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
Phonon anomalies in pyrochlores and 2D materials under extreme conditions: Insights from Raman spectroscopic studies
No full textIn solids, electrons do not behave as separate particles but instead interact intensely
to form quasiparticles, which couple with other quasiparticles such as phonon,
plasmon, and exciton to generate emergent quantum phenomena. The emergent
quantum phases in materials arise from the competition and cooperation of various
interactions. Competition and cooperation describe a situation where two entities produce
a result that cannot be achieved by one alone. Raman spectroscopy is a powerful tool to
probe various interactions in solids across the quantum phase transitions. Using Raman
studies, this thesis investigates emergent phenomena in novel quantum materials which
include pyrochlore and two-dimensional materials. n this thesis, we have carried out Raman studies of the following samples: (Eu1−xBix)2Ir2O7,
(Eu1−xBix)2Sn2O7, graphene, and tungsten disulphide (WS2) under extreme conditions of pressure and temperature. Using in-situ Raman measurements, we were
able to identify phonon anomalies in the pyrochlore oxides under the perturbation of
pressure and temperature. The X-ray studies showed the absence of structural change,
conforming that these anomalies were caused by phonon renormalization due to various interactions. Low-temperature Raman measurements on (Eu1−xBix)2Ir2O7 were
performed to investigate electron-phonon coupling and spin-phonon coupling across the
metal-insulator transition. We have also shown that electron-phonon coupling plays
an important role in the incoherency of the metallic phase in (Eu1−xBix)2Ir2O7. High pressure Raman measurements were done on single layers of graphene and WS2 to
detect structural change
Inducing Constraints in Paraphrase Generation and Consistency in Paraphrase Detection
No full textDeep learning models typically require a large volume of data. Manual curation of datasets is time-consuming and limited by imagination. As a result, natural language generation (NLG) has been employed to automate the process. However, in their vanilla formulation, NLG model are prone to producing degenerate, uninteresting, and often hallucinated outputs. Constrained generation aims to overcome these shortcomings by providing additional information to the generation process. Training data thus generated can help improve the robustness of deep learning models. Therefore, the central research question of the thesis is:
“How can we constrain generation models, especially in NLP, to produce meaningful
outputs and utilize them for building better classification models?”
To demonstrate how generation models can be constrained, we present two approaches for paraphrase generation. Paraphrase generation involves the generation of text that conveys the same meaning as a reference text. We propose two strategies for paraphrase generation:
(1) DiPS (Diversity in Paraphrases using Submodularity): The first approach deals with constraining paraphrase generation to ensure diversity, i.e., ensuring that generated text(s) are sufficiently different from each other. We propose a decoding algorithm for obtaining diverse texts. We provide a novel formulation of the problem in terms of monotone submodular function maximization, specifically targeted toward the task of paraphrase generation. We demonstrate the effectiveness of our method for data augmentation on multiple tasks such as intent classification and paraphrase recognition.
(2) SGCP (Syntax Guided Controlled Paraphraser): The second approach deals with constraining paraphrase generation to ensure syntacticality, i.e., ensuring that the generated text is syntactically coherent with an exemplar sentence. We propose Syntax Guided Controlled Paraphraser (SGCP), an end-to-end framework for syntactic paraphrase generation without compromising relevance (fidelity). Through a battery of automated metrics and comprehensive human evaluation, we verify that this approach does better than prior works that utilize only limited syntactic information in the parse tree.
The second part (meaningful outputs) of the research question pertains to ensuring that the generated output is meaningful. Towards this, we present an approach for paraphrase detection to ascertain that the generated output is semantically coherent with the reference text. Paraphrase Detection is the task of detecting whether or not the two input natural language statements are paraphrases of each other. Fine-tuning pre-trained models such as BERT and RoBERTa on paraphrastic datasets have become the go-to approaches for such tasks. However, tasks like paraphrase detection are symmetric - they require the output to be invariant of the order of the inputs. In the traditional fine-tuned approach for paraphrase classification, inconsistency is often observed in the predicted labels or confidence scores based on the order of the inputs. We validate this shortcoming and apply a consistency loss function to alleviate inconsistency in symmetric classification. Our results show an improved consistency in predictions for three paraphrase detection datasets without a significant drop in the accuracy scores.
