Special | A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | ALL
Q |
---|
Qubit: physical realisation | ||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Creating a low voltage to represent a logical “0” and a high voltage to represent a logical “1” is straightforward. 🤷♂️ Creating a superposition of low and high voltages however is not. A quantum computer is commonly envisioned to be a machine that exploits the full complexity of a many-particle quantum wavefunction to solve computational problems [LJL+10].
For the construction of quantum computers, laser serves as an inspiration because it is quantum mechanics that enables laser waves to be generated in phase [LJL+10]. Just as there are many possible materials for lasers (e.g., crystals, organic dye molecules, semiconductors, free electrons), there are many materials under consideration for quantum computers; see [LJL+10] and [MM12, Ch. 6]. Quantum bits are often imagined to be constructed from the smallest form of matter, e.g., an isolated atom, through ion traps and optical lattices, but they can also be made in components far larger than consumer electronics, e.g., a superconducting system [LJL+10]. Below we discuss three main technologies [LLSK22, Sec. 6]: 1️⃣ trapped ions, 2️⃣ photonics, and 3️⃣ superconducting qubits. Trapped ions: Main idea is to use the two different internal states of a trapped atomic ion as a two-level system (i.e., qubit) [LLSK22, Sec. 6.2]. An ion trap uses electromagnetic fields and laser cooling to control the spatial position of an ion in vacuum and reduce the temperature of the ion [LLSK22, Sec. 6.2; BCSH21, Sec. 2]. Watch an introduction to the ion trap: Lasers or microwaves are used to control the internal states of an ion [BCSH21, Figure 1]. The internal control plus the Coulomb repulsion between ions combine to form conditional logic gates [BCSH21, Figure 1]. 👍: State preparation, qubit measurement, single-qubit and two-qubit gates can be performed with fidelities (> 99%) higher than what is required for quantum error correction [LLSK22, Sec. 6.2]. 👎: A large array of bulk optical components are necessary and these are difficult to address and measure, challenging scalability [LLSK22, Sec. 6.2]. Trapped-ion quantum computers (e.g., IonQ) are enjoying a reasonable level of commercial success [LLSK22, Sec. 6.2]. Photonics: Photonics has always been a prominent candidate for realising qubits [LLSK22]. For generating qubits, photonics offers the following advantages [SP19, PAB+20, LLSK22]:
Superconducting qubits: They are currently the leading contenders in the race for large-scale quantum computing [LLSK22]. Superconducting qubits are the technology big-tech companies like Google and IBM have been focusing on. In 2019, a large research team consisting of Google and multiple American and European universities demonstrated “quantum supremacy” on a programmable superconducting quantum processor called “Sycamore”, which consists of a two-dimensional array of 54 transmon qubits [AAB+19]:
References
| ||||||||||||||||||
QuTiP (Quantum Toolbox in Python): setup | |||
---|---|---|---|
There is no shortage of quantum computing frameworks/toolkits, including Google’s Cirq, Rigetti’s Forest SDK (including pyQuil), Microsoft’s Q#-based Quantum Development Kit, IBM’s Qiskit, Quipper, QuTiP, ETHZ’s Silq, Cambridge Quantum Computing’s tket. We shall use Qiskit because 1️⃣ it is among the most well-established, and 2️⃣ it comes with rich learning resources.
We shall also use QuTiP, because while Qiskit is popular for quantum computing, QuTiP offers more features for quantum-dynamical simulations.
Watch an introduction to QuTiP: Using Qiskit and QuTiP means we are using Python but there are so many resources for learning Python it should not be an issue for a Bachelor/Master student to pick it up along the way. Even though the relevant computer pools will provide you with the required software, you need to install the software on your own computer(s) anyway, because you will not be able to finish any of the practicals within the allocated time. The allocated time is only for you to get enough supervision so that you can complete the remainder of the practical on your own. Our strategy here is to first install QuTiP, then Qiskit. The operating system of choice for many computer scientists and cryptograhers is Linux, but the setup guide here is only applicable to Windows, because UniSA computer pools have only Windows. Follow the instructions below (derived from the official instructions) to set up QuTiP:
Above, the backtick (`) is the line continuation character in PowerShell. Among the packages just installed, Official Conda documentation discourages using pip and conda together, but we need pip for Qiskit later. | |||
R |
---|
Reduced density operator/matrix and partial trace | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
The reduced density operator is an application of the density operator. The reduced density operator is so useful it is indispensable in the analysis of composite quantum systems [NC10, Sec. 2.4.3]. Consider a composite system, with density matrix , consisting of Alice’s system and Bob’s system, there is often a need to express the state of Alice’s or Bob’s system in terms of . The operation that takes us from to (denoting Alice’s density matrix) or (denoting Bob’s density matrix) is called the reduced density operator. The reduced density operator for Alice’s system is defined by where is a map of operators known as the partial trace over Bob’s system. Suppose and are any two vectors in Alice’s state space, and furthermore, and are any two vectors in Bob’s state space, then the partial trace is defined by the operation [NC10, (2.178)]: Above, the notation is equivalent to To see why the definition above makes sense, suppose a quantum system is in the product state , where and are the density matrices for subsystems and respectively, then since the trace of any density matrix is 1. Similarly, In general, if where is an orthonormal basis of , and is an orthonormal basis of , then the partial trace over is [WN17, Definition 1.6.1]: For a quick summary of discussion up to this point, watch edX Lecture 1.7 “The partial trace” of Quantum Cryptography by CaltechDelftX QuCryptox. Consider the pure (entangled) Bell state: . The system comprises 1️⃣ single-qubit subsystem with basis vectors and , and 2️⃣ single-qubit subsystem with basis vectors and . This system is entangled (i.e., not separable) because , but using the reduced density operator, we can find a full description for subsystem and for subsystem . The density matrix for the system is: Applying Eq. (1), the reduced density operator for subsystem is: Since , both and are mixed states. 🤔 How do we reconcile the preceding observation ☝ with the fact that is a pure state?
