PLUTO plus : Near-Complete Modeling of Affine Transformations for Parallelism and Locality

Affine transformations have proven to be very powerful for loop restructuring due to their ability to model a very wide range of transformations. A single multi-dimensional affine function can represent a long and complex sequence of simpler transformations. Existing affine transformation frameworks like the Pluto algorithm, that include a cost function for modern multicore architectures where coarse-grained parallelism and locality are crucial, consider only a sub-space of transformations to avoid a combinatorial explosion in finding the transformations. The ensuing practical tradeoffs lead to the exclusion of certain useful transformations, in particular, transformation compositions involving loop reversals and loop skewing by negative factors. In this paper, we propose an approach to address this limitation by modeling a much larger space of affine transformations in conjunction with the Pluto algorithm's cost function. We perform an experimental evaluation of both, the effect on compilation time, and performance of generated codes. The evaluation shows that our new framework, Pluto+, provides no degradation in performance in any of the Polybench benchmarks. For Lattice Boltzmann Method (LBM) codes with periodic boundary conditions, it provides a mean speedup of 1.33x over Pluto. We also show that Pluto+ does not increase compile times significantly. Experimental results on Polybench show that Pluto+ increases overall polyhedral source-to-source optimization time only by 15%. In cases where it improves execution time significantly, it increased polyhedral optimization time only by 2.04x.

Simple crystallizations of 4-manifolds

Minimal crystallizations of simply connected PL 4-manifolds are very natural objects. Many of their topological features are reflected in their combinatorial structure which, in addition, is preserved under the connected sum operation. We present a minimal crystallization of the standard PL K3 surface. In combination with known results this yields minimal crystallizations of all simply connected PL 4-manifolds of ``standard'' type, that is, all connected sums of CP2, S-2 x S-2, and the K3 surface. In particular, we obtain minimal crystallizations of a pair of homeomorphic but non-PL-homeomorphic 4-manifolds. In addition, we give an elementary proof that the minimal 8-vertex crystallization of CP2 is unique and its associated pseudotriangulation is related to the 9-vertex combinatorial triangulation of CP2 by the minimum of four edge contractions.

The cross-entropy method for blind multiuser detection

We consider the problem of blind multiuser detection. We adopt a Bayesian approach where unknown parameters are considered random and integrated out. Computing the maximum a posteriori estimate of the input data sequence requires solving a combinatorial optimization problem. We propose here to apply the Cross-Entropy method recently introduced by Rubinstein. The performance of cross-entropy is compared to Markov chain Monte Carlo. For similar Bit Error Rate performance, we demonstrate that Cross-Entropy outperforms a generic Markov chain Monte Carlo method in terms of operation time.

Rigorous finite-time guarantees for optimization on continuous domains by simulated annealing

Simulated annealing is a popular method for approaching the solution of a global optimization problem. Existing results on its performance apply to discrete combinatorial optimization where the optimization variables can assume only a finite set of possible values. We introduce a new general formulation of simulated annealing which allows one to guarantee finite-time performance in the optimization of functions of continuous variables. The results hold universally for any optimization problem on a bounded domain and establish a connection between simulated annealing and up-to-date theory of convergence of Markov chain Monte Carlo methods on continuous domains. This work is inspired by the concept of finite-time learning with known accuracy and confidence developed in statistical learning theory.

The Linear Ordering Problem Revisited

The Linear Ordering Problem is a popular combinatorial optimisation problem which has been extensively addressed in the literature. However, in spite of its popularity, little is known about the characteristics of this problem. This paper studies a procedure to extract static information from an instance of the problem, and proposes a method to incorporate the obtained knowledge in order to improve the performance of local search-based algorithms. The procedure introduced identifies the positions where the indexes cannot generate local optima for the insert neighbourhood, and thus global optima solutions. This information is then used to propose a restricted insert neighbourhood that discards the insert operations which move indexes to positions where optimal solutions are not generated. In order to measure the efficiency of the proposed restricted insert neighbourhood system, two state-of-the-art algorithms for the LOP that include local search procedures have been modified. Conducted experiments confirm that the restricted versions of the algorithms outperform the classical designs systematically. The statistical test included in the experimentation reports significant differences in all the cases, which validates the efficiency of our proposal.

