Transcript for:
Quantum Photonics Lecture by Professor Christine Silberhorn

hello everyone uh this is deepthurstree the coordinator of pix2day webinar series i cordially welcome you all on behalf of co ec pix iit madras i would like to exchange my heartiest welcome to professor christian silverhorn she is the chair of uh integrated quantum optics group in paterborne university professor christian we'll give a talk on quantum photonics before having this most demanding talk i would like to request professor vijay krishnadas to introduce our honorable speaker i'll also request professor deepa venkatesh to moderate our q and a session over to professor plus yeah thank you deep this way and i also welcome everyone and uh it is my pleasure to introduce christine silvaron professor christine uh of course she is a german physicist specialization in quantum photonics and currently heading integrated quantum photonics group quantum optics group i would say that university of paderborn i think uh just wearing the legacy of professor scholars which was formerly known as physique so from where i also got my phd degree and uh yeah silverhorn is best known for her role in leading research projects which develop mostly uh quantum devices and systems for quantum computation and quantum communication so in 2005 silverhorn branch of max planck institute for quantum optics gassing heading the junior research group uh junior research group namely integrated quantum optics again until 2008 where she completed her habitation i think that is a very honorary degree uh to enter if you want to be a faculty professor in german universities you have to complete that habitation degree so she had it in 2005 and then in 2008 uh onwards uh she was leading the max planck institute for the science of light in london that was established that time and then in until 2010 she continued there i think then afterwards she joined in university i i think yeah so her research work has been uh awarded by several prizes i think prestigious prizes like uh got fried wilhelm leibniz price in 2011 and 2017 her team was awarded with the prestigious european european research council award a consolidator project grant for quantum photonics research she is also elected member of the leopoldina which is also prestigious in northern west balian academy of sciences humanities and arts and also she is a fellow of the optical society of america so she became a fellow for the max planck school of photonics in 2019 and yeah her work in the in this area really actually really actually is you can just imagine for and i with this short introduction i invite professor christine uh just to take over and also in the panel we have uh professor tagarajan who is also our mentor from whom we learned photonics and uh also well known to christine i think uh you can put up all your q and a in the chat box as it instructed and deepa is the moderator and maybe 40 to 45 minutes after her talk we will have a very useful interrupting q and a sessions and also our cto arnab goshami is there she can also participate in the pilot discussion with this i just welcome a professor casting please over to you yeah well let me first try to start my presentation um one minute should work we tested it okay i hope you can see it now yes yes okay so let me start with thanking you very much for the nice introduction and also for that invitation and giving me the opportunity to present my work here um now the presentation i entitled quantum photonic using nonlinear integrated optics and pulse glide it's a pleasure to present you that work and it's basically an overview what we're doing in my group before starting to really go into details into these topics i would like to give you a physical motivation about the things we are doing and one of the core ideas is we want to build quantum networks why quantum networks well quantum networks i consider as an ideal model system for having a complex large scale structure and quantum networks of course one thing which you probably immediately have in mind is the quantum internet and of course this is also an important topic but it's going beyond that if you want to understand transport phenomena or neural networks you see also these network structure is in there and the question is what does these networks make quantum now this is actually the question which is very easy ones but i hope by the end of the talk you will understand more what i mean by that um secondly i'm a photonics person as you have also heard already and if you're coming from photonics what we are trying to do these days is to establish multi-dimensional photonic quantum systems and ideal really very multi-dimensional that means large-scale quantum systems why is that well one of the motivation is to have photonic computation and simulation um i will come back to that in a minute and we had a breakthrough experiments there from china in 2020 and but multi-dimensionality also plays a role in our quantum technologies like metrology where we're looking at multi-parameter estimation and last but not least we're talking about photonics we are talking about uh light as information carry and of course communication is also an important topic and it's interesting to note that also their high dimensional information encoding is an important topic which i also will address let me say a few words about photonic quantum computation and photonic and computation um where people started to look more seriously into that i think was around 2001 where nila from melbourne suggested their loqc this was a very important milestone from the theoretical side but it was also clear that the overhead we need to implement such systems is huge this is nowadays phrased different oil elaborate on measurement based quantum photonics and we still want to do these things to establish universal quantum computation the same time we are trying to figure out systems which are easier to implement and have a more near-term future and this brings me to the topic of command and simulation and here i will show you some of our works in quantum walks which are targeting towards quantum simulations and understanding the foundations of that last but at least in that context i want to mention both on sampling boson sampling is quite interesting from a photonics community because it's a rather simple system and here you see what kind of system it is you have a unitary matrix that means a linear network you import n single photon states in the original proposal and you just detect at which output ports the photons come out surprisingly this rather simple looking experiments can't be calculated computationally so you can by demonstrating this system you can demonstrate quantum computational advantage and this was the experiment which i was referring to for quite a computation uh which came from china in chiang mai punks and show young loose group it was a milestone for quantum photonic one simulation because actually a system was implemented which can't be calculated classically and this was quite exciting and it was particularly also exciting for us i have to say because if you look what kind of scheme they really used it was not the original classical boson sampling experiment something which we now taste called gaussian bows and handling and then instead of single photon states as the input states they used input squeeze states and um why did they do that it is because it's easier to scale the system in terms of photon numbers by using these squeeze states which are easier to produce and have can have high brightness we were looking at the topic words for quite some time to deliver with excuse in 2017 we have suggested that instead of using the standard approach from boson sampling where people are always looking at single photon input states that you can use such gaussian single mode squeeze state as input states they have a very good scaling experimentally so scaling means that you need fewer resources to implement that and you still implement in this system a system which is equally hard to do classical or cultural simulated classically so here you measure half means advance it was very exciting to see that this works and you can also say that you have to come up with clever ideas to scale up your systems and this is actually a topic which we follow in my group in different ways now let me introduce briefly the team the integrated quantum optics team is a team which has a different aspects different subgroups first of all you heard already um has been known for quite some time for integrated photonics and we have um this integrated photonics um on on-site fabrication facilities and i have