While these works address the research question via paraphrase generation and detection, the approaches presented here apply broadly to NLP-based deep learning models that require imposing constraints and ensuring consistency. The work on paraphrase generation can be extended to impose new kinds of constraints (for example, sentiment coherence) on generation, while paraphrase detection can be applied to ensure consistency in other symmetric classification tasks (for example, sarcasm interpretation) that use deep learning models
Design and Performance Investigation of Supersonic Air-Intakes
No full textSupersonic air-intake is an essential component of a ramjet engine. Typically, a ramjet powered missile that operate over a range of Mach numbers and maneuverability conditions uses a fixed geometry intake to ensure cost-efficiency. Such an intake system at off-design conditions would experience adverse flow conditions related to shock-shock interactions, shock wave/boundary-layer interactions (SWBLIs), and many other flow phenomena affecting its performance parameters. Moreover, the flow is likely to “unstart because of SWBLI”. Therefore, fixed geometry intakes must be designed to maintain desirable performance characteristics in the complete operating regime. The thesis focuses on the preliminary design and performance evaluation of a supersonic intake system for a “four-intake symmetric cross or X configuration missile”, to be operating in the freestream Mach number range of about 1.8-2.5, an angle-of-attack range of 0° ≤ AoA ≤ 10° and over a low altitude condition of 5-10 Km, corresponding to a unit Reynolds number of approximately 1.4 -4.0×107 m-1. More specifically, the objective is to provide an in-depth understanding of the effects of internal geometric factors, upstream flow and boundary-layer bleed system parameters that should be considered for the design of a ramjet air-intake. The performance has been evaluated in terms of total pressure recovery (TPR), captured and bleed mass flow ratio (MFR, BMFR), sustainable backpressure ratio (or peak TPR) and distortion index (DI). The highlights of the results could be summarized as follows:
In the first phase of the study, a theoretical one-dimensional (1D) optimization tool has been developed for the design of rectangular mixed-compression intakes. The tool makes use of various oblique and normal shock relations, Oswatitsch criterion for planar shocks and available experimental correlations to fix the geometric dimensions of different parts, and aims at maximizing the total pressure recovery. Secondly, a computational fluid dynamic (CFD) model has been proposed for the prediction of intake flow and performance characteristics. The numerical methodology has been verified using established techniques for scientific computing, that involves grid independence study and the estimation of grid convergence index (GCI). Moreover, the model is systematically validated against three experimental test cases/data available in the literature, where the possibility of “unstart because of SWBLI” arises.
In the third phase of the work, the effects of several internal geometric factors such as the multi-ramp compression structure, internal contraction ratio, subsonic diffuser internal ducting as well as the sidewall configurations, and several upstream flow conditions have been studied independently to clarify their impact on the compression performance. The analysis shows that, at Mach 2.2, the mixed-compression intake with two-ramp and three-ramp compression leads to a critical TPR of about 0.83 and 0.84, respectively, with a corresponding supercritical MFR of 0.98 and 0.967, respectively. Though the three-ramp configuration shows a higher TPR at the design condition, it leads to a rapid drop in the critical MFR and TPR at the off-design operations compared to that of the two-ramp intake. This is because of the higher losses across the bow shock that forms because of “unstart because of SWBLI” at low off-design Mach numbers and angles-of-attack conditions, as well as the losses across the separation region at the cowl at high off-design Mach numbers. Hence, a supersonic compression structure with lower degree of external compression is useful to achieve enhanced off-design performance. Next, a passive bleed system has been designed to improve the overall intake performance by diverting a fraction of the low energy boundary-layer (BL) at the throat. The analysis shows that with the use of bleed, the exit flowfield uniformity as well as the critical and peak TPR improves considerably at all operating conditions. This is because of the reduction in the flow separation as well as the total pressure loss across the SWBLIs (called as boundary-layer factor), and the stabilization of terminal shock at the bleed entrance (called as stabilization factor). In the last phase of the study, the optimized isolated intake is integrated to an ogive nosed forebody as a four-intake symmetric cross or X configuration missile, and a combined external-internal flow analysis has been carried out to evaluate the performance of the installed intakes. The data confirms that the considered forebody length ahead of the intakes, as well as the diverter height and its position downstream of the ramp lip are sufficient to maintain a good performance over the desired off-design operating conditions without significant flow separation or spillage ahead of the intakes.JATP/P-VIII/P-2018/159-DRD
Platform Technologies for In vitro Point-of- Care Diagnostics
No full textIn vitro diagnostic (IVD) tests are critical for making informed clinical decisions. These
tests are performed outside the body using specimens such as blood, sputum, saliva, and
urine. IVD tests are usually performed at centralized labs with complex and expensive
instrumentation, and associated protocols. There are only a few commercially available
point-of-care (POC) devices for performing IVD Tests. This thesis presents platform
technologies for POC testing of three categories of IVDs: Clinical Biochemistry,
Immunodiagnostics, and Molecular diagnostics. Additionally, a Biosafety Level 2+ (BSL-
2+) Mobile Laboratory was designed & developed for testing of infectious diseases at the
point-of-need.
Clinical biochemistry tests are quantitative tests that measure the concentration of
biochemicals in body fluids. Most clinical biochemistry tests involve the testing of serum.
