Indian Institute of Science Bangalore
etd@IISc Electronic Theses and Dissertations at Indian Institute of ScienceNot a member yet
6204 research outputs found
Sort by
Optoelectronic and Magnetic Properties of 2D Layered Organic-Inorganic Hybrids and Selected Transition Metal Oxides
No full textPerovskites with the general chemical formula ABX3 can be categorized into oxides and halides depending on the nature of the X anion. 3D organic-inorganic halide perovskites are extensively studied in the context of solar cell and photo- and electro-luminescence applications due to their outstanding optoelectronic properties, while oxide perovskites have attracted a great deal of attention for their many interesting physical properties such as structural, electrical, magnetic, and magnetocaloric effects. More recently, 2D layered organic-inorganic hybrid (OIH) materials have emerged as a new class of materials with rich optoelectronic properties. They exhibit proven advantages over their 3D counterparts due to their large structural diversity and improved environmental stability against heat and moisture.
2D OIH materials have exhibited many interesting physical properties such as high exciton binding energies, intense photoluminescence, ferroelectricity, and chiro-optical properties. While the Lead-based hybrid perovskites have been very well studied for their spectacular optoelectronic properties, lately there have been attempts to design new materials such as Cd2+, Cu2+, Sn2+ -based hybrid halide materials which offer a new playground in the field of photovoltaic research. The work reported in this thesis explores the ferroelectric properties of Cd2+ - and Cu2+ -based halide materials and discusses the possible microscopic mechanisms for the origin of ferroelectricity in these materials. Further, using experimental and theoretical inputs, it is shown that the Cu2+ -based hybrid materials have outstanding chiro-optical properties. In addition, we also explore the interesting magnetic properties of a few transition metal compounds, exhibiting inverse magnetocaloric effect as well as Griffiths phase in some temperature ranges.
Chapter 1 briefly introduces various concepts relevant to the investigations reported in subsequent chapters of this thesis. The present status of the research in the field of 2D organic- inorganic hybrid materials with an emphasis on various exciting properties such as ferroelectricity, bandgap modulation, and chiro-optical properties has been discussed. This chapter also presents discussions relevant to the family of oxide perovskites with reference to
magnetic properties from fundamental and technological standpoints.
Chapter 2 describes different experimental and theoretical methods that were employed to carry out the studies presented in this thesis.
Chapter 3 presents a detailed study of the successive structural phase transitions of BA2CdCl4. These results establish that these structural phase transitions are associated with intrinsic ferroelectric transitions, from room temperature paraelectric to intermediate temperature ferroelectric, followed by another low-temperature ferroelectric phase. It was widely believed that ferroelectricity in these 2D organic-inorganic hybrid materials originated due to the order-disorder transition of molecular dipoles associated with the organic spacer cations. However, results in this thesis show that there are dipoles also associated with the inorganic components due to local structural distortions. The thesis presents a combination of experimental and theoretical results suggesting that the dipoles associated with the organic spacers and the dipoles originating from these local structural distortions within the inorganic units both play significant roles in the observed ferroelectricity of BA2CdCl4.
Chapter 4 deals with the chiro-optical properties of quasi 2D (R-/S-MBA)2CuBr4 hybrid material. We discussed the role of chiral organic amine cations on the optical properties of these hybrid materials. Few Lead based compounds and one copper-based sample show giant chiro-optical properties but these are chirally active only for wavelengths < 490nm, limited by their large bandgaps. (R-/S-MBA)2CuBr4 shows a relatively large chiro-optical property in the orange-red part of the visible spectrum. Structural analyses of these compounds show that these are made of alternating layers of the chiral organic units and an inorganic layer of isolated CuBr4 units. Such isolated inorganic units distinguish this class of compounds from the more intensely investigated hybrid lead halide systems where the basic PbX6 (X = Cl, Br, or I) units
are linked together by corner sharing of the halide ions, making them intrinsically 2D systems. The present Cu-based system would have qualified as a 0D system but for the Cu-Br….Br-Cu interactions that allow the separated CuBr4 units to interact, making the system a quasi-2D system. This subtle structural aspect plays an important role in giving this system remarkable chiro-optical properties. The semi-isolation of the CuBr4 units allows them to be rotated along
the 21 screw-axis by the chiral organic units via strong hydrogen bonding, thereby imparting the giant chirality to the entire hybrid system. Simultaneously, the connectivity of the CuBr4 units via Br…Br interactions imparts a quasi-2D character helping to achieve a broadband absorption, thereby extending the chiro-optical properties to longer wavelengths.
Chapter 5 discusses tuning of the bandgap while retaining ferroelectricity through halide substitution in Cu2+ -based chiral 2D hybrid materials to obtain small bandgap ferroelectric materials. Search for such small bandgap ferroelectric materials has been popular in the literature, not only because most ferroelectrics tend to have large bandgaps but also because of their obvious applications in solar photovoltaics. The ferroelectric Lead-based hybrid perovskites have been very well studied for their rich optoelectronic properties that are relevant in photovoltaic applications, but all are having a relatively higher bandgap. We have lowered the ligand to metal charge transfer bandgap from ~2.53 to 2.09 eV while retaining ferroelectricity in a copper chloride based low-dimensional hybrid material through partial substitution of chlorine with bromine. Our results show that a complete substitution of Cl- by Br- leads to a bandgap of ~1.62 eV with a loss of the ferroelectric state in the pure bromide material.