a subgroup led by christoph eidner where we use all kinds of photonic devices in our own facilities then we bring these divide these structures um to the optics labs and we built devices from that um the group leader here is highlight helmet and there we really build up optical circuits we model things and we really show quantum devices and we'll see something for that too uh last but not least we also have the quantum networks group uh headed by benjamin braid and here we take the opposite view we kind of look what kind of theoretical proposals are out there and what can you be implementing there are there are quantum works which is related to the quantum simulations and ultrafast pulsed quantum optics is an important topic and you see that this all is in a circle and this is not a coincidence because you also see the philosophy we're pursuing at the end of the day if you really want to implement these quantum technology applications they really rely heavily that you have a good technology which is tailored to that needs and you can have all these different aspects in one place and this is what we're trying to do in parliament now i told you that one of the challenges i see is to have good strategy for scaling this photonic system and in my talk i want to present give you a glimpse in all of these different topics how you can scale photonic systems the one the first one is indigo optics integrate quantum optics and then i want to show you our work on in that area secondly i will talk about time multiplex quantum systems uh because these systems uh really scan very nicely in terms of experimental overhead and last but at least they also want to talk about temporal modes of light you might not see at the moment why this is scaling it's coming from fundamental fines and how to combine ultra-fast physics with quantum physics but it's also a new way how we try to make our systems larger in terms of having a harmonic dimensions we can encode information onto let me get start with non-integrated quantum optics and our work in this area now let me start it because also has a long tradition in integrated optics and i should say when i came to paderbon there was a very strong group professor solas group already there and we were kind of had to learn from each other and what i learned as a quantum optics person so i'm really coming from the background of an optics there is a lot of things which i have to learn from integrated optics and actually not only learn um we really build a lot on these uh expertise which is there that it's a good thing so i want to start with an integral optics introduction it's actually from 1969 so you see it's an old overview article the interesting thing is that this old article you can make very quickly in a very modern review if you want you just add between integrated optics the word quantum you sometimes replace um the word laser beam circuitry by quantum beam circuitry some places you have to replace lasers by photons well i don't want to go into detail but this is hard but anyhow um if you do that then you can come to a conclusion and already in 1969 people had the same vision if realized this new art would facilitate isolating quantum circuit assembly from thermal mechanical and acoustic amplitude changes through small oral sites economy should ultimately result and i think this should be very modern for you in terms of technologies what are we doing these days what is new well we keep all the good ideas which is behind the scenes of integrated optics but we adapt them to quantum optics well this adoption is not really easy because you have to tailor your system there's a lot of technology work which has to go into that and i have to say over the last 10 to 15 years we have been doing that and this is good fun to do that but it's also work which you have to do if you really want to succeed now this uh is another review i'm presenting here about photonic random simulators from alum aspect and philip alter and from 2012. so this is the vision how you can imagine an integrated optics photonic quantum simulator would look like so it's a hybrid integrated network where you have a source bit a circuit bit and a detection bit now if you look a little bit close the circuitry is linear optics so couplers differently while source is always you need if you have quantums of some non-union action and if you build a system like that the advantages is that you have many spatial modes available so the scalability is there which can be rather simply controlled you get the interferometric stability of large optical systems which is important if you want to scale systems up you have compact devices you can minimize your devices easy operation well if it's operating as you want you have to develop it but it can be easily operated and one of the challenges that we need high efficiencies for that now in the linear part integrate photonics has really become a very hot topic i think it's fair to say that it started in bristol uh with the around 2008 and now i just cite european papers but i think there's all over the world other work have been done in rome uh or also in oxford and nowadays i think integrated photonics has become a whole field most of the voxel is still doing really the lineart exhibit and the thing which we have been focusing over the years was non-linear integrated optics and in particular what we are looking for is um to look for material platform where this non-linear degree optics can be harnessed in the best way which is the lithium niobate platform the system we are looking at we build our system still on rather established technology because we have extremely low wave guides that have minimized the clearance and for all quantum optics experiments it's essential we get the stability and the scalability uh in contrast to many other platforms and it's we our platform has a cartoon linearity so we have highly efficient photography generation and equally important you have fast photon routing and this is what's becoming hot topic nowadays because this fast photo rooting is needed and lithium nail bed is an established material system to accomplish modulated systems and what we are trying to pioneer is the implementing of complex quantum circuits on this platform now for the source bit we are typically using parametric dom conversion sources where we have a waveguide and a periodically poled structure we pump this waveguide by pump light and due to the non-injection um a pair of photon signal and other photon pairs are generated in order for this process to happen we have to have energy momentum conservation such the frequencies of the signal and idler photons have that under the frequency of the pump the same has to be true for the propagation constants and there the periodic polling is important this allows us to have this momentum uh phase matching in other words well the momentum conservation otherwise the classy phase matching i'm not going to detail um but we have therefore we can kind of customize our systems there having waveguide allows us to have a strong confinement of all optical waves so we have we can have long interaction life and very efficient processes and the samples we're looking at can be as long as 90 millimeter in our case now you can ask what kind of different elements can you build on that platform and this is kind of trying to categorize them we have on this platform can have passive routing like in all integrated optics um lithium iodide has been used for photograph generation but we also did a lot of work on that up and down conversion for transfusing a different between different wavelengths and also what memories have been demonstrated in this platform very importantly you can have active manipulation with electro optic switches and last but not least one thing which typically quite frequently don't talk about but you need also interfacing so either fiber waveguide coupling or dielectric coatings for example to build up resonators in these systems now all these things have been developed and most of them for classical optical applications but the thing we need to do is we have to tailor them now for quantum applications or redesign optimizations for example in quantum optics experiments quite frequently we hunt very short waveguides this is simply because the coherence is so important that we have to reduce the losses and this was when i arrived in powderpoint people were quite surprised that i didn't want to have the long wavelength when we get along more non-linearity but a shortwave guide because this is quite frequently beneficial for quantum experiments now i can't go into detail about all our devices but i