Tobias Lindstrom, Head of Science for the Department of Quantum Technologies at the National Physical Laboratory (NPL) in the UK, is interviewed by Yuval Boger. They discuss the critical role of national measurement institutes in the quantum ecosystem. Tobias explains how NPL bridges the gap between academia and industry, delves into the complexities of quantum benchmarking and standardization, and explores advancements in quantum communications, sensing, and computing. They also touch on the evolving quantum supply chain, the role of neutrality in measurement services, and what the future holds for quantum technologies.
Transcript
Yuval Boger: Hello Tobias and thank you for joining me today.
Tobias Lindstrom: Hello and thank you for having me.
Yuval: So who are you and what do you do?
Tobias: My name is Tobias Lindstrom. I guess I have two hats. I’m the head of science for the Department of Quantum Technologies at the National Physical Laboratory in the UK. I’m also one of the principal scientists in the quantum computing circuits group in that department. As a scientist I work on various problems related to scaling up for quantum computing and then as a head of science I take more of a strategic role around the science we do and the type of metrology we work on for quantum technologies in general. So that includes not only quantum computing but also quantum comms, materials and quantum electrical metrology.
Yuval: Quantum technology in general is getting a lot of attention and funding these days, so what is the role of a national institute like the NPL in this industry?
Tobias: NPL is part of the National Quantum Technology Program in the UK. We sit alongside the academic efforts that are typically from the UK hubs. I know you had Professor Tim Spiller as a guest a few episodes ago and I think he described what the hubs do really well. Obviously the quantum technology program in the UK industry also has a very prominent role, both as users of technology and in the development and as investors. And I guess NPL sort of sits somewhere in between there. So we sort of think of what we do, we function a bit like the link between industry and academia.
But more specifically our role is typically about being the home of measurement. So we work with both academia and industry developing in some cases bespoke measurement solutions. We also help them with things like working on standardization of quantum technologies where we represent the UK and many other roles. So obviously quantum technologies is still an emerging field so it’s not quite at the level of established industries where NPL typically works where we provide measurement services. We do not yet provide measurement services for most of the quantum technologies because they do not yet exist but that’s what we’re working towards. So our role differs a bit depending on what we’re doing.
Yuval: If I think about a US analogy NIST, the National Institute of Standards in the US has played an outsized role in PQC standards. Are you driving any standards in the NPL today?
Tobias: We are the UK lead on standardization of all quantum technologies. So we work together with BSI which is the standardization organization that represents the UK. But we provide most of the, not all but many of the experts and we also do a lot of the groundwork in terms of communicating with industry and trying to figure out what industry needs in terms of standardization. So we are active in most of the organizations around the world. So ISO, IEC, ETSI and also CEN -CENELEC which is the European organization as well as most of the industry organizations that are interested in things like benchmarking. For example, for quantum computing there are a lot of industry organizations working on benchmarking, QED-C for example, we are a member of that as well. So we would pop up all over the place.
Yuval: And that actually leads me to something I wanted to ask earlier about measurement. So do you intend to have a quantum algorithm or quantum computer benchmarking service at the NPL?
Tobias: That’s a very good question. We do have a team which sits within my group at NPL that does a lot of work especially on benchmarking and metrics and most of that work is done together with the National Quantum Computing Center and also the Quantum Software Lab up in Edinburgh as well with industry. Whether or not we will ever provide a service is something I think is a bit uncertain. As you know it’s quite complicated to benchmark a quantum computer. So at the moment the field is not mature enough for us to say we’re going to do it as a service. We are very happy to work with industry. Someone might come to us and say they have this quantum computer or QPU and they would like our help to benchmark it and figure out how well it works. So then we have to do so.