Chapter 6 discusses the low temperature magnetic state along with the main ferromagnetic ordering at ~200 K and the magnetocaloric effect of double perovskite, Nd2NiMnO6. An earlier study on this compound established that for any applied magnetic field lower than 3 T, these samples show a downturn in M(T), and any field higher than 3 T, shows an upturn in M(T) for the temperature range below 100 K. This has been interpreted as Nd moments experience an effective ~3 T internal magnetic field due to the presence of the ordered Ni-Mn ferromagnetic sublattices. This indicates that the low temperature magnetic state of this compound is easily influenced by an externally applied magnetic field in the tune of 3 T, suggesting possible interesting magnetocaloric effects in this material. This chapter presents a detailed study of the magnetocaloric properties of Nd2NiMnO6. Interestingly, it shows a significant inverse magnetocaloric effect (IMCE) at low temperatures (T < 50 K) together with a significant conventional magnetocaloric effect (CMCE) at the ferromagnetic ordering temperature (Tc ~200 K). IMCE and CMCE correspond to the antiferromagnetic arrangement of Nd and Ni–Mn sublattices and ferromagnetic ordering of Ni–Mn sublattices, respectively. Nd2NiMnO6 with its second order phase transition follows the universal behavior of magnetic entropy change, ΔSM(T); it also shows a power law dependency on the magnetic field as ΔSM ∝ .
Chapter 7 deals with Griffiths phase-like magnetic anomalies in disordered La0.85Sr0.15CoO3 induced by chemical doping. In earlier studies on doped LaCoO3 with doping concentration higher than the percolation threshold (18%) shows non-Griffiths phase-like behavior. But no reports are available for the composition just below the percolation limit in this context. So, we chose the 15% doping concentration and explored its magnetic properties carefully. Our results
establish that this composition shows typical Griffiths phase-like behavior in the intermediate temperature range and followed by spin glass behavior below 60 K. The existence of nanoscale ferromagnetic clusters below 240 K contributes to the total magnetization of the system for low applied magnetic fields resulting in a downturn of the χ-1 vs. T plot. The extent of this downturn is strongly suppressed by increasing the dc applied magnetic field, a typical signature of
Griffiths phase.
In the appendix, we present results of investigating Cs- and Na-doped WO3 exhibiting strong absorption in the near-infrared (NIR) and transmittivity in the visible range. Despite several publications, there is a lack of agreement in the community on the origin of this strong optical absorption with competing claims of polaronic and plasmonic origins. We address this controversy by first investigating bulk samples that are relatively free of complications arising from any shape anisotropy; we show by combining experimental and theoretical results that all spectral features in both bulk and nanoparticle samples are consistent with plasmonic excitations, without any need to invoke a polaronic mechanism. Doped WO3 exhibits strong optical absorption primarily due to surface plasmon resonances in colloidal nanoparticles,
while their bulk counterparts are dominated by bulk plasmonic features. Investigating systems with different crystal structures and charge doping levels, we established that the complex spectral features of these plasmonic absorption bands for both bulk and nano samples are dominated by the underlying structure-dependent anisotropic electronic properties that determine the plasmonic feature
Monitoring and Control of Intermittent Water Distribution Network
No full textMillions of people in low and middle-income countries worldwide receive water only for a small fraction of the time in a day through piped networks. Even though the percentage of the world’s population with access to piped water has increased, the number of individuals receiving water intermittently is projected to rise. Water supply systems are designed for a continuous supply of water to consumers with a standard demand pattern. However, an unforeseen increase in demand, reduced network capacity, ease of operation and climate change results in Intermittent Water Supply. Because the intermittent water supply goes through filling and draining cycles regularly, the water is not always safe to drink. Due to ageing infrastructure, poor management, and system operation, the amount of Unaccounted For Water in these cities is extremely high and leads to inequitable water distribution. The water supply in developing countries is insufficient to meet demand, and there is a significant imbalance between supply and demand. This thesis examines the influence of intermittently supplied piped water on water inequity, the impact on the quality of water delivered, and possible solutions to intermittent supply.
A framework for understanding the effect of inequitable distribution due to intermittent supply is developed first. Economic and geographical characteristics of the DMA, such as insufficient infrastructure, high Unaccounted For Water, socio-economic status, and so on, all contribute to inequity in the intermittent water supply. The inequality in intermittent water supply between 10 divisions and 83 DMAs in the South division of Bangalore, India, are examined in this study. The inequity is divided into four categories, namely: (i) inequity in water sharing, (ii) inequity in time of supply, (iii) inequity in the duration of supply and (iv) inequity in alternate supply. The supply, UFW, and consumption data were collected for a period of 36 months. The presence of inequity in water sharing was indicated by the Lorenz curve and Gini coefficient. The Kullback-Leibler divergence, Cauchy-Schwarz inequality, and Euclidean distance among supply
zones revealed inequity in supply time and duration. Inequity in the supply duration and an alternate source of supply was also discussed using the Lorenz curve and Gini coefficient. The current analysis revealed a severe disparity in all four categories, showing that the IWS with poor infrastructure leads to inequitable water distribution. It was also clear that, despite significant savings in UFW, inequity remained largely unchanged. The most efficient way to counter the
inequity in IWS is to move towards a continuous system. We study this phenomenon of Intermittent Water Supply with the city of Mysore, India, as an example, understand the issues therein and address them with a systematic procedure to achieve Continuous Water Supply. It is essential to solve the problem of Intermittent Water Supply due