A POC serum separation device from micro-volumes of blood using an air-displacement
pipette and a chemical-preloaded disposable pipette-tip cartridge is introduced in Chapter-
2. To address the issue of reagent stabilization and room-temperature storage, trehalosebased
enzyme reagent matrix (TERM) cartridge with dried reagents was developed. The
cartridge was tested for triglycerides measurement in blood and found to offer good shelflife.
In addition, the development of reagent kits for conducting sickle-cell disease
confirmatory test at point-of-need was reported. Finally, as an alternative to conventional
biochemical analyzers, a portable multi-spectral absorbance reader capable of detecting
the results at POC of a wide range of clinical biochemical assays was developed and
validated.
Immunodiagnostics includes tests that can detect and quantify antigens and
antibodies in a sample. Chapter-3 of the thesis proposes the development of platform
technologies in this category based on introducing innovative (i) protocols and (ii)
processes: (i) Counting target CD4+ T-Cells in blood was demonstrated using anti-CD4
antibodies conjugated to superparamagnetic iron oxide nanoparticles (SPIONs) and a
commercially available haematology analyser. (ii) A bare fibre Bragg grating (bFBG)
sensor-based technique is developed for pathogen detection, and demonstrated its ability in
detecting E. coli in water by coating anti-E. Coli antibodies onto the bFBG sensor.
Molecular diagnostics comprises of techniques that can detect the presence or
modification of nucleic acid (DNA/RNA). These tests usually require sophisticated and
iv
expensive instruments. To address this issue, Chapter-4 reports the development of
Portable PCR System: i) Isothermal Amplification Device (LAMP assays), ii) Handheld
Thermal Cycler (tested for HCV & Dengue), and iii) Fluorescence Reader (tested for
CoViD-19). In Chapter-5, a mobile laboratory for infectious diseases testing and reporting
(MITR Lab) was conceptualized, designed, fabricated, and found to offer 100% match
with static lab results in testing CoViD-19. The MITR lab is the first ICMR approved
Mobile BSL2+ lab for COVID-19 testing and aided in conducting few lakh tests during the
past one year.
All the technologies developed as part of the thesis are validated with clinical
samples and showed sensitivity and specificity above 90%. These innovative, affordable
platform technologies are expected to create huge social impact by providing timely
diagnoses at the point-of-need
Development and Studies of Open-Pore Metal Foam for Diverse Applications
No full textThis thesis presents a method for producing open-cell metal foams by vibration-assisted
casting (VAC) and its studies for diverse engineering applications. Metal foams have a wide
range of applications due to their several beneficial properties such as high specific strength,
low relative density, high effective thermal conductivity, and permeable porous structure with
a high surface area to volume ratio. However, their applications are currently limited due
to the complexity and high manufacturing cost. Traditional techniques involve a complex
production setup that includes an inert environment and high-pressure compressed gases to
accomplish molten metal infiltration into the preform to process metal foam. As a result, we
have developed a method to simplify and minimize manufacturing costs by eliminating the
usage of inert environments and high pressure compressed gases. The VAC method carefully
controls the pouring temperature and cooling rate with the additional use of a low-frequency
vibrator for infiltration of molten metal into the preform. The VAC method facilitates
the design and manufactures several metal foams with high specific surface area with pore
diameters ranging from 100 μm to 10 mm. Additionally, the VAC method produces various
types of metal foam, homogeneous, heterogeneous, functionally graded, and composite metal
foams for diverse engineering applications. Furthermore, the VAC method also enables the
production of compact integral foam heat exchangers (HXs) or heat sinks in single-step
casting. Controlling the cooling rate (preheating temperature) allows the metal foam to be
fused bonded to a metal substrate. Thus, it eliminates many intermediate processes that are
unavoidable in conventional metal foam HXs, such as machining, use of epoxy glue, brazing,
press-fitting, or welding, making it easily recyclable. Fusion bonding between the metal
foam and substrate reduces thermal contact resistance (RTCR) and improves the thermal
performance of foam HX. The current technology facilities development of various metal
foam heat exchanger designs such as integral foam heat sink (HS), foam-plate-fin HS, and
foam-pin-fin HS. The following major findings are obtained from the contribution chapters
that are enumerated below:
(i) The effect of foam height, pore diameter, number of foam fins, orientation, and bonding
methods on thermal performance are examined. The heat transfer rate per unit mass
of foam-fin heat sink is approximately double that of commercially available fin heat
sinks.
(ii) The thermal contact resistance (RTCR) of fused-bonded foam heat sinks is approximately
19 times lower than an epoxy-glued foam heat sink, resulting in ∼30% higher
Nusselt number (Nu) of the fused-bonded foam-fin heat sink than the epoxy-glued
foam-fin heat sink.
(iii) Two types of foam heat exchangers (HXs): single-sided fused bonded foam tube heat
exchanger (SFMF) and double-sided fused bonded foam tube heat exchanger (DFMF)
with spherical and ovoid shape pores are developed.