Extending Distance-based Ranking Models In Estimation of Distribution Algorithms

Recently, probability models on rankings have been proposed in the field of estimation of distribution algorithms in order to solve permutation-based combinatorial optimisation problems. Particularly, distance-based ranking models, such as Mallows and Generalized Mallows under the Kendall’s-t distance, have demonstrated their validity when solving this type of problems. Nevertheless, there are still many trends that deserve further study. In this paper, we extend the use of distance-based ranking models in the framework of EDAs by introducing new distance metrics such as Cayley and Ulam. In order to analyse the performance of the Mallows and Generalized Mallows EDAs under the Kendall, Cayley and Ulam distances, we run them on a benchmark of 120 instances from four well known permutation problems. The conducted experiments showed that there is not just one metric that performs the best in all the problems. However, the statistical test pointed out that Mallows-Ulam EDA is the most stable algorithm among the studied proposals.

Analysis of a transcriptional network involving PU.1, Notch, and Gata3 in the lymphomyeloid lineage decision during early T-cell development

Hematopoiesis is a well-established system used to study developmental choices amongst cells with multiple lineage potentials, as well as the transcription factor network interactions that drive these developmental paths. Multipotent progenitors travel from the bone marrow to the thymus where T-cell development is initiated and these early T-cell precursors retain lineage plasticity even after initiating a T-cell program. The development of these early cells is driven by Notch signaling and the combinatorial expression of many transcription factors, several of which are also involved in the development of other cell lineages. The ETS family transcription factor PU.1 is involved in the development of progenitor, myeloid, and lymphoid cells, and can divert progenitor T-cells from the T-lineage to a myeloid lineage. This diversion of early T-cells by PU.1 can be blocked by Notch signaling. The PU.1 and Notch interaction creates a switch wherein PU.1 in the presence of Notch promotes T-cell identity and PU.1 in the absence of Notch signaling promotes a myeloid identity. Here we characterized an early T-cell cell line, Scid.adh.2c2, as a good model system for studying the myeloid vs. lymphoid developmental choice dependent on PU.1 and Notch signaling. We then used the Scid.adh.2c2 system to identify mechanisms mediating PU.1 and Notch signaling interactions during early T-cell development. We show that the mechanism by which Notch signaling is protecting pro-T cells is neither degradation nor modification of the PU.1 protein. Instead we give evidence that Notch signaling is blocking the PU.1-driven inhibition of a key set of T-regulatory genes including Myb, Tcf7, and Gata3. We show that the protection of Gata3 from PU.1-mediated inhibition, by Notch signaling and Myb, is important for retaining a T-lineage identity. We also discuss a PU.1-driven mechanism involving E-protein inhibition that leads to the inhibition of Notch target genes. This is mechanism may be used as a lockdown mechanism in pro-T-cells that have made the decision to divert to the myeloid pathway.

Essays in mechanism design

This dissertation contains three essays on mechanism design. The common goal of these essays is to assist in the solution of different resource allocation problems where asymmetric information creates obstacles to the efficient allocation of resources. In each essay, we present a mechanism that satisfactorily solves the resource allocation problem and study some of its properties. In our first essay, ”Combinatorial Assignment under Dichotomous Preferences”, we present a class of problems akin to time scheduling without a pre-existing time grid, and propose a mechanism that is efficient, strategy-proof and envy-free. Our second essay, ”Monitoring Costs and the Management of Common-Pool Resources”, studies what can happen to an existing mechanism — the individual tradable quotas (ITQ) mechanism, also known as the cap-and-trade mechanism — when quota enforcement is imperfect and costly. Our third essay, ”Vessel Buyback”, coauthored with John O. Ledyard, presents an auction design that can be used to buy back excess capital in overcapitalized industries.

Descriptive set theory and the ergodic theory of countable groups

The primary focus of this thesis is on the interplay of descriptive set theory and the ergodic theory of group actions. This incorporates the study of turbulence and Borel reducibility on the one hand, and the theory of orbit equivalence and weak equivalence on the other. Chapter 2 is joint work with Clinton Conley and Alexander Kechris; we study measurable graph combinatorial invariants of group actions and employ the ultraproduct construction as a way of constructing various measure preserving actions with desirable properties. Chapter 3 is joint work with Lewis Bowen; we study the property MD of residually finite groups, and we prove a conjecture of Kechris by showing that under general hypotheses property MD is inherited by a group from one of its co-amenable subgroups. Chapter 4 is a study of weak equivalence. One of the main results answers a question of Abért and Elek by showing that within any free weak equivalence class the isomorphism relation does not admit classification by countable structures. The proof relies on affirming a conjecture of Ioana by showing that the product of a free action with a Bernoulli shift is weakly equivalent to the original action. Chapter 5 studies the relationship between mixing and freeness properties of measure preserving actions. Chapter 6 studies how approximation properties of ergodic actions and unitary representations are reflected group theoretically and also operator algebraically via a group's reduced C^*-algebra. Chapter 7 is an appendix which includes various results on mixing via filters and on Gaussian actions.