to the adverse effects caused. IWS can create inequity at various levels: command areas, DMAs, supply and sub-supply zones. On the contrary Continuous Water Supply has additional benefits like improved water quality, increased customer satisfaction, and better supply and demand man- agement. To achieve CWS, we formulate a convex cost minimisation problem with appropriate penalties to i) maintain the reservoirs at the desired level, ii) obtain a periodic solution and iii) avoid frequent changes in the valve configuration. The network-wide problem formulation helps vii compensate local infrastructure limitations using the excess infrastructure available elsewhere in the network. We then optimally control the water inflow into the reservoirs and meet daily demand. We propose a Simultaneous Perturbation Stochastic Approximation (SPSA) based gradient algorithm on solving the optimisation problem. A hydraulic modelling toolkit EPANET is used to compute the flows in the water network. We then initialise the reservoirs to their desired target levels using a similar approach, starting from arbitrary initial levels. The proposed algorithm was applied to two water networks in Mysore, India. The essential contribution here is to verify the possible enhancement of water supply systems by network-wide control. Risks to water quality are exacerbated due to intermittency of supply, mainly due to intrusion and back-flow, as well as variation in bio-film, deposits, and microbial growth. Chlorine is a widely used disinfectant in the drinking water treatment process, mostly in developing countries, due to its efficiency and affordability. Therefore, assessing the effect of intermittency in the drinking water supply on chlorine decay is needed to ensure safe drinking water delivery through piped networks in these countries. In this work, field sampling for chlorine decay measurement was carried out in adjoining continuous and intermittent water supply networks in Bangalore,
India. The two networks received water from the same source. The first-order model was used to calculate chlorine decay rate in different pipe segments distributed across the networks. The effect of network parameters such as pipe velocity, pipe material, and pipe diameter on chlorine degradation rate was investigated. Further chlorine decay rates in continuous and intermittent networks were compared, considering the influence of these network properties. The analysis indicates that decay rates varied significantly with velocity in both networks, with reactive ma- terials being more sensitive to velocity variation than non-reactive materials. Reactive materials had higher decay rates than non-reactive materials, and decay rates increased with a decrease in pipe diameter, concurring with trends found in previous literature. The comparison of decay rates across networks showed that the average decay rate in the intermittent network was 1.5 times higher than the average decay rate in the continuous network, indicating that intermittency
in the water supply has a long-term effect on chlorine decay. Contaminants can infiltrate a water distribution network at any time or location due to the ageing infrastructure, accidental discharge, or malicious actors. Contaminants such
as total coliform bacteria, organic compounds, and others create serious health issues in a large population served by the WDS. This is especially prevalent in intermittent water supply, and it can lead to a loss of faith in the water utility. The detection of intentional or accidental contamination in a water distribution system (WDS) is necessary to ensure that society has access to adequate safe drinking water. One of the techniques for detecting contamination in
WDS is to install the sensor. In the literature, determining the best location for the sensor is still a work in progress. We proposed ”EQ-water,” a new method for identifying sensor location which works for both continuous and intermittent water distribution systems. The EQ-water is based on complex network theory and utilizes very little hydraulic information. We evaluated the proposed method’s performance on four real networks, BWSN 1, BWSN 2, JPN, and D2B Bangalore networks. Furthermore, we compared the results to all of the BWSN’s proposed approaches and existing complex network-based methods. We employed TEVA-SPOT to assess the methods’ performance on four different objective functions, as well as the cumulative objective function. EQ-Water has the best performance in BWSN 1, JPN, D2B Bangalore and average performance in BWSN 2. Utility boards in India are ill-equipped to deal with Unaccounted For Water (UFW) in intermittent water supply, with leakage being the primary cause of UFW. Conducting controlled
independent experiments in the field to detect leaks or study water quality is a time-consuming process. It is relatively difficult to interfere with a running water supply to conduct experiments and collect measurement data. As a result, at our R&D centre in IISc, Bangalore, we developed an experimental water supply setup. Detecting leaks in intermittent water supplies have not been studied in the literature; thus, in this study, we created such scenarios and used
a data-driven approach to detect leaks in laboratory conditions. The goal of the work is to create an intermittent supply in the lab that mimics the real-world water supply of Bangalore, south, and to introduce various leak scenarios. Using the proposed data analytic approach using a log-likelihood model in conjunction with CUSUM, we then use this data to train the model. The testing data results show that leaks in intermittent supply systems can be detected and localised.IMPRINT initiative of Ministry of Human Resource \& Development and Ministry of Housing and Urban Affairs, Govt of India [Project code: 5786 and Sanction No: F.No.3-18/2015-TS-TS.I.] and Robert Bosch Center For Cyber Physical System
Carbon Dots for Next-generation Artificial Lighting Devices
No full textEnergy conservation has become a major concern globally in today’s world and artificial lighting (AL) accounts to ~50% of total energy consumption. With increase in population accompanied by modern civilization there is a tremendous rise in the use of artificial lighting sources leading to energy crisis in near future. So there rises a high demand for the fabrication of energy efficient sources for artificial lighting. Traditional AL sources involve the use of Incandescent lamps (ILs)/Halogen Lamps (HLs) and Compact fluorescent lamps (CFLs) for lighting and the commercially used AL sources use white LEDs. Recent advancements in the fabrication of white LEDs involve the use of Quantum dots as down converting phosphor material coated on blue emitting InGaN/GaN LEDs. Quantum dots (QDs) are semiconductor nanoparticles with confinement of charge carriers and exhibit size/surface dependent emission properties. Semiconductor quantum dots as phosphor material for WLEDs exhibit size-tuneable emissions, high photoluminescence quantum yields, low scattering compared to traditional phosphors. White LEDs reported using semiconductor QDs, show potentially toxic behaviour to be used for indoor and outdoor lightening due to the toxicity of precursors involved. Further these dots show self-absorption loses effecting the color rendering index and efficiency of the fabricated device. Due to these shortcomings, there is a high demand for the synthesis of benign nanomaterial with similar or better optical properties.