would just give you a kind of an overview what kind of things did we do we had we have produced the plug-in placing the photon source single photons we had noon state generation on chip we have the generation of polarization entanglement which is kind of i call the clear type source for those of you are interested that so we are generating tangled photon pairs on chip with the same frequency two spatial modes with cascaded nuclearities we also have generate on chip three photon states um we have been building optical squeezing using waveguide resonators narrowband photon pair sources by adding waveguides of combining waveguides with specifically designed mirrors lately we also put waveguide resonators together with integral modulators to tune them with electro-optic tuning and we are looking at quantum interferometers uh for applications something like su-1 into quantum information now one of the most advanced systems we have been implemented was a hormonal interference on chip and this i want to present you a little bit more into detail i think many of you know what hormonal interference is it's if two photons are imping on a 50 50 beam splitter we have one interference you see the four classical possibilities for two particles who are kind of uh redistribute via this beam spreader because of destructive quantum interference the possibility that two photons leave at the same prime space cancels out and this is the whole famous hormonal experiment um it is the basis of measurement induced non-near linearities uh it's a fundamental quantum property um it's at the heart of linear optical computing or any networking the experiment of hormone typically in free space is done such that you have a photon pair here in the mods a and b and ping in the 50-50 printer and to introduce a variable delay and look for coincidence detection the trace you expect is if you have a large delay that you have not not the effect that both photography at the same output part so the coincidence uh occur while if you have zero delays no coincidence detection is done now one of the tasks we want to do what you do is a benchmark device how do you do these things on chip and here you see our experiments of a monolithic integration of parametric down conversion source and a two photon interferometer and a delay line of chip let me go briefly through the system so we have a pdc source of type 2 so our photon pairs actually in two different polarizations then we have electro optic control which allows us to change the polarization of these devices and actually by the bio-refrigerants modeling we are able to implement a fast optical delay and here we have a 50-50 beam spread and this 50-50 beams better than source as the hormonal interference being together in detail so we start with photo pair h and we've been generated we separate them by the integral processing beam splitter here you see the segmented another polarization convert and now because uh the photons travel with different uh polarizations at different speed if you switch the first segment you see here the 50 50 beams that you just saw the horizontal interference now if you switch at different positions here in our segmented processing beams where you see that the two ports in the upper and the lower paths travel at different speed and we effectively introduce a delay and this delay should then allow us to see the home dip this is theory next experiment depending each which segment we are switching we should see the red trace um such that get the bunching if we wouldn't have the first switch at the beginning which i haven't explained then there would be always a delay and you see the blue curve there and this is the reason how this is the way how we implemented that and this is how we expected it from theory experimental data looks very much the same and now we can kind of retrace which kind of delays you've introduced with that switching and what you can see is that we have implemented a nice homo monitor with a high visibility of over 93 percent and the time delays we can introduce in that way around 12 picoseconds this is interesting uh for us it was a benchmark device but not only as a benchmark device you can also imagine that an element like that you can have in a communication network and this will help can help you for example for source synchronization with that i would like to switch gears and i told you we are not doing only integrated optics i'm going completely to the other side now i'm talking about the networks here is another picture of our lab of our time multiplex systems and in this systems we try to get scaling not by different spatial channels by different time pin channels we interested in quantum simulation experiments and kind of one of the paradigms of the most classical ways of looking in such large scale system is by comparing random classical works now a classical random walk you probably have heard from your studies is one of the most basic elements if you want to study propagations or um it's the following we have this small little guy he walks on a line but he doesn't know what so which direction he wants to go so he throws a coin depending on head or tail he takes a step to the left or right you see here this famous golden board if you repeat that for several steps you turn on the time what you end up is at the end if you look at the distribution uh where the walker will end up independence on the different positions you get the famous binomial probability distribution it corresponds from a physical perspective to a diffusive propagation and is characterized by the fact that the variance of this distribution scales the number of steps now we can make this random volkswando by introducing a quantum coin and then the walker goes simultaneously depending on the coin again this is a two dimensional hilbert spec a step to the left and to the right and then we kind of effectively look at the same distribution but instead having at every position a decision point we put 50 50 beam splitter and we get an interference pattern well the coil is are quantum state we're dealing here with superposition states and the distribution at the out uh foot of such a device looks completely different uh it corresponds to a ballistic distribution and is characterized by a variance scale with a square root of step size and here you see quite different quite distinct difference between classical and quantum a classical random work has this binomial distribution while a quantum walk will be um has this wide spread functions and this is how quant box came up has been reduced now one might ask what makes this quantum system quantum well in the classical world we have particles travel at several paths and while in the quantum world we have here these surplus different states so the wave particles for single photos traveling through the systems makes this quantum i hope you remember that later on um this is only partially of the story why is that interesting well it's because random walks have applications for simulating in quantum computation in different ways classical processes like the brownian motion biological systems like deer wandering and actually also the google search algorithms i was told is also based on a random walk algorithms now the idea is um or this is people are for sure that quantum box can be used in the same framework we can model semiconductor physics systems with such quantum marks quantum biology is a topic which comes up over different locations and of course this is also where quantum works were introduced it's one way of developing new quantum algorithms now how do we implement such a system because we want many many steps that means many modes and the way we came up with was a time multiplex quantum mark here very briefly this is experimental setup you're using we have here a switch which can when our input states are switched in the system then we have electro optical modulators or half wave bits so the encoding for the coin is polarization with h and b um as two different coin states or coin possibilities if you want and then here you see that we introduced by a loop geometry optimization dependent delay um i will explain a minute how this works and then we feedback our system and we either detect our photons or we bring it back into the loop why is that a quantum box system well let me go to the beginning we couple in we meet the half wave plate this is the coil operation vaporization degree of freedom now we have two polarizing beam splitter and a different delay between those two and you see what's happening depending on the pronunciation we either introduce a delay or we don't include the delay this