Structural requirements for catalysis: studies on RTEM-1 β-lactamase

β-lactamases are a group of enzymes that confer resistance to penam and cephem antibiotics by hydrolysis of the β-lactam ring, thereby inactivating the antibiotic. Crystallographic and computer modeling studies of RTEM-1 β-lactamase have indicated that Asp 132, a strictly conserved residue among the class A β-lactamases, appears to be involved in substrate binding, catalysis, or both. To study the contribution of residue 132 to β-lactamase function, site saturation mutagenesis was used to generate mutants coding for all 20 amino acids at position 132. Phenotypic screening of all mutants indicated that position 132 is very sensitive to amino acid changes, with only N132C, N132D, N132E, and N132Q showing any appreciable activity. Kinetic analysis of three of these mutants showed increases in K_M, along with substantial decreases in k_(cat). Efforts to trap a stable acyl-enzyme intermediate were unsuccessfuL These results indicate that residue 132 is involved in substrate binding, as well as catalysis, and supports the involvement of this residue in acylation as suggested by Strynadka et al.

Crystallographic and computer modeling studies of RTEM-1 β-lactamase have indicated that Lys 73 and Glu 166, two strictly conserved residues among the class A β-lactamases, appear to be involved in substrate binding, catalysis, or both. To study the contribution of these residues to β-lactamase function, site saturation mutagenesis was used to generate mutants coding for all 20 amino acids at positions 73 and 166. Then all 400 possible combinations of mutants were created by combinatorial mutagenesis. The colonies harboring the mutants were screened for growth in the presence of ampicillin. The competent colonys' DNA were sequenced, and kinetic parameters investigated. It was found that lysine is essential at position 73, and that position 166 only tolerated fairly conservative changes (Aspartic acid, Histidine, and Tyrosine). These functional mutants exhibited decreased kcat's, but K_M was close to wild-type levels. The results of the combinatorial mutagenesis experiments indicate that Lysis absolutely required for activity at position 73; no mutation at residue 166 can compensate for loss of the long side chain amine. The active mutants found--K73K/E166D, K73KIE166H, and K73KIE166Y were studied by kinetic analysis. These results reaffirmed the function of residue 166 as important in catalysis, specifically deacylation.

The identity of the residue responsible for enhancing the active site serine (Ser 70) in RTEM-1 β-lactamase has been disputed for some time. Recently, analysis of a crystal structure of RTEM-1 β-lactamase with covalently bound intermediate was published, and it was suggested that Lys 73, a strictly conserved residue among the class A β-lactamases, was acting as a general base, activating Ser 70. For this to be possible, the pK_a of Lys 73 would have to be depressed significantly. In an attempt to assay the pK_a of Lys 73, the mutation K73C was made. This mutant protein can be reacted with 2-bromoethylamine, and activity is restored to near wild type levels. ^(15)N-2-bromoethylamine hydrobromide and ^(13)C-2-bromoethylamine hydrobromide were synthesized. Reacting these compounds with the K73C mutant gives stable isotopic enrichment at residue 73 in the form of aminoethylcysteine, a lysine homologue. The pK_a of an amine can be determined by NMR titration, following the change in chemical shift of either the ^(15)N-amine nuclei or adjacent Be nuclei as pH is changed. Unfortunately, low protein solubility, along with probable label scrambling in the Be experiment, did not permit direct observation of either the ^(15)N or ^(13)C signals. Indirect detection experiments were used to observe the protons bonded directly to the ^(13)C atoms. Two NMR signals were seen, and their chemical shift change with pH variation was noted. The peak which was determined to correspond to the aminoethylcysteine residue shifted from 3.2 ppm down to 2.8 ppm over a pH range of 6.6 to 12.5. The pK_a of the amine at position 73 was determined to be ~10. This indicates that residue 73 does not function as a general base in the acylation step of the reaction. However the experimental measurement takes place in the absence of substrate. Since the enzyme undergoes conformational changes upon substrate binding, the measured pK_a of the free enzyme may not correspond to the pK_a of the enzyme substrate complex.