The emerging carbon dots (CDs) seems to be an alternative to semiconductor dots for white LEDs due to the wide availability of carbon in nature and broad emissive behaviour of CDs with least toxicity increasing their potential applications in fabricating white LEDs. Compared to semiconductor QDs CDs have drawn a great attention among researches across the globe due to its good bio-compatibility, non-toxicity, photo bleaching, photo blinking, excellent water solubility, chemical stability, ease of surface functionalization and large two photon cross section areas, which makes them highly beneficial for applications in fabricating white LEDs. Also, photoluminescence properties of carbon dots can be controlled by modifying the size and surface of carbon dots fabricated. This surface modification of CDs enhances the intensity of PL emission from CDs that has a direct impact on increasing efficiency of white LED fabricated. Despite these advantages of CDs, they have few drawbacks in excitation dependent and dilution dependent emissive behaviour that can alter the CRI of the WLED in the long run. Also, although few reports have been published using CDs for white LED fabrication, However, the luminance, color-rendering index (CRI) doesn’t yet meet the requirements for practical application. Therefore, it is still urgent; to develop novel CD phosphor with enhanced properties for white LEDs. This thesis focuses on the fabrication of highly stable single system white light emitting carbon nanoparticles. The fabricated CDs show non photo bleachable behaviour and dilution independent emissive behaviour. Further luminescence from CDs is tailored using surface passivation and functionalization routes. Photoluminescence mechanism in CDs and reaction parameters effecting quantum yield of CDs are explored. In our next work plastic waste generated during covid-19 in the form of face masks, face shield, gloves, syringes and other plastics are recycled to white light emitting nanoparticles. This work of generating light from waste will reduce environmental plastic pollution (raised beyond limited due to Covid-19). Further we use the fabricated nanoparticles as fluorescent markers, active emitters in LEDs and in bio imaging
Weights of highest weight modules over Kac-Moody algebras
No full textIn this dissertation, broadly, we treat the weight-sets of arbitrary highest weight modules (uniformly) over all general complex Kac-Moody Lie algebras ,
achieving the below.
We obtain a uniform, explicit, cancellation-free and positive formula for the weight-sets of all highest weight modules (of all highest weights) over all Kac-Moody .
Interestingly, our formula for the weights of involves basic ingredients, namely the independent subsets in the Dynkin diagram of determined by , and nothing else!
Prior to our work, it seems that even for all (non-integrable) simple highest weight -modules - despite these being treated for over a half century - their weight-formulas were not known until a few years ago.
Namely, these were written-up in Khare [J. Alg. 2016] and Dhillon-Khare [Adv. Math. 2017 and J. Alg. 2022]; and our general formula recovers the formulas for simples in these papers.
Leaving out general Kac-Moody or even general semisimple settings, given an arbitrary highest weight module over or in particular , even in this case the weights of were not written down in the literature to the best of our knowledge.
More importantly, our weight-formula naturally drives us to introduce and study a finite family of ``higher order Verma modules" over Kac-Moody , for every highest weight and (any) collection of independent subsets in the Dynkin diagram of .
This family generalizes and includes: all Verma modules at zeroth order level, and the parabolic Verma modules (which were introduced and studied by Lepowsky, Kumar, Mathieu,... to name but a few) at first order level.
Importantly, our higher order Verma modules are crucial and universal for weight considerations:
their weight-sets (which are finitely many when we fix their highest weight) are pairwise disjoint and exhaust the weight-sets of all highest weight -modules.
In this thesis, we also initiate the study of the characters of these universal modules, by computing them via BGG-type resolutions in certain cases.
Next, a phenomenon in root systems which rewarded us with many applications on the weights side.
First, recall the partial sum property for Kac-Moody root systems: every root of is an ordered sum of simple roots such that each partial sum is also a root.
The course/journey to finding and proving our weight-formula mentioned above begins from proving a parabolic-generalization of this property, which we call as the parabolic partial sum property.
Given a subset of simple roots, it allows one to write any (positive) root involving some simple roots from as: an ordered sum of roots, in which each root involves exactly one simple root from (unit -height roots) and with each partial sum also being a root.
The parabolic partial sum property has been devised in order to obtain a ``minimal" description for the weights of all (non-integrable) highest weight simples, which was posed by Khare.
This thesis exhibits such minimal descriptions as an immediate application of the parabolic partial sum property;
in fact we will more strongly show this property at the level of Lie words for any general Lie algebra graded over any free abelian semigroup.
There is also another generalization of the partial sum property due to Khare and Kumar for highest weight simples in finite type, which we extend to the Kac-Moody setting in this thesis.
There is another notable application of the parabolic partial sum property shown in this thesis. Chari and her coauthors [Adv. Math. 2009 and J. Geom. Phys. 2011] introduced and studied certain combinatorial subsets called weak faces and -closed subsets of finite root systems. This is in order to construct Koszul algebras, study Kirillov-Reshetikhin modules over specializations of quantum affine algebras, and also to classify nilpotent ideals in the parabolic Lie subalgebras of finite type , etc.
These subsets (say subsets of a set ) are some discrete analogues/ generalizations of the faces for convex sets (the faces of the convex hull of ).
In this thesis, we are concerned with such subsets of a weight-set, generalizing the faces for convex hulls of weight-sets.
Using the weak faces for weight-sets of finite-dimensional simples (in finite type) Khare extended the aforementioned results of Chari et al.
Motivated by the applications of these two modern notions to representation theory, we completely classify the weak faces and -closed subsets of weights of all highest weight modules (again uniformly) over all Kac-Moody ; extending and completing the partial classification results of Chari, Khare, and their co-authors from finite type.
We more strongly show that both these notions are the same as the weights falling on the faces, for the convex hulls of these weight-sets.
