Understanding inherent bottlenecks to efficient algorithm design lies at the heart of computer science. This question is significant both in the classical computing domain as well as in the emerging context of quantum computing. In this thesis, my goal is to characterize bottlenecks in designing efficient algorithms through the lens of a parameter called concentration of functions, starting with the domain of quantum information. My primary focus is probabilistically approximately correct (PAC) learning. I chose this domain since it allows us to approach the important subject of supervised learning in a rigorous and principled manner. For PAC learning, I propose a quantum algorithm to learn the class of concentrated Boolean functions with complexity $ O(\frac{M}{\epsilon^2})$ which offers an advantage over the best known classical PAC algorithms with complexity $\tilde{O}(n^2M)$, where M denotes the number of the concentration terms. I also show a lower bound $\Omega(M)$ to PAC learn this class of functions in distribution-independent settings. All this work is done in the context of the standard query model for PAC learning where the complexity measure is the number of queries, dubbed query complexity. I extend this work to include the learning model where functions are learned without any error, which is often called exact learning, and prove a query complexity lower bound of $\Omega(\frac{\epsilon \log M}{n} 2^n)$ in exact learning the class of concentrated Boolean functions. In the next part of the thesis, I focus my work on classical algorithms and explore a combinatorial counterpart of concentration called degree of symmetry. In this arena, graph isomorphism is my problem of choice. Once again, my goal is to characterize the efficiency of algorithms, in particular parallel algorithms, for graph isomorphism based on a concentration-related parameter. In particular, I propose a parallel algorithm in polynomial time using a quasi-polynomial number of processors for the Graph Isomorphism problem. My work builds on Babai’s celebrated quasi-polynomial algorithm and is work-preserving. The idea behind the parallelization explores the symmetry of the input structure for easier parallelization.
Understanding inherent bottlenecks to efficient algorithm design lies at the heart of computer science. This question is significant both in the classical computing domain as well as in the emerging context of quantum computing. In this thesis, my goal is to characterize bottlenecks in designing efficient algorithms through the lens of a parameter called concentration of functions, starting with the domain of quantum information. My primary focus is probabilistically approximately correct (PAC) learning. I chose this domain since it allows us to approach the important subject of supervised learning in a rigorous and principled manner. For PAC learning, I propose a quantum algorithm to learn the class of concentrated Boolean functions with complexity O(ϵ2M) which offers an advantage over the best known classical PAC algorithms with complexity O~(n2M), where M denotes the number of the concentration terms. I also show a lower bound Ω(M) to PAC learn this class of functions in distribution-independent settings. All this work is done in the context of the standard query model for PAC learning where the complexity measure is the number of queries, dubbed query complexity. I extend this work to include the learning model where functions are learned without any error, which is often called exact learning, and prove a query complexity lower bound of Ω(nϵlogM2n) in exact learning the class of concentrated Boolean functions. In the next part of the thesis, I focus my work on classical algorithms and explore a combinatorial counterpart of concentration called degree of symmetry. In this arena, graph isomorphism is my problem of choice. Once again, my goal is to characterize the efficiency of algorithms, in particular parallel algorithms, for graph isomorphism based on a concentration-related parameter. In particular, I propose a parallel algorithm in polynomial time using a quasi-polynomial number of processors for the Graph Isomorphism problem. My work builds on Babai’s celebrated quasi-polynomial algorithm and is work-preserving. The idea behind the parallelization explores the symmetry of the input structure for easier parallelization.
This thesis studies strengths and weaknesses of quantum computers. In the first part we present three contributions to quantum algorithms. 1) consider a search space of N elements. One of these elements is "marked" and our goal is to find this. We describe a quantum algorithm to solve this problem using essentially sqrt{N} queries and other operations, improving over the gate count of Grover's algorithm. 2) We give a succinct characterization of quantum algorithms in terms of polynomials, and develop a new technique for showing upper and lower bounds on quantum query complexity based on this. 3) One generic technique used to compute the minimum of a given function is "gradient descent". We present a quantum gradient-calculation algorithm and a gradient-optimization algorithm that is quadratically faster than classical algorithms. In the second part of this thesis we look at quantum learning theory. 1) We survey quantum learning theory, describing the main results known for three models of learning, using classical as well as quantum data: exact learning from membership queries, the probably approximately correct (PAC) learning model, and the agnostic learning model. 2) We then consider if a quantum learner can PAC-learn a given concept class from fewer quantum examples. We give a negative answer by showing that quantum examples are not more powerful than classical random examples, both for PAC learning and for agnostic learning.