Coding for information storage

Storage systems are widely used and have played a crucial rule in both consumer and industrial products, for example, personal computers, data centers, and embedded systems. However, such system suffers from issues of cost, restricted-lifetime, and reliability with the emergence of new systems and devices, such as distributed storage and flash memory, respectively. Information theory, on the other hand, provides fundamental bounds and solutions to fully utilize resources such as data density, information I/O and network bandwidth. This thesis bridges these two topics, and proposes to solve challenges in data storage using a variety of coding techniques, so that storage becomes faster, more affordable, and more reliable.

We consider the system level and study the integration of RAID schemes and distributed storage. Erasure-correcting codes are the basis of the ubiquitous RAID schemes for storage systems, where disks correspond to symbols in the code and are located in a (distributed) network. Specifically, RAID schemes are based on MDS (maximum distance separable) array codes that enable optimal storage and efficient encoding and decoding algorithms. With r redundancy symbols an MDS code can sustain r erasures. For example, consider an MDS code that can correct two erasures. It is clear that when two symbols are erased, one needs to access and transmit all the remaining information to rebuild the erasures. However, an interesting and practical question is: What is the smallest fraction of information that one needs to access and transmit in order to correct a single erasure? In Part I we will show that the lower bound of 1/2 is achievable and that the result can be generalized to codes with arbitrary number of parities and optimal rebuilding.

We consider the device level and study coding and modulation techniques for emerging non-volatile memories such as flash memory. In particular, rank modulation is a novel data representation scheme proposed by Jiang et al. for multi-level flash memory cells, in which a set of n cells stores information in the permutation induced by the different charge levels of the individual cells. It eliminates the need for discrete cell levels, as well as overshoot errors, when programming cells. In order to decrease the decoding complexity, we propose two variations of this scheme in Part II: bounded rank modulation where only small sliding windows of cells are sorted to generated permutations, and partial rank modulation where only part of the n cells are used to represent data. We study limits on the capacity of bounded rank modulation and propose encoding and decoding algorithms. We show that overlaps between windows will increase capacity. We present Gray codes spanning all possible partial-rank states and using only ``push-to-the-top'' operations. These Gray codes turn out to solve an open combinatorial problem called universal cycle, which is a sequence of integers generating all possible partial permutations.

Diagonal forms, linear algebraic methods and Ramsey-type problems

This thesis focuses mainly on linear algebraic aspects of combinatorics. Let N_t(H) be an incidence matrix with edges versus all subhypergraphs of a complete hypergraph that are isomorphic to H. Richard M. Wilson and the author find the general formula for the Smith normal form or diagonal form of N_t(H) for all simple graphs H and for a very general class of t-uniform hypergraphs H.

As a continuation, the author determines the formula for diagonal forms of integer matrices obtained from other combinatorial structures, including incidence matrices for subgraphs of a complete bipartite graph and inclusion matrices for multisets.

One major application of diagonal forms is in zero-sum Ramsey theory. For instance, Caro's results in zero-sum Ramsey numbers for graphs and Caro and Yuster's results in zero-sum bipartite Ramsey numbers can be reproduced. These results are further generalized to t-uniform hypergraphs. Other applications include signed bipartite graph designs.

Research results on some other problems are also included in this thesis, such as a Ramsey-type problem on equipartitions, Hartman's conjecture on large sets of designs and a matroid theory problem proposed by Welsh.

On erasure coding for distributed storage and streaming communications

The work presented in this thesis revolves around erasure correction coding, as applied to distributed data storage and real-time streaming communications.

First, we examine the problem of allocating a given storage budget over a set of nodes for maximum reliability. The objective is to find an allocation of the budget that maximizes the probability of successful recovery by a data collector accessing a random subset of the nodes. This optimization problem is challenging in general because of its combinatorial nature, despite its simple formulation. We study several variations of the problem, assuming different allocation models and access models, and determine the optimal allocation and the optimal symmetric allocation (in which all nonempty nodes store the same amount of data) for a variety of cases. Although the optimal allocation can have nonintuitive structure and can be difficult to find in general, our results suggest that, as a simple heuristic, reliable storage can be achieved by spreading the budget maximally over all nodes when the budget is large, and spreading it minimally over a few nodes when it is small. Coding would therefore be beneficial in the former case, while uncoded replication would suffice in the latter case.