But what we can’t do yet and probably to be honest probably wouldn’t want to do anyway is say yes this QPU is better than the QPU of one of your competitors because one of our key roles in the UK landscape and also internationally is to be seen as neutral. So we don’t want to pick favorites. We can help you with the R&D by doing benchmarking and so on. But you’re not going to sort of say you know again as I said pick winners. That said I mean benchmarking and metrics are coming up very very frequently in international discussions about standardization and also in other fora. And there I think there is a need not necessarily for services but there is a need I think to find some sort of agreement about which benchmarks and also which metrics should be used in different contexts and again for computing that’s really complicated because you know the different modalities and so on. But if you ask end users and potential end users and also ask governments and so on asking who are thinking about this and what should I invest in. One thing they really won’t help with is basically to figure out what is best for their particular application. So again in those situations we have to help but we’re not going to be benchmarking different platforms and then say that company A is better than company B. That’s not our role.
Yuval: Before we go back into benchmarking for quantum computers, it sounds like a big part of measurement would actually be sensing and maybe communications. Are you also engaged in that?
Tobias: Yes, we are, especially quantum communications, is something NPL does a lot of work on so we do and obviously that’s more mature both in terms of as a technology and also much more mature in terms of as a market because there is a supply chain. There are companies selling practical products. So yes we are providing there we are actually at the level where we can provide services and we can help companies validate their products.
That said, even in quantum communications it’s still an evolving area so there are still new technologies coming online especially more recently, for example, satellite-based QKD. So there’s still a lot of R&D to do even on the measurement side that needs to be done and understanding how to do that. Same thing goes with sensing. Sensing obviously is a very big field. In some areas it is fairly well established how you do things because frankly a quantum sensor is just another type of sensor so what you’re benchmarking is not just the quantumness if you can call it that but just how well it works as a sensor and then in a way it is much easier because they can compare it with the classical version of the same sensor. But again for some sensors especially sensors that use say entanglement and things like that it’s much harder and again that’s still very much a field that needs more R&D.
Yuval: It’s easier for me to envision what you measure in a sensor but you actually started your response talking about comms. What is it that you can measure about quantum communications?
Tobias: In quantum communications, so if you think about this as a network, when you have something transmitting and then something detecting and then you have a link in between so a lot of the work we do in the department is about things like measurements of single photon emitters or single photon detectors. Qualifications of quantum random number generators which are part of the quantum communications systems for say QKD that’s how you generate the random numbers. But there are also a lot of what I guess we could call classical measurement that goes into that so typically because quantum comms you know most of the time will use very low level signals. One thing you want is very low loss fibers and lower loss patch cables and all of these measurements need to be very precise, especially if you’re part of the supply chain selling into that field. You might come to NPL and say hey I have this fiber and I think it’s very low loss, can you help me do the measurements that can put that in a data sheet. So there’s nothing quantum about the measurements or even about the product as such but it’s just that the end users in the quantum industry or is in quantum technologies.
Yuval: You mentioned supply chain do you assist companies or the government in understanding critical components of the supply chain, where they come from, what’s at risk, in terms of geography or otherwise?
Tobias: Yes, we do I mean again this is we’re not the only ones in the landscape doing this because again we work very closely with other organizations but yeah that’s definitely one of our roles because NPL is directly owned by the government where we operate as a company but we’re directly owned by DSIT (Department for Science, Innovation and Technology) which is a government department so it follows naturally that they will come to us for advice and seek information about what’s happening in the various fields, but I don’t want to say that we’re the only ones doing that because especially in the computing field we work very closely with a number of organizations and the government, including the hubs.
Yuval: I did an episode on quantum benchmarking quite some time ago and I think some benchmarks are more on the lower level, you know gate fidelity or coherence, and some are more application related what’s the largest algorithm or longest algorithm that I could run. Do you have an opinion if one is more important than the other or you get asked about one more than the other?
Tobias: I think that this is a very good question and it’s a question that’s being discussed a lot especially in the context of standardization. My personal view is that ultimately what people care about if especially if you’re an end user, if you want to use quantum computer for something, is application level benchmarking. You want to know how well does this computer work for my particular application. I need to be a bit careful. I know people are working on these but I think everyone would agree that there is as of yet no really good way of doing that type of benchmarking simply because we don’t have practical quantum computers yet, so the types of problems you can run are simply too small to be of any practical interest. So when you do a benchmark, are you really testing a real world problem? The answer right now I say is no.
