felt it was too authoritarian and restrictive. How comfortable he feels today in his specialist field, which he discovered more or less by chance – he still seems a little disbelieving when he talks about it.

Santana cannot yet tell what the future holds for him; he does not yet have any firm plans. “Should I start a Ph.D. or not, that's the main question on my mind at the moment,” he says. It would really tempt him, but as a young father he wonders whether he can and wants to find the time for it. Family versus career – he doesn't really like this cliché, but he finds it difficult to juggle everything. The fact that, at 33, he is later than most people who start a Ph.D. also bothers him. But not too much. “I look forward to my birthday every year, I love getting older,” he emphasizes. “I found my first gray hair the other day, I was very proud,” he adds with a laugh. Whatever happens, Santana is very happy at the moment and feels at home at the DLR. He doesn't really care much for big dreams anyway: “I prefer to think from small dream to small dream.”

Santana is always motivated to keep learning, try out new things and develop his professional skills. He likes to start his days by “calculating something for fun. So that I can shake it out of my sleeve later,” as he describes it, and it becomes clear how much he enjoys his work. The entire environment at DLR inspires him immensely and motivates him intrinsically. “Satellites are simply much more exciting than cars,” he says with a laugh, referring to a previous employer. In addition to his work in the quantum field, Santana also works in satellite operations in telemetry. Using machine learning methods, he analyzes the large amounts of data sent by the satellites every time they fly over the ground station in Weilheim in order to make risk predictions about the satellite system. Then it's back to programming quantum systems. If his colleagues have problems of a mathematical nature, they come to Santana, who implements them. And when everything gets too much for him, he goes back to the control bridge: “It's dark on the bridge and you can get away from everything for a moment, which helps tremendously. Then you come back after a few minutes, look at your problem and think to yourself... Ah!”

Published 29 November 2024; Interview 12 August 2024