This shows the equivalence of these two notions, to the classical faces for the convex hulls of weights (which have been pursued from the 1960s).NBHM Ph.D. Fellowship (Ref. No. 2/39(2)/2016/NBHM/R&D-II/11431) and by a Swarnajayanti Fellowship from the DST and SERB (Govt. of India)
Explainable and Efficient Neural Models for Natural Language to Bash Command Translation
No full textOne of the key goals of Natural Language Processing is to make computers understand natural language. Semantic Parsing has been one of the driving tasks for Natural Language Understanding. It is formally defined as the task of generating meaning representation from natural language input. In this work, we focus on using the Bash command as the meaning representation. Bash is a Unix command language used for interacting with the Operating System. Recent works on natural language to Bash command translation have made significant advances on this problem. The best performing solutions employ a neural network architecture called the Transformer. In this work, we explore the aspects of explainability and efficiency for this task and use the Transformer as one of the baselines for comparing the proposed approaches.
In the first part, we utilize documentation data from Linux manual pages and the Abstract Syntax Tree for Bash to generate explanations for the translated Bash command. We propose a novel architecture that incorporates tree structure information in the Transformer and provides explanations for its predictions via alignment matrices between user invocation and manual page text. We find that the proposed method performs on par with the Transformer performance. Our method performs better than fine-tuned T5, a Transformer-based neural model pre-trained on a large amount of text data in a self-supervised manner.
In the second part, we use the problems inherent synchronous structure and propose the Segmented Invocation Transformer (SIT) that utilizes the information from the constituency parse tree of the natural language invocation. Our method is motivated by the alignment between segments in the natural language text and Bash command components. By utilizing this structure, the proposed method outperforms the state-of-the-art approach while achieving a 1.8x improvement in the inference time (as measured on a CPU) and a 5x reduction in model parameters. We also conduct an attribution analysis using Integrated Gradients to empirically confirm the identified structure and the ability of SIT to capture it
A systems-level exploration of host responses across the disease spectrum of Tuberculosis and COVID-19
No full textThe host immune system orchestrates a defense mechanism against any invading pathogen with a combination of many mechanisms including antibody production, T-cell response, and memory response. The manner and extent of host immune response play a key role in deciding the fate of any infection, whether it manifests only as a mild disease or a lethal one. While a timely and robust immune response is necessary for a successful defense against a pathogen, in some cases a dysregulated immune response can also cause increased immunopathology. This work aims to explore different aspects of immune response across the disease severity spectrum of tuberculosis and COVID-19. Tuberculosis (TB), caused by Mycobacterium tuberculosis, is among the oldest known infectious disease to mankind. Despite over a century’s research on tuberculosis, it is not been successfully managed on a global scale. In late 2019, a coronavirus, SARS-CoV-2 induced COVID-19 emerged as the newest pandemic claiming over 6 million lives in only 2.5 years. Despite their many differences, host genetics and immune response are often the major deciding factors of the outcome of infection in both tuberculosis and COVID-19. Upon getting exposed to Mycobacterium tuberculosis the host can either completely clear the bacteria, or successfully prevent it from replicating and causing disease but fail to remove it completely (latent tuberculosis), or fail to control the infection leading to symptomatic and contagious disease (active tuberculosis). Stages of the disease have also been documented that fall in between latent and active tuberculosis based on the bacterial burden and state of symptoms, such as incipient tuberculosis and subclinical tuberculosis. In the case of COVID-19, most of the patients successfully mounted an immune response to clear the virus with no to mild symptoms (asymptomatic and mild). On the other hand for several others, especially those with a compromised immune response caused by aging or other comorbidities, the disease caused more serious symptoms often requiring external oxygen support (moderate), ICU and ventilation (severe), and even causing organ failure and septic shock (critical and fatal). Understanding different aspects of the host response are critical to finding strategies for successful management of both of these diseases. With the rapid technological advances in different unbiased data generation such as genomics, transcriptomics, and proteomics and sharing of data among the scientific community, the systems-level analysis provides unique opportunities to study a disease from a global perspective and identify the most important perturbations in a system in an unbiased manner. This thesis describes a systems biology approach to understanding the host and pathogen biology across the spectrum of these two infectious respiratory diseases - tuberculosis and COVID-19.
The first part of the thesis (Chapters 2-4) focuses on tuberculosis from a host and pathogen perspective. The whole blood transcriptome can capture systemic immune perturbations in a disease. Although traditional transcriptome analysis has been successful in elucidating different aspects of immune response in active tuberculosis, no clear pattern of systemic host response has so far been linked with latent tuberculosis, as molecular correlates of latent infection have been hard to identify. This work (Chapter 2) uses individualized response network analysis (an in-house algorithm to mine information on important biological processes for large-scale data) to find global perturbations in the host response to latent tuberculosis as compared to uninfected individuals. We specifically focus on the heterogeneity in such responses observed across latent tuberculosis cohorts. We identified the most frequently perturbed immune pathways despite the heterogeneity that are used by the host to maintain the latency of tuberculosis, such as the interferon-γ/interleukin-12 axis, tumor necrosis factor α (TNFα), epidermal growth factor receptor (EGFR) signaling, transforming growth factor β (TGFβ) signaling, etc. The analysis identified patterns of perturbation of these immune response pathways among latent tuberculosis patients despite the high heterogeneity in the gene expression profiles, grouped them into immune subtypes, and identified the subtype most likely to undergo reactivation into active TB. On the other end of the TB spectrum is the active disease. The immune response undergoes significantly large perturbations at both systemic and localized levels in active tuberculosis. Among the different pathways that are activated only in active tuberculosis patients, but not in latent or uninfected cases, pro-inflammatory leukotriene molecules play a significant role. Next, we studied the Mtb pathogen physiology in latent tuberculosis, where the bacteria stays in a dormant non-replicating yet viable condition. Since most anti-TB drugs target essential processes in actively replicating mycobacteria, their efficacy is highly reduced against dormant Mtb. Thus an ideal drug for latent tuberculosis treatment should target cellular processes that are essential for the survival of the non-replicating pathogen. In Chapter 3 we utilized systems modeling of dormant Mtb using flux balance analysis and network analysis to understand the dormancy mechanisms and identify 6 potential drug targets specific for latent tuberculosis treatment. These targets could be associated to lead molecules from approved drugs from DrugBank, showing the possibility of drug-repurposing for these targets. Following this, the focus was shifted to active TB (Chapter 4) and on the role of leukotriene B4 in this stage of tuberculosis. The important effector molecules of leukotriene B4 signaling in tuberculosis patients, STAT1/2 and NADPH oxidase, are identified with transcriptome integrated response network analysis. Further, the identified downstream mediators of leukotriene B4 signaling are experimentally validated in Mtb infected macrophages. The potential for targeting this pathway as a possible mode of host-directed therapy has also been assessed which showed that inhibition of leukotriene B4 signaling significantly reduces bacterial growth in infected THP-1 cells. The same was also seen in murine infection models.