The strange property, that the joint state of a system can be pure (completely known) yet a subsystem be in mixed states, is a hallmark of quantum entanglement.
⚠ Caution [KLM07, Exercise 3.5.5]
The partial trace contains all the relevant information about subsystem if subsystem is discarded. Similarly, contains all the relevant information about subsystem if subsystem is discarded. These local descriptions do not in general contain enough information to reconstruct the state of the whole system. Expressing a bipartite vector in the Schmidt basis makes it much easier to compute the partial trace of either subsystem. For this reason, let us discuss Schmidt decomposition. References
| ||||||||||||
S |
---|
Schatten norm | ||||
---|---|---|---|---|
For , suppose its singular values are . Then, the Schatten -norm of , where , is defined as [Ber09, Proposition 9.2.3]: When , we have the Schatten 1-norm, which is also called the trace norm or nuclear norm [Ber09, p. 549]: When , we have the Frobenius norm: References
| ||||
Schmidt decomposition | ||||
---|---|---|---|---|
References
| ||||
Secret key rate | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
The secret key rate is the fraction of secure key bits produced per protocol round, where a round is the transmission of a quantum state through the quantum channel [Gra21, p. 38]. The secret key rate generally depends on the total number of rounds performed. The asymptotic secret key rate (often just asymptotic key rate) is the secret key rate when is assumed to simplify analysis.
Assuming is not realistic, and the security of a QKD protocol has to be analysed assuming a finite and generally finite resources. Analysis of the secret key rate and associated security properties of a QKD protocol is called finite-key analysis [TLGR12], to be covered in the future. References
| ||||||||||
Security of quantum key distribution (QKD): overview | ||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Methods for analysing the security of quantum key distribution (QKD) schemes/protocols are still being developed [PAB+20, PR22].
The security of a QKD scheme is typically analysed in terms of the level of success of 1️⃣ individual attacks, 2️⃣ collective attacks and 3️⃣ coherent attacks; in 🔼 increasing order of power given to the adversary [Wol21, Sec. 5.3.1].
The security of a QKD scheme is also often analysed in terms of composability, short for universally composable security.
TODO: Trace distance criterion [PR22], security = correctness + secrecy [Gra21]. Asymptotic vs finite-key security analysis. Security evaluation of practical QKD implementations involves evaluating the level of success of “quantum hacking” (i.e., side-channel attacks on QKD). References
| ||||||||||||||||||
Separable vs entangled | ||||
---|---|---|---|---|
The joint state, , of two quantum systems and is separable [WN17, Definition 2.1.1] if there exists a probability distribution , and sets of density matrices and such that If such an expression for does not exist, is entangled [WN17, Definition 2.1.1]. Specifically, if is a pure state, then is separable if and only there exists and such that Example 1: [WN17, Example 2.1.3]
Example 2: [WN17, Example 2.1.2]
Example 3: [WN17, Example 2.1.1]
Example 4: [WN17, Example 2.1.4]
This example is meant to highlight the difference between the following two states: where . is separable whereas is not. Consider the outcomes of measuring subsystem in and in the standard basis and in the Hadamard basis. Measuring subsystem of in the Hadamard basis: Define the measurement operators to be and , where and . Clearly, and . Using projective measurement, the post-measurement state conditioned on measurement outcome is Let us work out the numerator and the denominator separately, starting with the numerator: The preceding equality follows from these identities, which you will derive in the practical: The denominator is: The preceding computation is tedious but straightforward given the right tool (e.g., NumPy). Combining the results for the numerator and denominator, we get Thus, upon measuring a on subsystem , subsystem can be in either or at equal probabilities; we say the reduced state on is maximally mixed. Measuring subsystem of in the Hadamard basis: The pair of and can be re-expressed as and , because ‘’ is orthogonal to ‘’ just as ‘’ is orthogonal to ‘’. The preceding statement implies when is measured on subsystem , subsystem is in state as well. References
| ||||
Singular value | ||||||
---|---|---|---|---|---|---|
Singular values, like eigenvalues, are an intrinsic property of a matrix. Unsurprisingly, they can be defined in terms of eigenvalues: Definition 1: Singular value [Ber09, Definition 5.6.1]
Singular values can also be defined through the operation called singular value decomposition: Definition 2: Singular value decomposition and singular values [Hog13, Sec. 5.6]
References
| ||||||
Spectral theorem and spectral decomposition | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Spectral theorem is one of the fundamental theorems of linear algebra [ABH09, Sec. 3.6]. Based on the spectral theorem, spectral decomposition is an essential tool in quantum theory [NC10, Box 2.2]. Multiple equivalent interpretations of the spectral theorem exist, e.g., [ABH09, Theorems 3.6.4 and 3.6.12]. The interpretation in Theorem 1 directly defines spectral decomposition, and is hence also called the spectral decomposition theorem. Theorem 1: Spectral (decomposition) theorem [Hol13, Theorem 8.23; Zha11, Theorem 3.4; KLM07, Theorem 2.4.3]
Any normal operator, , has an outer product representation [KLM07, Sec. 2.4; Mey00, p. 517]: where The outer products are projectors that satisfy
References
| ||||||||||||||