is our step so the position is done in time and the step operation corresponds to the relative shift in time in the position space now we are behind our loop geometry we feedback the same signal and you see here we come back to the exactly the same system we take the two pulses into the system again and you see that we have the case where the pulses traveled once the upper one the step one the upper and the step two the lower one and then we get these interferences and a nice thing because it's a loop geometry we can repeat that over and over again and get our quantum wall moreover i told you that we can use half wave plates or eons use electro-optical modulation this system is quick enough that only individual pulses are switched so we have full control over our corner box systems and we can simulate all kinds of things now this is one of our earliest experiments here we implemented a quantum walk over 28 steps and here you can see that this works perfectly so the red bars would be classical um and you see theory against experiment and we have a very good um quality of the data there it corresponds to quantum work if you want to do it in the special domain of the 406 beams and 29 detectors you see also the good scalability um and we see extremely good process preservation of the coherence this is because we have a very resource efficient system and the ideal homogeneous comes from the fact that we're using always the same elements now again we have been doing that for more ten years we demonstrate this time with exponent works we started d coherence and disorder elements you could also translate that to two-dimensional quantum walks we study decoherence and losses percolated graphs graph engineering state transfer then we went to more applied things we looked how we can study systems with topological invariants how we can simulate them and we looked for quantum box before the coins and also for measurement induced dynamics so we saw that this is a very nice system to study these coherence properties however um some of you might have asked in particular if you're coming from classical optics what makes this system quantum because effectively if one is honest all of these experiments uh were done with coherent light at the single photon level and then the natural question what makes the system quantum and what can if you want to go from classical optics here to some real quantum optics we need many particles effectively home dips and we have another boundaries that we do not only look at these coherence properties which also classic lighthouse but at multi-particle effects where we really enter the framework of one optics and this is what we achieved in the last years too so now we started to do this quantum box not only with individual uh input pulses but with two single photons as an input and here you see that we have to implement a specific source for that it's a parametrically dumb conversion source which gives us very nice photons we have the network which we can fully control and we showed that the green bodies are nice and then we have detection and then we wanted to demonstrate an experiment where we really can study how the relationship between coherence and quantum effects meaning quantum interference is done for that we devised a very specific network a specific experiment so we put in two single photons these were heroics in the photon states we see in that generation area then we programmed our network such that we could program two different scenarios um in both cases um the blue one and the red ones the single photons were put in the supposition states of uh uh photons impinging simultaneous and the blue and the two beam splits from the blue side on the right side um but we programmed it such that the coherence properties between these individual photons which are now delocalized over these networks are different they were different in such a case that in the one we had complete superposition which were completely equal between red and blue photons and being on these two beams plus the second one we introduced the phase change so we changed kind of um the coherence property are the first order coherence properties of these single photons and for these two different cases you see them again here here the two red pins pinging here the photo the two blue pins and the two red pins uh we looked now for different detection schemes if you look individually and we call it a time resolved manner to our system and we ignore that there's a second system there so we recorded the counts and looked for coincidences from the left side and the right side individually you see for both of the assists for the cases that we get nice home tips this is not unexpected because at the end you just neglect that there's another thing the two single photons have been impenetrable in that system well then this is what expect the second thing we did look for global interference properties and we kind of didn't care on which side of these beamsplitter the photons in pink um we only compared was it red or blue and again we looked for a global two-photo interference property and what you find is that in the first case that we again get the lights home depth this is because we have the equivalent uh coherence properties well in the second case this global interference will be completely vanished and so you can see that by controlling the coherence properties of these individual photons which are delocalized over these different sites we could control the global interference property and such control of quantum interferences might become really important for quantum networking later on now let me switch to the last topic temple mode positive light i don't know maybe how much time do i have for that okay and keep going please stop me so we have seen two ways of kind of um present a scaling up system and now i want to show you the third one we're pursuing and i'm talking about temporal modes of past light um let me first start start talking about temple mode encodings in channel so you could have time bin encodings where we kind of have different timestamps corresponds to different quantities different bins different qubits and the information is important in time you can do the same with frequency where if every individual frequency corresponds to a specific property and you see that we are leaving leaving uh living in this time frequency domain now what we're doing here is now kind of an encoding which is neither time being no frequency been encoding but it's also called pulse temporal motion coding where we encode uh simultaneously in this time frequency space why is that interesting it's a different way of encoding information of choosing a different high dimensional alphabet and if that might look strange for you uh in spatial modes you're used to that to have something very similar to this pulsed temple mode this high dimension alphabet has the advantage if you do such an encoding that it's fully compatible with wafer devices in fiber networks um and that we can really process it nicely now the problem is if you're talking about these pulse temporal modes how do you actually measure these past temporary modes and how do you manipulate them and this is where we started to look into that and where we got interested in these systems now let me start again from a kind of introduction to that topic um if you're talking about frequency modes or quantum optics you use that effectively you're talking about monochromatic creation operators um basically um we say we want to have a quantized field and before you in quantitative field you say have a monochromatic frequency and a quantize that now this monochromatic operators can't be the full truth because if you have something which is monochromatic you do a fourier transform and you immediately will see that you have something which is infinite time now if you want to have a realistic scenario you can't have something which is infinite in time but you have to introduce something which has a spread in frequency basically a wave packet as we are using it for outer first past laser system for example and this wave package is securing suppressions of different frequencies and we've introduced an envelope function if you do a void transform we have also in time now a nice uh system which is localized in time and you see that now we are talking about realistic functions our realistic objects the interesting thing if you have accepted that you can introduce envelope functions here i had the first example a gaussian envelope function we can now start to introduce different envelope functions and here you see i use their meat modes and with that we