The second part (Chapters 5-6) of the thesis focuses on COVID-19. Despite the large number of systems-level studies being performed on cohorts from China, North America, and European countries, similar studies from India have been very limited, despite the huge number of cases and fatalities observed. Understanding the immune response in different severities in the Indian population would provide important insights into the disease manifestation in this distinctly different genetic and environmental background. In Chapter 5, an Indian cohort has been generated at the early stages of different COVID-19 severities, and whole blood transcriptome has been utilized to study the differences between their systemic immune response. This study identified some of the critical differences among perturbed pathways between mild and severe COVID-19 at the early stage of disease onset. Suppression of the systemic immune system was observed in the early stages of severe COVID-19 patients. Timely mounting of robust type-I interferon response and classical complement pathway was found to be critical for reduced severity whereas repressed class I
MHC mediated antigen presentation and overall activation in translation pathways were key features in more severe patients. During the outbreak of COVID-19 in 2020 and 2021, another major concern faced by the clinicians in the over-burdened healthcare facilities was the presence of bacterial coinfection with COVID-19 which increased morbidity. It was a major challenge to correctly diagnose this condition to decide whether or not to administer antibiotics. The efficacy of the gold standard test for bacterial infection, the culture sensitivity test, was severely affected due to the widespread use of antibiotics in COVID-19 patients, leading to presumptive bacterial coinfection diagnosis and prescription of high doses of antibiotics in suspected coinfection cases. The unreliability of these tests could lead to both overprescriptions of antibiotics in a false-positive diagnosis and lack of necessary treatment in a false-negative diagnosis. In Chapter 6, the differences between the immune responses in a COVID-19 patient and a suspected coinfection case were explored. From a pool of host blood genes (identified in a previous meta-analysis) known to be perturbed only in confirmed bacterial infections but not in viral infections, a 9-gene signature was identified as a diagnostic tool to increase the confidence of bacterial co-infection diagnosis in COVID-19 patients. The gene signature and the score formulated from the same showed high accuracy, sensitivity, and specificity in distinguishing probable bacterial coinfection cases from only COVID-19 patients.
In summary, this work provides systems-level insights into the systemic immune responses in different stages of tuberculosis and COVID-19 diseases. The heterogeneous systemic host response in latent tuberculosis was successfully classified into subtypes. Our analysis also led to the identification of key effector molecules of leukotriene B4 signaling in active tuberculosis and explored its potential as a mode of host-directed therapy. The study of the dormant pathogen allowed us to identify 6 potential drug targets against latent tuberculosis. Together these works open up new possibilities in TB treatment and can be explored further for their translational potential. On the other hand, this thesis explored the key differences between the systemic immune responses in mild and severe COVID-19 at early stages and identified the key responses whose timely activation is necessary for lower disease severity. At the same time, a 9-gene signature and score were also identified from COVID-19 patients’ blood transcriptome that can assist in the diagnosis of bacterial coinfection in culture-negative cases.DB
HYDRA: A Dynamic Approach to Database Regeneration
No full textDatabase software vendors often need to generate synthetic databases for a variety of applications, including (a) Testing database engines and applications, (b) Data masking, (c) Benchmarking, (d) Creating what-if scenarios, and (e) Assessing performance impacts of planned engine upgrades. The synthetic databases are targeted toward capturing the desired schematic properties (e.g., keys, referential constraints, functional dependencies, domain constraints), as well as the statistical data profiles (e.g., value distributions, column correlations, data skew, output volumes) hosted on these schemas.
Several data generation frameworks have been proposed for OLAP over the past three decades. The early efforts focused on ab initio generation based on standard mathematical distributions. Subsequently, there was a shift to database-dependent regeneration, which aims to create a database with similar statistical properties to a specific client database. However, these mechanisms could not mimic the customer query-processing environments satisfactorily. The contemporary school of thought generates workload-aware data that uses query execution plans from the customer workloads as input and guarantees volumetric similarity. That is, the intermediate row cardinalities obtained at the client and vendor sites are very similar when matching query plans are executed. This similarity helps to preserve the multi-dimensional layout and flow of the data, a prerequisite for achieving similar performance on the client’s workload. However, even in this category, the existing frameworks are hampered by limitations such as the inability to (a) provide a comprehensive algorithm to handle the queries based on core relational algebra operators, namely, Select, Project, and Join; (b) scale to big data volumes; (c) scale to large input workloads; and (d) provide high accuracy on unseen queries.
In this work, motivated by the above lacunae, we present HYDRA, a data regeneration tool that materially addresses the above challenges by adding functionality, dynamism, scale, and robustness. Firstly, extended workload coverage is provided through a comprehensive solution for modeling select-project-join relational algebra operators. Specifically, the constraints are represented as a linear feasibility problem, in which each variable represents the volume of a partitioned region of the data space. Our partitioning scheme for filter constraints permits the regions to be non-convex and ensures the minimum number of regions, thereby hugely reducing the problem complexity as compared to the rectangular grid-partitioning advocated in the prior literature. Similarly, our projection subspace division and projection isolation strategies address the critical challenge of capturing unions, as opposed to summations, in incorporating projection constraints. Finally, by creating referential constraints over denormalized equivalents of the tables, Hydra delivers a comprehensive solution that also handles join constraints.