Second, we study how distributed storage allocations affect the recovery delay in a mobile setting. Specifically, two recovery delay optimization problems are considered for a network of mobile storage nodes: the maximization of the probability of successful recovery by a given deadline, and the minimization of the expected recovery delay. We show that the first problem is closely related to the earlier allocation problem, and solve the second problem completely for the case of symmetric allocations. It turns out that the optimal allocations for the two problems can be quite different. In a simulation study, we evaluated the performance of a simple data dissemination and storage protocol for mobile delay-tolerant networks, and observed that the choice of allocation can have a significant impact on the recovery delay under a variety of scenarios.

Third, we consider a real-time streaming system where messages created at regular time intervals at a source are encoded for transmission to a receiver over a packet erasure link; the receiver must subsequently decode each message within a given delay from its creation time. For erasure models containing a limited number of erasures per coding window, per sliding window, and containing erasure bursts whose maximum length is sufficiently short or long, we show that a time-invariant intrasession code asymptotically achieves the maximum message size among all codes that allow decoding under all admissible erasure patterns. For the bursty erasure model, we also show that diagonally interleaved codes derived from specific systematic block codes are asymptotically optimal over all codes in certain cases. We also study an i.i.d. erasure model in which each transmitted packet is erased independently with the same probability; the objective is to maximize the decoding probability for a given message size. We derive an upper bound on the decoding probability for any time-invariant code, and show that the gap between this bound and the performance of a family of time-invariant intrasession codes is small when the message size and packet erasure probability are small. In a simulation study, these codes performed well against a family of random time-invariant convolutional codes under a number of scenarios.

Finally, we consider the joint problems of routing and caching for named data networking. We propose a backpressure-based policy that employs virtual interest packets to make routing and caching decisions. In a packet-level simulation, the proposed policy outperformed a basic protocol that combines shortest-path routing with least-recently-used (LRU) cache replacement.

Biotechnologies for cancer diagnostics : cell sorting, protein analysis and imaging of cellular metabolism

This thesis presents the development of chip-based technology for informative in vitro cancer diagnostics. In the first part of this thesis, I will present my contribution in the development of a technology called “Nucleic Acid Cell Sorting (NACS)”, based on microarrays composed of nucleic acid encoded peptide major histocompatibility complexes (p/MHC), and the experimental and theoretical methods to detect and analyze secreted proteins from single or few cells.

Secondly, a novel portable platform for imaging of cellular metabolism with radio probes is presented. A microfluidic chip, so called “Radiopharmaceutical Imaging Chip” (RIMChip), combined with a beta-particle imaging camera, is developed to visualize the uptake of radio probes in a small number of cells. Due to its sophisticated design, RIMChip allows robust and user-friendly execution of sensitive and quantitative radio assays. The performance of this platform is validated with adherent and suspension cancer cell lines. This platform is then applied to study the metabolic response of cancer cells under the treatment of drugs. Both cases of mouse lymphoma and human glioblastoma cell lines, the metabolic responses to the drug exposures are observed within a short time (~ 1 hour), and are correlated with the arrest of cell-cycle, or with changes in receptor tyrosine kinase signaling.

The last parts of this thesis present summaries of ongoing projects: development of a new agent as an in vivo imaging probe for c-MET, and quantitative monitoring of glycolytic metabolism of primary glioblastoma cells. To develop a new agent for c-MET imaging, the one-bead-one-compound combinatorial library method is used, coupled with iterative screening. The performance of the agent is quantitatively validated with cell-based fluorescent assays. In the case of monitoring the metabolism of primary glioblastoma cell, by RIMChip, cells were sorting according to their expression levels of oncoprotein, or were treated with different kinds of drugs to study the metabolic heterogeneity of cancer cells or metabolic response of glioblastoma cells to drug treatments, respectively.

Design and analysis of nucleic acid reaction pathways

Nucleic acids are a useful substrate for engineering at the molecular level. Designing the detailed energetics and kinetics of interactions between nucleic acid strands remains a challenge. Building on previous algorithms to characterize the ensemble of dilute solutions of nucleic acids, we present a design algorithm that allows optimization of structural features and binding energetics of a test tube of interacting nucleic acid strands. We extend this formulation to handle multiple thermodynamic states and combinatorial constraints to allow optimization of pathways of interacting nucleic acids. In both design strategies, low-cost estimates to thermodynamic properties are calculated using hierarchical ensemble decomposition and test tube ensemble focusing. These algorithms are tested on randomized test sets and on example pathways drawn from the molecular programming literature. To analyze the kinetic properties of designed sequences, we describe algorithms to identify dominant species and kinetic rates using coarse-graining at the scale of a small box containing several strands or a large box containing a dilute solution of strands.