now define the feet of the temporal modes these passwords i was talking about these are orthogonal in the sense that if you integrate over them that we get 0 and effectively you can think about these um false temporal modes equivalently in terms of information encoding to different monoclimatic modes so then typically you can encode your information these are just considering a high dimensional system as you would have would be used for other systems the only difference is that the overlapping frequency in time and you have to have better sorting elements now over the years we have learned to control these temporal modes by spectral engineering and you need them because effectively if you want to have a state preparation and you want to use these parametry down conversion sources for having a good two-photon source what you want is that the pulse pump light generates a pulsed signal idler photon pair but for that you need already to control these sample modes that only these scorching modes are present and the way we kind of typically look into that is that we look at joint spectral amplitudes and we have very specific dispersion properties uh cupula matching such that we have here for example a circular system in these uh joint spectral intensity measurements now by shaping the pump pulse we can play with the transparent intensities or amplitudes and we can play with a higher dimensional system we can control these temporal modes and this is what we're doing these days now i told you that i also need a device for manipulating such systems and the way that we can manipulate them is we use another nonlinear process so for manipulating and detecting these things we now do not use parametric dump conversion but some frequency generation the some frequency generation has a gain very specific dispersion properties such that we have transfer functions here instead of the joint spectral amplitude function this is the input frequency and we transfer that to output frequencies and if you have the right properties of crystals you're employing you can manage that the transfer function is aligned along one line so to speak all frequencies which come into the systems are converted to the same output if i say all frequencies only the frequencies at the end which have a specific shape can be used then so the pump which we basically put into this device defines which of the modes here we impinging in that system uh where we have different modes present different pulse temper modes and then we are converting only one of them in the first case the gaussian mode to the to different frequencies which we can detect changing our pump mode then addresses difference of our input frequencies and this some frequency generation allows us now to have this additional possibility to play with these temporal modes and to manipulate and detect them now a little bit more into detail so this is a parametric down conversion source we pump it with a pulse pump light and the signal idler generated and i tolerate that energy momentum conservation has to be preserved now i also told you already that this resultant transpector now i want to explain you a little bit more detail how that works now this joint um spec from contour plots so the energy conservation of this is the frequency of sigma and the idler and you see the energy conservation just gives you a curve which is aligned at minus 45 degree if we have different frequencies available or possible where we cannot have all of these different frequencies possible now the phase matching condition if you're in a waveguide they're traveling in the same direction but nevertheless they're traveling at different speed because they're traveling at different speed we get a second contour plot in that frequencies domain so this is signal in idler where the angle of these pha or the properties of these phase matching properties are defined by dispersion of the waveguide all together every process is defined by energy and phase matching and we get these joints back lambda use where this is signal frequency and idler frequencies and you see get general correlated function this is the maths behind the things you have these correlation functions and c and idler frequencies are prepared now what does that correlation function mean well this is basically what you see immediately is that this function is not a function which factorizes so in general we don't have just one one photons but we have to take this internal structure into account and this people have done by doing schmidt decompositions now schmidty composition is a mathematical tool we use this function to inspect the amplitude function and we decompose it and we decompose it in the following way and we define temporal nodes as i introduced them beforehand and these temporal modes now are defining these pairs so effectively what this process does is if we have a correlated joint spectral amplitude our process produces pairs of pulsed temporal modes and the effective number of modes corresponds to the amount of entanglement and here you see in a pictorial way what kind of state is generated so we have pair photons where signal is in a gaussian temple mode and idle in the corresponding one or the first mid mode and so on so in channel if you have a waveguide and you just pump it with a pulse place a light you will have all these different modes or represent now if you want to have only one pair present then you have to get rid of them and you have to modify your joints back lambitude in such a way that you have no correlations in the system and this is nowadays called source engineering and then you have only one wave packet present only these temporal mode present one of them and this source engineering is an ideal ptc source for integrated networks with multiple spatial channels because you have past light you can synchronize that you can have high purity single photon states by hydraulic you can have very high brightness source and you have pure squeezed light sources if you want to go to the contest continuous viral machine now what we are doing is we start from these engineered sources and now we play around with different envelope functions for the pump pulse and interestingly this allows us to have a complete control about the temper modes so it's not that we just have one temporal mode set but we can really fully control which temporal modes and what the dimensioning of these temple modes are generated and we combine this now with the detection of the temporal modes where we use the quantum pulse gate and the quantum pulse kit is the device which explained for the intro already um we have a dispersion internet some frequency generate with a very specific um dispersion engineered some frequency generation such that the sun frequency generation is also a single mode process and it only selects the modes we are aiming for now a few more details about the pulse gate this is coming back to our technology it's a home a parity portal it's a live waveguide with a small polling period of 4.5 micrometers the nice thing is that our input states are around 1550 nanometers so in the telecom regime uh our pump pulses which we use for conversion um is in the ties of regime and this device has the specific properties that we can really address individual modes we can select other temple modes in time frequencies and all that in superposition modes we reconstructed the pvm elements of seven dimensional mutual unbiased basis states and here you see now we are coming to the area of quantum communication so these seven dimensional states with all kinds of mutually unbiased days can be used for information encoding now and this is what we are working on one experiment we also did is we combined the ptc state with these engineered pdcs that we can control the modes with the quantum path gate and here you see an example we had only one mode present and we analyzed here the joint spectral intensity and then with the quantum pathway also the modal properties we see that here in this graph only that one hermit mod was present then we changed that um to a meet gaussian mode and then we kind of could implemented that an entangled state in these mode bases we could prove that again with that quantum pulse gate setup now maybe is there any time left anybody still hearing me yes yes a person okay everyone hearing you yeah okay do you have five more ten five more minutes or should i stop here i think five minutes is fine i guess yeah sure i just want to give you a glimpse where these quantum pulse gate where we can operate on these modes can be also used and this is quite interesting because there is um a way for parameter estimation