Secondly, a unique feature of our data regeneration approach is that it delivers a database summary as the output rather than the static data itself. This summary is of negligible size and depends only on the query workload and not on the database scale. It can be used for dynamically generating data during query execution. Therefore, the enormous time and space overheads incurred by prior techniques in generating and storing the data before initiating analysis are eliminated. Our experience is that the summaries for complex Big Data client scenarios comprising over a hundred queries are constructed within just a few minutes, requiring only a few MBs of storage.
Thirdly, to improve accuracy towards unseen queries, Hydra additionally exploits metadata statistics maintained by the database engine. Specifically, it adds an objective function to the linear program to pick a solution with improved inter-region tuple distribution. Further, a uniform distribution of tuples within regions is modeled to obtain a spread of values. These techniques facilitate the careful selection of a desirable database from the candidate synthetic databases, and also provide metadata compliance.
The proposed ideas have been evaluated on the TPC-DS synthetic benchmark, as well as real-world benchmarks based on the Census and IMDB databases. Further, the Hydra framework has been prototyped in a Java-based tool that provides a visual and interactive demonstration of the data regeneration pipeline. The tool has been warmly received by both academic and industrial communities
Design and Development of Hybrid Metal and Polymer Additive Manufacturing System
No full textAdditive manufacturing (AM) is a layer-based manufacturing process aimed at producing parts directly from a Computer-aided design (CAD) model. There are various types of AM systems, which can be classified based on: (i) the base material being used for fabrication, such as polymers, ceramics, and metals; (ii) indirect and direct processes depending on the bonding method; and (iii) the state of the input raw material, i.e., liquid, molten, powder, and solid layer. The current research in AM processes includes technology development for the printing of multi-material parts using two or more materials such as metal, polymer, glass, ceramics, and graphene. The multi-material additive manufacturing (MMAM) processes are complex and challenging due to the significant differences in material deposition techniques, material processing temperature, or pre/post-processing methods involved for an individual material. Our work focuses on the hybrid metal/polymer printing process where liquid metal printing is achieved using a novel design of a molten metal droplet-on-demand (MMDoD) system. The metal is fed into the MMDoD system in the form of solid wire and is melted using a zero-voltage-switching circuit based induction heater. The magnetic field, eddy current density, and power transfer from the induction coil to the molten metal pool are studied using experiments, theoretical formulation, and FEM simulations. The influence of workpiece geometry on the induction heating process is also studied for solder alloy and aluminum billets. These studies show that for a given geometry of induction coil and workpiece, the power transferred to the workpiece is a non-monotonic function of the workpiece’s resistivity. Also, the heating rate of the workpiece depends on the thermal mass and the magnetic field flux in and around the workpiece. Using these studies, the resistivity of the workpiece, and the geometry of the workpiece and induction coil, can be chosen to achieve faster heating and melting of the metal.
Once the raw material is in the liquid state, it can be used to generate molten metal droplets (MMDs). To generate the MMD, a novel MMDoD system is designed and developed using a thermally insulating piston and magnetostrictive actuator. Using the MMDoD mechanism, the molten metal is deposited on the printing bed surface (glass) or partially formed part (metal or polymer – PLA/ABS). To find the optimal parameters of MMD generation process, a parametric study of the MMDoD mechanism is conducted by varying the size and material (Brass, Stainless-Steel, Nickel-plated steel alloy) of the nozzle, the gap between nozzle and piston, unfiltered vs. low-pass filtered actuation pulse, and the actuation pulse amplitude. This shows the following regions where, DoD process is not achieved, and the DoD is achieved with the generation of single or multiple droplets for each actuation. The droplet size, Feret width and length, and standard deviation are measured using snapshots from the high-speed camera of the droplet formation process. The region where a single MMD is generated for each actuation of the MMDoD mechanism with the least standard deviation is most desirable for the reproducible metal AM process. A parametric study is conducted to find the optimal printing parameters of the metal AM system by varying the gap between each droplet to print the 2D connected metal lines on the substrate. Other parameters like the size of MMDs, droplet ejection rate (20Hz), and liquid metal temperature are kept fixed. The 3D metal printing can be achieved by printing these metal lines layer-by-layer.
MMDoD system is extended to multi-material additive manufacturing (MMAM) system by combining it with polymer extrusion system. The designed MMAM system consists of a controller board to control the overall system, an induction heater, a computer numeric control (CNC) build platform/positioning system, MMDoD mechanism, and the polymer extruder. The system is designed and developed to print metal (Solder alloys - Sn99Cu1, Sn63Pb37, and Sn96.5Ag3.5) with polymer (PLA and ABS). To demonstrate the hybrid AM of metal and polymer, a few mechanical structures (2D planar text, hollow tube, hollow square pyramid, hollow hexagon) and electronic device (RC-LED circuit) are fabricated. The deposition of molten metal on polymer substrate leads to good bonding of metal on polymer due to remelting of the polymer surface. The working of the printed, electronic device is tested and found satisfactory. The testing is conducted by checking the electrical connectivity along the track and the functionality of the electronic device by measuring the output signal waveform. In the future, the combined metal and polymer AM system can be combined with a pick-and-place mechanism that can help achieve a rapid AM of functional 3D electronic devices
Efficient and Convergent Algorithms for High-Fidelity Hyperspectral Image Fusion
No full textHyperspectral (HS) imaging refers to acquiring images with hundreds of bands corresponding to different wavelengths of light. HS imaging has a wide range of applications such as remote sensing, industrial inspection, environmental monitoring, etc. A fundamental consideration with multiband sensors is that the amount of incident energy is limited and this creates an intrinsic tradeoff between spatial resolution and the number of bands---current optical sensors can either generate images with high resolution but a small number of bands or images with a large number of bands but reduced resolution. For example, HS images have hundreds of bands but low spatial resolution, whereas the opposite is true for multispectral (MS) images. An extreme case is a panchromatic (PAN) image with very high spatial resolution but just a single band. Image fusion refers to techniques where multiband images with high spatial resolution are synthetically generated using image processing algorithms. It includes pansharpening (MS+PAN), hyperspectral sharpening (HS+PAN), and HS-MS fusion (HS+MS). Reconstructing a fused image from the observed images is ill-posed and needs regularization. Diverse regularization methods have been proposed over the years for general imaging problems, many of which perform very well for fusion. This includes vector total variation, sparsity and dictionary-based penalties, generalized Gaussian- and GMM-based priors, etc. This thesis proposes novel regularization models and algorithms that can outperform state-of-the-art image fusion techniques. We can broadly group these into two classes---explicit and implicit regularization.