But coming back to your question I think the hardware benchmarking is more established if you do things like obviously randomized benchmarking, different types of fidelity to get the various types of fidelities, quantum volume, and all of these things that are more established. I think one thing we’ve had to do is that even if someone puts a paper out saying they did randomized benchmarking and this is what they got, it’s not. If you actually then check how they did it, it turns out the methodologies are actually still somewhat different. And I need to be clear and I’m not saying anyone is being dishonest. Even the benchmarks people are assuming are well established, if you actually look at the details it tells you that there are slightly different ways of doing these so you can get slightly different results.
So that’s one thing. Another issue which is close to my own research as a scientist is about if you especially if you look at solid state quantum computers, especially superconducting quantum computers like the ones by IBM, if you do benchmarking and say you know use the afternoon and get one value and then you run the same benchmark the following day you’re probably going to get another value. So then the question is how do you report this data and what data is correct, because the end user probably doesn’t necessarily care too much about what’s the absolute best value you can ever get they want to know what’s the type of value I will get when I’m running it for my application and and sometimes those differences are really significant. I think this is widely known in the industry, but it’s something that sometimes gets lost when people are reporting their data, in say a press release.
Yuval: You mentioned your scientific background, focusing on superconducting qubits. Are you excited about another technology? Have you seen wonderful things happen in photonics or in trapped ions or in something else that you’re excited about?
Tobias: Yeah, I think all of these modalities are making really good progress. The oldest of the modalities you just mentioned are doing really well. One thing to pick up on which I probably wouldn’t have predicted two years ago is the great progress on neutral atoms and then some because they seem to be doing really well at the moment even when you are using gate-based approaches, so I think that’s quite exciting. One of the things I do find most exciting is that all the platforms seem to be making good progress. No one seems to have hit a point where they say no we can’t develop it further and people lose interest because in all the main modalities there are companies doing really good progress and are coming up with exciting results very frequently.
So I don’t know if that answers your questions but again if I could pick one it’s probably neutral atoms, but superconducting quantum computing is making good progress and ion trapping is doing extremely well. There are lots of companies, especially in the UK, there are lots of spin-out companies working on photonic quantum computing, so that’s doing very well as well. So yeah, there’s lots of stuff going on throughout the whole industry.
Yuval: And it seems like photonics and superconducting or trapped ions are merging a little bit once you talk about optical interconnects between multiple QPUs. Is that also an area that you’re working on at the NPL?
Tobias: It’s something we started looking at. So quantum networking is definitely something we’re really interested in. We’vee already done a lot of work on quantum communications. And we also do work on quantum computing so this obviously would be a nice fit. There are also quite a lot of interesting measurement problems there. One thing you certainly will need to do if you have a superconducting qubit and you want to convert to optical frequencies you need transducers and how you actually do the measurements of the transducers and all of these things is something we’re really interested in both on the measurement side and also in terms of working with various collaborators and developing new technologies for transduction.
Yuval: We’re recording this talk towards the end of 2024. Let’s fast forward say three years from now. What would you like to see? What would you like your department to achieve or to see a couple years into the future?
Tobias: I think I would like to see it would be very good I think if we reached a point where at least some companies are coming to us on a sort of fairly routine basis to do some validation of their products. This is starting to happen and in quantum communications this is more established, but if you’re asking about the computing side it would be nice to get to the point where a company can come to us we can do some testing and then we can then get back to them and say yeah these are the results and this is the measurement data, these are the uncertainties, and so on. It’s going to take longer than that to get to the point where we can you know potentially provide people with sort of calibration certificates because those type of services take a long time to develop but yeah that’s probably where I would like to get to on the quantum computing side, where we reach a level of maturity where there is some agreement both among the measurement institutes of the world but also among industries of what you want to measure what you want to show your potential customers.