I hereby declare that I am the sole author of this thesis. This is a true copy of the thesis, including any required final revisions, as accepted by my examiners. I understand that my thesis may be made electronically available to the public. ii In this thesis we study the power of quantum query algorithms and learning graphs; the latter essentially being very specialized quantum query algorithms themselves. We almost exclusively focus on proving lower bounds for these computational models. First, we study lower bounds on learning graph complexity. We consider two types of learning graphs: adaptive and, more restricted, non-adaptive learning graphs. We express both adaptive and non-adaptive learning graph complexities of Boolean-valued functions (i.e., decision problems) as semidefinite minimization problems, and derive their dual problems. For various functions, we construct feasible solutions to these dual problems, thereby obtaining lower bounds on the learn-ing graph complexity of the functions. Most notably, we prove an almost optimal Ω(n9/7/ log n)
This dissertation explores results at the intersection of two important branches of theoretical computer science: quantum computation, which studies the power of devices based on quantum physical phenomena, and computational theory, which studies the foundations of machine algorithms. We refer to the area of research at this intersection as quantum computational theory. Like its classical counterpart, quantum computational theory aims to understand the computational and information theoretic requirements of problems and algorithms. However unlike the classical models, the models for quantum learners generally involve quantum sources of information and quantum gates employing quantum physical phenomena such as superposition and entanglement, which do not have classical equivalents.
Our results can be summarized as follows: (1) In Chapter 3, we study the information theoretic requirements of quantum learning, partition and Probably Approximately Correct (PAC) learning. First, we develop a new general quantum algorithm, resolving a conjecture by Hunziker et al. [HMP+03]. Next, we present positive and negative results towards the rather general problem of partition of which both exact learning and computing a binary property are special cases. Finally, we derive an improved lower bound on the number of quantum examples required for PAC learning. (2) In Chapter 4, we consider the problem of unions of high dimensional rectangles over the domain [b]n using membership queries and uniform quantum examples. These classes oral generalizations of classes of DNF formulae over {0,1}n that have been extensively studied in the theory literature. (3) In Chapter 5, we develop quantum algorithms for and testing juntas (Boolean functions that depend on a few variables) and sparse functions. These algorithms are based on the quantum subroutine due to Bshouty and Jackson [BJ99] that enables sampling from the Fourier spectrum.
This thesis focuses on two different aspects of quantum computation: quantum complexity and quantum semantics. In the first part, I study the quantum complexity mainly within the quantum query model. Together with my colleagues, I introduce a new framework for quantum query complexity, phrased in terms of the minimal oracle and analyse limits and strengths of this new model in comparison to their classical and quantum counterparts. Working within the query model I study quantum one-way functions. I show that in the quantum setting, the problem of the existence of a quantum one-way permutation can be reduced to the problem of constructing polynomial size networks for performing the specific task of the reflection about a sequence of states. Furthermore, I extend these results to the domain of the state and operator complexity. I show that if a quantum one-way function exists, then we can construct a sequence of so called “hard” states with the property that the reflection operators about those states are efficiently implementable. In the second part, I study the extension of domain theory to the quantum setting and develop the semantics of quantum computation. By defining a quantum domain I introduce a rigourous definition of quantum computability for quantum states and operators. Furthermore I show that the denotational semantics of quantum computation has the same semantical structure as the denotational semantics of classical probabilistic computation introduced by Kozen. This could be considered as a foundation for designing functional programming languages for quantum computation. Finally, I continue with an abstract mathematical approach to study a general formalism for describing entanglement manipulation and introduce a new approach to derive a unique measure of entanglement for bipartite quantum pure states.