of incoherent emitters this is a new topic which came up and which one we find quite interesting it was introduced again in the spatial domain the question there was you have two stars which you want with uh to estimate the distance between those two things that there's incoherent light coming from the stars and you want to estimate the distance how well can you do that and of course what you can do in space you can do also in the frequency domain or in the time domain and you can ask if i have two increment emitters and there's a distance between then how well can i do that now how well can i do that um you can do in a classical treatment or in a quantum treatment and you see just the theory which is behind you can use estimation theory to treat this problem and and you find and this kind of a trick that in the quantum domain if you use a quantum information approach that's called quantification information you do better than you could do in a classical framework in which sense do i mean that well here you see the two different emitters which are kind of sent lighter and you want to get the distance and you see that the statistics very quickly doesn't allow you to distinguish between these two systems and you know that very well all of you this is nothing else but the rayleigh limit right if you have two emitters which have broad distribution you bring them too close together then you can't estimate this difference anymore interestingly enough if you do a quantum information treatment of that you find that if you do the right measurement a more selective measurement that this curse is not true anymore and can always estimate how far they are away if you do the correct measurement what you need for that is really the coherent detection and this is what we did in the in the time frequency domain in our lab so we generated two different states and we estimated the difference between them and we did not only estimate it for two beams which have individual the same height but also with different heights here's the experimental setup if you're interested i can explain you more in detail how it works but the idea what we really wanted to have we wanted to estimate the um different uh distance between that here in the frequency domain where we had the constant width that we brought them closer and closer together and what you find is that you can always estimate that separation even here you see that in this graph if but by just looking at uh detection the intensity measurement you say would be not possible to see that at all and you see that this works very nicely the tool we use that was exactly this one pulse get where we can measure on these coherency positions now i told you that we are also doing it for different intensities and you can go further you can ask not only what is the distance between what's the centroid and what is the height between those two things uh quantum mechanically this is not trivial because these properties from a measurement point of view are non-computative so we have something like a heisenberg uncertainty principle but if you are able to shape the selection of your coherences you can do these measurements and effectively we were quite proud of that um last year we could show that we can do such a multi-parameter estimation and this quantum pulse gate can help you to really get enhanced performances here you see for all kinds of different parameters um here the distance between them the centroids and the different heights um you can do better than classical systems could do and the measurements are the quantum karma rauban simultaneously this is basically the limit which is analogy to heisenberg okay this was a long talk thanks for listening uh it was an overview of all kinds of things we're doing um to sum up i showed you first of all all non-integrated optics uh quantum devices that and i think this is the basis for practical quality technologies and we are trying to develop that on top of that there's other ways of scaling up systems and i think we will need them i talked about timely fixed quantum systems and our idea and approach is to see what is quantum and classical last but not least i talked about temple modes which brings you to the field of pulsed light and how you can play with different ways of manipulating this path night of the intrinsic mode structure to really also accomplish high dimensional system for example for quantum communications what's next well the technology and we see that already now goes from uh well these large uh wavelength systems to the niobate um leno i literally have an insulator we're also interested in that this will give the whole field another push and makes these little systems even more important you have to scale systems and timely blitzing is a nice way so we have an experiment where we have g substitute generation it's a boost of time multiplexing we need the benchmark in the quantum characterization so we also do work on on characters in large networks and last but not least i didn't have time we're looking for different applications of these pulsed light for example quantum spectroscopy with time frequency entangled light and quantum communication with one pulse gate devices too so i'm at the end let me thank you again for your attention um i would also like to emphasize that this is not my work in principle there's a lot of people are involved and i'm proud of my group all of them are doing great work i would like to acknowledge the founding agencies and while we're always looking for people so thank you for your attention i see that anil is here joined anil would you like to take over perhaps he's not there i think oh shall we move directly to the q a yes i think probably that is yeah so that was a wonderful talk giving a big great overview of different technology possibilities devices integrated devices there are several questions let's try to get to them before that let me take this opportunity to ask you my question first you know i'm completely from a classical communication background and this uh temporal nodes that you were talking about sounded very similar to alternate frequency division multiplexing where we give different pulse shapes to uh different frequencies and i mean so of course these are all done at uh single photon levels but is that the right parallel i think it is to be honest we are this is something thanks for the wonderful question this is something i have to say the quantum optics people discover things which classical objects people also know um i guess we are we are running these systems and you have to tell me if that's correct at different time scales because this femtosecond pulses and as far as i'm aware in classical systems you don't do that and i don't know how you do the sorting there but i think this is the different thing but at the end of the day you're absolutely correct these things could be combined and of course you could use it for dense classical encoding too i think this is where it's nice and you can use them for squeeze slide equivalently to all of the other things the the thing which excites us is that device this quantum pulse plate okay i'll probably take it up later with you for discussion of that uh i should not have a question in the experimental results of quantum work there are some minor deviations from the theory what could be the source of deviation i mean is that just the statistics in the experiment um yeah i think it it this is minor imperfections like these walks are extremely sensitive to any deviation of half-wave plate rotations or my minimal losses so it's an exponential scaling and we are at power to the 28th so if you look at us at quantumworks in the spatial domain we did that i should say with our they look much worse but you still see minor deviations because i said you're extremely sensitive to anything which is not ideal how do you make a 90 i think 90 millimeter long uh this is this is just depends it's a standard lithography and um this patterning is still this is depends on your machines but this is still feasible yeah this is i think this is just i know that because i worked in that layer so it is a contact lithography it is not like a stapler you don't need to bother about field filter whatever things are there that is in the mask actually but uh once you get the mask it is just a simple contact which of course differences yeah it's it's this is why it's a standard mask contact lithography we are not at the level where we have nano's systems already we saw that it's a micrometer and this is from the from the size this is pretty normalizing this is what you find in mass production too um it's um the more tricky