Yuval: You mentioned collaboration with the QED-C around benchmarking to what extent do you collaborate with other European or even Far East measurement or National Institute of Standards?
Tobias: We work a lot with national measurement institutes from across the world. At the moment we have projects with NIST, PTB which is the German NMI, INRiM which is the Italian NMI. Most of the European countries, so France, Denmark, Norway. And Japan and South Korea. One of the nice things about doing this type of work at an NMI is that it is in the DNA of a national measurement institute to do collaborations because if you come to us and you want us to calibrate something – and I’m not talking about only just quantum technologies now but just general you want to calibrate something – and we give you a certificate. One of the things we guarantee is not only have we done the measurements but we have also checked that our measurements agree with the measurements among the other NMI’s across the world so we do a lot of what’s known as intercomparison so we exchange samples and so on and make sure we all get the same result. So that means that we already have very very strong links to these organizations and we are now seeing that in the area of quantum technologies we’re now picking that up, so that as those NMI’s also start getting involved in quantum technologies, including quantum computing, those things are already there and then it’s quite easy for us to start collaborating with them in that area as well.
Yuval: You mentioned how you’re helping industry. How can industry help you or your department?
Tobias: I think there are a couple of things that would be really useful. I should emphasize we do have very good links with industry but we’ve found that especially and this is now I should stress this is changing as the industry is becoming more mature but many companies especially in the beginning are if you’re a startup you’re really protective of your IP, you don’t necessarily know what you can and cannot say and you don’t necessarily trust an NDA. Frankly you don’t necessarily know if even though you have an NDA can you really tell people that. So we found it quite difficult sometimes to work with industry and get them to tell us what they actually need. There are some good reasons why they want to keep their roadmap secure and not tell us everything, but it’s slightly frustrating sometimes when especially in the context of standardization we go to a company and ask what kind of measurement services do they think they will need in a number of years when they’re a bit more established and think they’ll be selling products and what can we help them with, and they don’t really want to answer or maybe they do want to answer but maybe they’re very worried about sharing too much. And I should say that when we do this in more established industries, we can talk to really big companies, sometimes about products that are worth a lot of money and they are typically a bit more open so that’s one bit.
I think the other bit is it would be nice to get more direct involvement from the quantum industry in international standardization and I think some countries are doing this really well so you go to international meetings and you find quite a few SMEs from say France and Germany and so on but in some countries and unfortunately that includes the UK although again they’re starting to change they haven’t been as interested in the work on standardization. I can sort of understand that to some extent they probably don’t say okay our focus is on developing new products and we don’t really see the value of standard yet, but at the same time it’s really the case that now the standardization is starting if you’re not at the table from the beginning you don’t have a voice, so you risk losing out in the long run.
So I’m going to ask industries to be a bit more open about what they actually need not just to us but more generally and also to try to get a bit more involved with standardization. But things also are slowly starting to change.
Yuval: When I last visited NPL there’s a tree outside in the garden which is said to be a descendant of Newton’s apple tree so putting Newton aside I wanted to ask you a hypothetical if you could have dinner with one of the quantum greats dead or alive who would that be?
Tobias: I’ve listened to your podcast for a while and I know some of your previous guests have already said some of the names that would be high on my personal list as well, but I think Leo Szilard would probably be someone I would have liked to meet, not sure if you count his counts as one of the quantum greats but he definitely made some really fascinating contributions especially around the relationship with quantum information and statistical physics and thermodynamics, and I will say I think he was a fascinating person because he had a really fascinating career when moving from physics to biology and obviously his involvement with the first atomic bomb. So he had a very fascinating life. I would have loved to sit down and have dinner with him towards the end of his career and discuss all the things he had seen and done and all the people he had met.
Yuval: And as far as I can tell, you are the first guest to mention his name.
Tobias: So okay that’s good.
Yuval: So, Tobias, thank you very much for joining me today.
Tobias: Thank you for having me. I really enjoyed this discussion. Thank you.