Explicit regularization refers to the design of hand-crafted penalty functions that impose desirable properties (e.g., smoothness) on the reconstruction; this is used along with the observed data for fusion. We propose a convex regularizer that is motivated by nonlocal patch-based methods for image restoration. Our regularizer accounts for long-distance correlations in hyperspectral images, considers patch variation for capturing texture information, and uses the higher resolution image for guiding the fusion process. Unlike local pixel-based methods, where variations along just horizontal and vertical directions are penalized, we use a wider search window in terms of nonlocality and directionality. This is shown to yield state-of-the-art results. The catch is that the resulting optimization problem is non-differentiable and we cannot use simple gradient-based algorithms. However, we show that by expressing patch variation as filtering operations and judiciously splitting the original variables and introducing latent variables, we develop a provably convergent iterative algorithm, where the subproblems can be solved efficiently using FFT-based convolution and soft-thresholding.
In the implicit approach, we rely on a recent paradigm known as plug-and-play (PnP) regularization, where powerful off-the-shelf denoisers are used for regularization purposes. While this has been shown to give state-of-the-art results for general restoration tasks, it has not so much been explored for fusion. In fact, we faced few technical challenges in applying PnP for hyperspectral fusion. Firstly, existing denoisers are slow when applied to multiband images and we need to apply such denoisers several times with the PnP framework. Secondly, convergence is generally not guaranteed for PnP regularization since the mechanism is ad-hoc. Along with efficiency and good denoising performance, we need to come up with a denoiser with specific properties that can guarantee convergence. We proposed a couple of approaches to solve this problem. In the first approach, we have developed a high-dimensional kernel denoiser with low cost yet good denoising performance, which can guarantee PnP convergence. The overall algorithm is fast and competitive with state-of-the-art methods. In the second approach, we leverage the power of deep learning to develop a trained patch denoiser which has a couple of advantages over conventional end-to-end learning:
(1) Unlike end-to-end networks which require excessive ground-truth data for training, we can be trained the denoiser from patches extracted from the observed images. For example, in HS+MS fusion, the MS image captures the same scene and has the same spatial resolution as the target image. We train the denoiser by sampling clean patches from the MS image and corrupting them with noise.
(2) Compared to end-to-end learning, where the training is done with a fixed forward model, our method can be deployed for different forward models. This is possible thanks to the decoupling of the inversion (of the forward model) and denoising steps in PnP.
We use the trained denoiser for PnP regularization and establish convergence of the PnP iterations under a technical assumption that we verify numerically. As far as the reconstruction quality is concerned, our method outperforms state-of-the-art variational and deep-learning fusion techniques
Investigating the role of AMPK in mammary gland alveologenesis and lactation
No full textAMP-activated kinase (AMPK) is an energy sensor that regulates cell growth and metabolism. Reports from our laboratory and others have shown the context-specific role of AMPK signaling in breast cancer. However, its role in normal mammary gland growth and function is unclear. Here, we showed that AMPK expression and activity within murine mammary epithelia increased from puberty to pregnancy, reaching its highest levels during lactation, and then declined post-lactation. Further, induction of prolactin (PRL) signaling increased AMPK expression and activity in ex vivo organotypic cultures of mammary epithelial cells (MECs), whereas PRL failed to do so in 2D monolayer culture of MECs. To understand the role of AMPK in mammary gland morphogenesis in vivo, we generated mice with conditional knockout of the catalytic AMPK isoforms 1 and 2 (AMPK⍺1,⍺2 homo cDKO) in mammary gland. Whole mount analysis of AMPK⍺1,⍺2 homo cDKO mammary glands demonstrated precocious alveolar development with increased epithelial content due to enhanced proliferation and altered differentiation. This was corroborated by ex vivo organotypic cultures wherein pharmacological inhibition of AMPK in primary MECs led to the formation of bigger acini with a significantly increased number of cells AMPK⍺1,⍺2 homo cDKO mice also showed increased beta-casein expression with significantly increased pups’ weight when compared with wild-type control mice. Interestingly, AMPK⍺1,⍺2 homo cDKO epithelia showed increased phosphorylated STAT5 which is known to drive alveologenesis downstream of PRL signaling, suggesting a negative correlation between the two pathways. Interestingly, Akt inhibition led to reversal of phenotype in AMPK⍺1,⍺2 homo cDKO MECs cultured in 3D LrECM, demonstrating a negative cross talk between AMPK and Akt in maintaining cellular homeostasis during alveolar morphogenesis. Our study thus identifies a novel interplay between AMPK and Akt that determines mammary alveologenesis and differentiation through PRL-JAK2-STAT5 signaling