Hi Daniel, Very nice talk. It's the first time I've heard that the no-cloning theorem can be used to illustrate why a mesoscopic superconducting circuit could manifest a truly quantum behavior. Thank you for preparing such an excellent talk. All the best.
Hi Daniel, do you know if there is an online recording of Joe Bardin's talk that you mentioned in the introduction of the video? Thank you for posting this.
Very good explanation. Most "tutorials" just repeat the description of qubit, interlaced and superposition, but don't provide any further detail. P.S. You move the pointer too much.
You're welcome. I agree that more detail would be nice. It would help me to know the backgrounds of people watching the video. Are you a physicist / programmer / electrical engineer?
@@danielsank2286 I'm a programmer with some physics background. If I can add one more question, it seems like one of the key enablers of the transmon qubit is the josephson junction and how it allows us to oscillate between the 0 and 1 states on a constant drive. Do you have any resources that would allow me to understand the math behind this? I didn't understand the leap from "more current -> more inductance" to the blue/red loops u drew between |0> and |1> states.
@@slashhashdash Well, there are a few ways of understanding what's going on there. One way is to treat the transmon like a nonlinear resonator and use a quasi-classical approach, and the other way is to go full-on quantum mechanics. As for references, try appendix D of my thesis (web.physics.ucsb.edu/~martinisgroup/theses/Sank2014.pdf). I have an updated version of that appendix with more detail that I can share directly if you want (it will eventually be published as a book chapter).
Hi Daniel, Nice presentation. Couple of questions from my side. What kind of calibration do you use assuming launched signal needs to propagate through different temperature/environment? SOLT, TRL or In-situ. In the implementation phase what kind of yield analysis you do considering possible highest level of precision you need, Is this Monte Carlo with +-1 sigma distribution yield or something more robust? What kind of nonreciprocal device (circulator) do you use in the readout (Rx chain), is magnetbased or magnetless parametric circulator? Assuming amplifier needs to be used right after the circulator at a very low temperature stage, if so what kind of amplifier? Is it JPA? Understand temperature T is linearly related to frequency f which dictate the temperature margin and required energy level, mentioned 6GHz is the optimum frequency for this technology or is it possible to use other frequency region as well like 1GHz? Thanks
Hi Swadesh. These are good questions. We mostly do not calibrate (de-embed) the path between the pulse generator and the qubits. Instead, we vary the pulse magnitude and phase and observing the response of the qubit, therefore directly calibrating the pulses. This approach works but it is somewhat annoying that every time we change the qubit frequency etc. we have to re-calibrate the pulses. We do use de-embedding measurements to learn things about e.g. the readout circuit, specifically impedance mismatches and losses -- because in readout these things really matter. I do not understand your second question. We do use magnet-based circulators in the readout system. Smaller circulators will be a critical element in our next steps. The first stage amplifier is an IMPA (a Josephson parametric amplifier with impedance matching). For your final question, certainly other temperatures could be used, but note that 1 GHz corresponds to 48 mK.
@@danielsank2286 Thank you very much for the detailed explanation. If I understand correctly, the complete RF signal chain can be simplified into 3 parts. 1. Signal generator to Qubit input, 2. Qubit itself, 3. Qubit output to Cryostat output (Readout) For segment 1, are you referring TDR (Time domain reflectometry)? Why do you need to change frequency? Is it to achieve different anharmonicity and different energy/temperature? when you change frequency all RF component should be changed/adjusted as well, like circulator, attenuator, amplifier considering each component are in line with resonance. For segment 3, how do you calibrate when there are non reciprocal device in the chain (just curious). My 2nd question was mostly related to signal integrity, Say for example, you have a nominal/expected result from your device, However, in a 6GHz RF environment, there may be issues such as crosstalk, interference, coupling, signal bounce, TCF (temperature compensation factor) etc. Now for example if you have 15 different variables that is controlling these KPIs, you may want to know their sensitivity on the performance which require yield analysis/Monte carlo simulation with ~1 sigma distribution. to understand in real time how much performance can vary from nominal/simulation. Magnet based circulator might be expensive and heavy and assuming for this part of of readout chain impedance matching has to be perfect to eliminate any reverse isolation, is that correct? Thanks
@@swadeshpoddar6918 For segment 1 we do not routinely do any of the usual microwave measurements. We shoot pulses at the qubits and adjust their magnitude, shape, and phase until the qubit does what we want (because it is quantum, it takes many repetitions of the same experiment before we have enough statistics to know whether or not the qubit is doing what we want). The qubits change frequency as part of normal operation, i.e. to do logic gates. Also, we change their frequency to avoid frequencies where the performance is hindered by defects. This is an important point: there seem to be charged microscopic defects in the qubits such that when the qubit is operated at certain frequencies, the electromagnetic fields of the qubits interact with the defect and mess up the qubit. Fortunately, we can tune the qubits' frequencies to avoid those defects. About the crosstalk, etc. historically we have measured it directly via qubit performance. However, that approach needs to be improved, i.e. we need to measure the performance of each piece of the signal lines carefully. This is work in progress as it involves proper simulations and measurements in the cryogenic environment.
@@danielsank2286 what semiconductor process (CMOS, GaAs, etc) was the chip fabricated in? I assume all the qbits need to be read and written at the same time, with coherent LO and sampling across everything?
Can you pls send me your calculation with which u came to the result of Kirchhoffslaw? I got different answer with 3. degree derivation of the voltage, not the second. Tnx
Mostly great explanation thank you. But I dont seem to understand how quantum state machine matrix relates to map on the bottom right corner. You state that probability is zero everywhere but state 10. While matrix shows clearly only 00 is 1 and rest is 0. Also in stochastic machine you say every outcome is possible while output vector shows 0 probability for 10. Thank you.
There are two issues here. First, I did not make sure that the quantum matrix would correspond to the diagram in the lower left; this was a lazy mistake on my part. Second, I just do not understand what you mean about "only 00 is 1 and rest is 0".
@@danielsank2286 Thank you for your quick response. That was related to matrix not being in sync with the dot chain. You pointed out that the blue lines destructively interfere therefore state machine is absolutely in state "10". But the vector up on the screen indicates that state machine is %100 in state "00". I guess that was only to provide some visual representation and they are not really related.