thing is how do you get this periodic polling down to 4.5 micrometer and this is not limited by the lithography i should say but by the way how you can control your periodic polling and there we did over the last 10 years quite some work but our smallest pollings we have is 1.7 micrometer and below and this is really quite something but it's an optimization of the way how you process the polling but the lithography is nothing special if you want the next question is regarding the sources what is the maximum value of the purity of the source achieved so far what is the which value sorry it's just the the purity of the source uh the purity of the source um well we have been optimizing the purity's really a lot um but we don't have upperturization so our purity corresponds um to a t2 value of 1.9 somehow and it depends how you filter that so it's something around ninety percent these days um but having said that with uportization so if you have periodic polling which you also structure at the edges you can achieve even better purities and in the literature nowadays people have achieved really extremely high purities also for integrated systems the next question is something related to what you probably touched upon the end lithium level insulator what is the prospect of that in this research i mean some materials i have to say this what i i presented you the concepts there and just recently if you look um there's a very fascinating work by a long cast group together i think with feyer um and also franco mo where you they show extremely good quality on these sources and personally i think that this will be the future um because uh well you can minoritize things now having said that for many other applications like for example for communication one of the challenges is to have extremely high efficiencies and also interfacing and to be honest um there's probably a long time where also these conventional devices will have their place but if you think about computation i think it's a very promising platform so uh i can have a related question here uh to christine uh question uh that you know lny at this longer group he was just referring that is really promising i'm also thinking of switching back to lithium nightmare college president tagrajan also asked me to why don't you go but lithium rabbit again so one uh problem i see that you see normally whenever you are going for a pdc parameter down conversion so your wavelength is just from 750 to come coming down to 1550 nanometer that's a wide wavelength range and in lnoi this good part is that you can have a very tightly confined guided mode but that will lead to a very uh high dispersion yeah so so that means you have to have your dispersion compensation broad band discussion compensation needed for your conversion right so how that can be managed is there any way out um i think that the dispersion compensation you have to design your system correctly you're right that the dispersion then is not governed only by the material dispersion anymore but by the modal dispersion but you need to control that actually a lot of the ideas i showed you in the last slides i didn't go into detail about these pulsed systems use a dispersion control so yes you have to control your system very precisely you have to design them in advance so you have to make clear that the geometric structures you're using are the correct ones but actually this is even advantage of lnri so you can't just produce wave cuts and see what they do you have to model and simulate them beforehand and i can tell you because we also you're looking to these things it's uh longcast and they're doing a good job in that they know exactly what they do but this is not an issue not at the end you have to care for it but then it's not an issue is that the answer enough yeah yeah okay yeah okay so the next question is probably more general uh it's about the term single photon source we attach with signal idler generating sources based on uh spontaneous over mixing parametric down conversion etcetera in these modes the photon found in a single policy is probabilistic and we cannot assure that there is only one photon in a given pulse hence it's really misleading to an entire title resources are single proton sources um yeah absolutely right thanks for pointing that out um i started off with gaussian boson sampling and there you see only one way out um if you look at the full statistics of these sources actually a photon pair source is exactly the same like a two mode squeezed source um so one way is you look at continuous variable systems but maybe that's not what you want but you have to drive them at low efficiencies such if you really want to use them at single photon sources and the way out there is what people have also already demonstrated is multiplexing so you have to produce several of them and you stack them together and select things um i think honestly uh of course there's other ways but the more the systems become integrated the more technological advance they get this multiplexing becomes more realistic in the quantum walk experiments what's the scope for studying 3d quantum networks using synthetic dimensions such as polarization or frequency control since your systems already have polarization control control capabilities this is the question how we can make it 2d a 2d system effectively 3d 3d can you add polarization for instance you have time frequency so um i hope that wasn't confusing in the in the temple in the quantum walks we use the polarization already the pronunciation is the coin space uh and we're using temple bins now if you want to make it more dimensional what you can introduce is different time scales and we did that for implementing a 2d system already now if you use three different time scales you can go in-dimensional and the question is how do we uh get that coin space up because if you want to cover the whole thing um now we also have demonstrate four dimensional coins and this is a clever design how you combine different timings with different conversations together so we demonstrated a 2d system with a real 4d coin i would say we demonstrate a 1d system for four-dimensional coin but this time multiplexing can be used in a clever way to also try to synthesize higher dimensions now how efficient that is at the end it's always the question because the more times because you introduce the longer you have to wait but again also these ideas um have been looking people have been looking into that and you can't play around here in implementing different dimensionalities in coin as well as in uh time space and of course in principle you can also use frequency dimensions if you want and this is also the overall idea this is why i present you these different ways the spatial encodings integrate optics the temporal encoding with the time and frequency encoding and then you can play all kinds of different games if you want at the end you will have to find efficient ways of scaling our systems as a photonic as a whole the next question is how efficient is the generation of pulsed temporal modes with frequency sum generation as compared to narrow bands how efficiently can you generate the pulse temperatures um again here we have an engineered system so the efficiency is not not less but actually at least i think it's actually more efficient well if you're only going with with monochromatic modes you can have and probably uh equally efficient but these condom pulses are very efficient this is because we make sure that the group velocities of the pump and the input is the same and then you have a very efficient system so we demonstrate efficiencies for the pulse gates up to 50 percent um and then you other things kick in like time ordering issues and then quantum mechanics also plays a role but then you can have cascade pulse gates in principle 100 is feasible and there has been a work by mike reimer where he has created two of them showing the principle also 100 efficiencies can be done so there's twofold um there's it is more efficient than at least as efficient with monochromatic light this path that is good and um maybe the second answer 100 efficiency is feasible which is important if you want you use these things as add drop failures didn't dispersion be a problem the shorter the pulse you uh you must make sure that your face matching is flat such that means that within the length of your crystal that the pulses really don't have a third order dispersion effects which they experience so these kind of things come into game but this is a question of engineering at the end um the next question is there a concept of periodic polling that exists in lnoi