In this thesis, we study quantum computation and algorithms for isomorphism problems. Some of the problems that we cover are fundamentally quantum and therefore require quantum techniques. For other problems, classical approaches are more appropriate; however, we often give quantum variants of our classical algorithms as well. The field of quantum computation aims to accelerate algorithms by exploiting the laws of quantum mechanics. Several quantum algorithms are known that are exponentially faster than the best classical algorithms known for the same problems, including several which are relevant to cryptography. Quantum computing is therefore of great importance. On the theory side, it has already transformed our notions of which problems are tractable and which are not. On the practical side, the construction of a quantum computer would render many popular techniques for encryption obsolete. We derive several results in quantum computation. Quantum algorithms are normally formulated in an abstract way that ignores practical details such as where different parts of the computation are located physically. Using naive methods for translating these algorithms into practical quantum computing technologies increases the time complexity by a linear factor, while previous work reduced this factor to a square root. We further reduce this factor to a constant at the cost of requiring additional space. This retroactively justifies an important assumption made in many quantum algorithms. Another interesting problem is to determine the query complexity of extracting different types of information from physical processes. In the case of deterministic processes, it is well known that quantum algorithms cannot outperform deterministic algorithms by more than an exponential factor. We show --- somewhat surprisingly --- that when the process is allowed to be randomized, there are problems that have a constant quantum query complexity but cannot be solved classically no matter how many queries are made. In fact, we show how to construct such an infinity-vs-one separation from any weaker separation between the classical and quantum query complexities. In an isomorphism problem, we seek to determine if two algebraic or combinatorial objects have the same structure. One of the most well-known of these is the graph isomorphism problem, which is interesting since it is suspected to be of complexity intermediate between P and the NP-complete problems. We transition from quantum computation to isomorphism testing by considering the tree isomorphism problem. While linear-time classical algorithms are known for this problem, we show that a promising framework for developing efficient quantum algorithms for graph isomorphism, known as the state preparation approach, can also solve tree isomorphism. While this result might seem modest, it is important to know that the state preparation approach works for trees, since otherwise there would be little hope of using it to obtain efficient algorithms for more interesting classes of graphs. We also derive a powerful primitive along the way that is of independent interest. Next, we study the group isomorphism problem, which is a special case of graph isomorphism. Group isomorphism is suspected to be significantly easier than graph isomorphism, but still has unknown complexity. While group isomorphism is already of interest since it is a fundamental computational question about groups, searching for faster algorithms for group isomorphism is therefore also one way of approaching the graph isomorphism problem. We use collision detection methods to give a classical algorithm that obtains a square-root speedup over the best algorithm previously known for the general group isomorphism problem. For the solvable groups (which are conjectured to be difficult and contain almost all groups), we combine our collision detection techniques with group-theoretic methods to obtain a classical fourth-root speedup over the best algorithm known previously. Prior to this work, it was a longstanding open problem to obtain any improvement for either of these classes of groups. Finally, we give a general framework that yields speedups for many isomorphism problems. In particular, it yields the square-root speedup for testing isomorphism of general groups mentioned above. All of our group isomorphism-testing algorithms also have quantum variants that are slightly faster than their classical counterparts.
The last decade witnessed a great surge of fruitful studies in the new paradigm of quantum computing. Remarkable progress has been made in all the areas including quantum algorithms, quantum complexity theory, quantum error-correcting code, quantum information theory, quantum cryptography and physical implementations of quantum computers. However, quantum algorithm design, probably the most important task in quantum computing, did not make as much progress as people expected. This thesis aims at both designing new quantum algorithms and studying why efficient quantum algorithms are so hard to design by proving many lower bounds on quantum query complexity for various problems.
We first review (and generalize) known techniques for designing quantum algorithms and for proving quantum lower bounds. Using these techniques, we then study the quantum query complexity for many problems, including some specific problems such as Graph Matching and other natural graph properties, the general tuple search, and local search. We also study the lowest possible quantum query complexity for a class of functions as a whole, such as graph properties, circular functions, and functions invariant of a transitive permutation group, where tight results are given in all these cases.
Finally for the quantum lower bounds techniques themselves, we study the main two techniques and show that they are incomparable in power. For the most widely used one, which has a lot of parameters to choose, we show limitations for it in terms of certificate complexity. This implies that we cannot use the method to prove better lower bounds for almost all known open problems in this area.