sorry is there a periodic polling pplm uh periodic polling can be implemented can that be a concept that can be implemented in illinois yes there's several work which people have demonstrated that i don't know how many but there's if you look in the literature this is absolutely feasible so those are the questions in the chat how about using uh different special modes of waveguides instead of using signal mode waveguides or increasing the dimensionality of the fundamental system yes it's also a question of control yes you can do it i know that with optical fibers people have been doing that but then this wave guides um we have rectangular shapes and controlling yes i think there is of course uh in lithium airbag people have been using different spatial modes and this is definitely also a possibility why not the question is um we're not doing it i guess but there has been work also doing that that you use different spatial modes and then you have to control all the conversion into these different spatial modes the question of engineering yeah another question when you have this gaussian pulse modes a gaussian pump put a down convert to the gaussian signal and a second order hermit gauss idler that has a non-zero parity that parity is conserved so is that is that allowed at all or is it disallowed in your process are we talking about spatial or spectrum modes here so the time time pulse i think uh it is a lot it is allowed you see it in the decomposition and one reason why you immediately see that it's allowed it with playing with different wavelengths so i think this parity conservation because you had pairs of gaussian and hermit gauze first order hermet go second order i'm just asking there is also a possibility that a gaussian splits into a gaussian signal and a second-order hermite goes either there's no problem uh yes but these these timbre modes are defined no there's no problem but you maybe you're right the parity is conserved anyhow because there's three fields involved right exactly there's the pump and then there's signal idle and they kind of then there's the preservation but we have free fields correct yeah but uh you don't see that at all in your experiments this kind of output coming out um this this is a little bit tricky i like um we do see all these different modes uh but the process which is defining that is the correlation which are between signal and idler so this conversion we control and we see that if we have uh the first amine mode as a pump then we see exactly what you say then we are generating a signal uh gaussian with an idler first meet mode and vice versa this is how this entanglement generation is done yeah okay okay yeah all right one last question uh what gives a better resolution quantum imaging of two concurrent sources that beats the rally limit what is the physics behind improved resolution the physics behind is exactly in the spatial domain if you have um if you have a goal if you can measure on a coherent pulse and you ask what is kind of from a coherent path the first derivative meaning a displacement then you find it the first to meet gauss mode now if you're measuring on the gaussian mode and the meat mode in the coherent fashion then you can measure this separation very precisely that's exactly what's also the physics which is behind that it's basically playing around with measuring these derivatives and otherwise exactly the symmetry properties and if there's any small displacement you see that nicely then you see that the symmetry properties are broken and this is how you measure it okay so thank you i'm done deeper yes i have a question regarding this for full scale integration probably you may need the detector as well so can you please comment on that so what kind of detectors for single photon detection available on this lithium nibel is there any hybrid options are there or not yes there are also there's snsps which you can implement on top of this lithium nia but we are developing a junior professor in part one is also developing them but not only here and there's in the literature you find them already also an electrolyte illinois you need some nairobi on these so basically what you do you deposit superconducting films on top of your structures and then you have to to kind of also design the everlasting coupling but um snspd's are completely feasible uh i i have to just add one question here and then so uh you know your religion i have it uh you are operating at a higher temperature i guess that right so now you want to integrate snspd and then you have to put in refrigerator that's um yes you're right in order to to restrict photoreflective index in this you typically heat your waveguides to 200 degrees celsius now there's two facts that several of our experiments are done at room temperature or below first of all we drive them pulsed which changes apparently the situation that you don't have and we drive them at extremely low temperatures so btc extremely low power so pvc we launch next to no pulsed energy and therefore the refraction doesn't play that role um we also demonstrate already that we can have ptc this is a new uh optical paper uh in this cryostat so it's absolutely feasible it's not trivial but it's doable yeah okay okay if not i'll get back to my original arguments that in the concept of you know in the this orthogonal frequency division multiplexing you do also have the time frequency mapping and you pick orthogonal functions and of course there we do everything in the classical 50 domain and here you seem to be doing it in more of a physical modulation kind of right but otherwise the concept looks the same to me and then again we say that the resilience uh it's like you know you say using these uh central modes you're able to be able to for example uh improve the stellar resolution uh here again we say in the corresponding uh resolution is the frequency resolution here in the classical picture and we say that uh the resilience to right so noise see an exact parabola um okay so i don't quite understand if i get the answer the question but let me try to ask you if that's the question you say the signal so you are referring to the signal to noise ratio and you're kind of passionate um why that should be better with uh these time frequency mapping and you basically encode your information on these kind of eigen functions which are optimal functions like what you showed uh you do get a better resilience to the signal against the series ratio so i see something like that is happening here as well it is it absolutely is and there's work from contact banner actually for looking at communication systems actually classical communication systems what you could gain if you use pulse gates for noise filtering and if you're interested look out for that he demonstrates that you can gain a lot because um there's two things yes it's a perfect noise filter because um you kind of take these coherent things out right and then for the single photon level uh you also gain if you do not use homodyne detection but click detection these on off detectors can have advantages in terms of noise there too and um is that my colleague here demonstrated also for classical system this kind of approach of selecting with a pulse gate of these things on specific modes and doing at the single photon level detections actually has also advantages there and yes you're absolutely right in terms of noise um these encoding also is very beneficial this is what people are starting to look into and then you combine immediately the advantage from these different detection scheme which is also classically and you can directly combine that with quantum communication i think this is a very attractive thing i told you that we're looking for applications for quantum pulse gate all these things and yes it is an extremely useful tool for noise filtering yeah absolutely correct thank you thank you for that questions questions shall we close the session yeah i think i think if any other questions i think or not you can note down we can pass down to uh success lunch probably it is 1 30 no worries okay thank you so much christine for spending a lot of time patiently answering thanks to the wonderful audience asking wonderful questions thank you process for you see you sure asking us the thought progress queen questions bringing us very interesting questions professor christine say it's a wonderful talk listen to you again thank you thank you very much for inviting me to the opportunity thank you thank you thank you have a wonderful day bye bye thank you thank you deepak thank you thank you for talking thank thank you see you then yeah bye hello [Music]