(iv) Increment in foam height has a significant influence in reducing the thermal resistance
(Rth). For the same pumping power (W), the DFMF heat exchanger provides up to
∼25% lower Rth than the fin heat exchanger (Fin HX) and SFMF heat exchanger.
(v) Composite metal foam (CMF) is an efficient energy absorber as compared to foam since
it provides a larger stress plateau region while undergoing deformation. Moreover, the
damping capacity of CMF is ∼7.5 times higher than aluminium foam and synthetic
elastomer, and ∼3.5 folds higher than solid metal due to the high loss modulus of
CMF.
This study concludes that the composite metal foams are efficient energy absorbers and
vibration dampers. Furthermore, an integral foam-fin heat sink with minimal thermal contact
resistance is an excellent solution for thermal management in high-powered electronics with
the benefits of being lightweight, compact, and easily recyclable
Mitigating Domain Shift via Self-training in Single and Multi-target Unsupervised Domain Adaptation
No full textThough deep learning has achieved significant successes in many computer vision tasks, the state-of-the-art approaches rely on the availability of a large amount of labeled data for supervision, collection of which is expensive and time-consuming. Moreover, the performance of these models suffer when there is a mismatch between training and test data distributions. Motivated by this, we design Unsupervised Domain Adaptation (UDA) algorithms that address the distribution shifts without requiring label information in order to adapt to the new environment. In the first part of this work, we show that the class-aware frequency transformation obtained via self-training helps to reduce the style bias in the source dataset, thereby improving the target adaptation performance. Further, we address a more challenging and practical setting of source-free multi-target domain adaptation, where there is only one source, but multiple unlabeled target domains, and the source labels are assumed unavailable during target adaptation.
We explore the utility of frequency transformation for reducing the style bias between the source and target domains (e.g., the bias between the synthetic images and the natural images, respectively). The performance of the existing UDA methods degrades when the domain gap between source and target distributions is significant. In order to bring these domains closer, we propose ‘Class Aware Frequency Transformation’ (CAFT), which utilizes pseudo label-based class aware low-frequency swapping of image-magnitude spectrum to reduce the domain gap and thereby improve the performance. When compared with the state-of-the-art generative methods, our proposed approach is computationally efficient and can easily be plugged into an existing UDA algorithm to improve its performance. Additionally, towards mitigating the pseudo-label noise, we introduce a novel approach based on the absolute difference between top-2 class prediction probabilities (ADT2P), which separates target pseudo labels into clean and noisy sets. Our proposed UDA strategy substantially benefits from utilizing these ‘clean samples’ only, thereby resulting in a further improvement in overall performance.
We introduce the novel task of Source-free Single and Multi-target Domain Adaptation and propose a novel framework named Consistency with Nuclear-Norm Maximization and MixUp knowledge distillation (CoNMix) as a solution to this problem. The primary motivation of this work is to address the Single and Multi-target Domain Adaptation in the source-free paradigm, where access to labeled source data is restricted during target adaptation due to various practical privacy-related restrictions on data sharing. Our source-free approach leverages self-training using target pseudo labels to improve the target adaptation performance. We propose consistency between label preserving augmentations and utilize pseudo label refinement methods to reduce noisy pseudo labels. Further, to build one source-free Multi-target Domain Adaptation model using multiple single-target DA models, we use the concept of the MixUp Knowledge Distillation. We also demonstrate that by utilizing the modern Vision Transformers as backbones, we can obtain better feature representations leading to improved domain transferability and class discriminability. This further helps to boost the performance of source-free Single-target and Multi-target Domain Adaptation
Design of Increasing Block Tariff (IBT) for Pricing Domestic Water: A Simulation Approach
No full textIn recent times, urban water utilities have shown increasing preference for increasing block tariff (IBT) for pricing domestic water. Although appealing at the outset, IBT’s can fall short of achieving its goals such as adequate cost recovery and effective cross-subsidization when water utilities arbitrarily design the IBT structure without any rational basis. To solve this problem, this study presents a simulation method to design an IBT for Bangalore, taking into consideration the nature of distribution of monthly consumption of domestic households and the revenue requirements of the water utility. The simulation method aims to find the best set of IBT parameters that simultaneously satisfies goals such as cost recovery, affordability, cross-subsidization and most importantly, fairness of marginal prices in the higher blocks of IBT. The study presents the design of a 3 block IBT for pricing domestic water in Bangalore and finds that an increase of the marginal price in the primary IBT block increases the likelihood of achieving the IBT goals simultaneously, when compared to marginal prices in the higher blocks. At the same time, the block sizes have a negative significant influence on achieving the IBT goals simultaneously. The relative influence of the 3 block IBT parameters on achieving the IBT goals simultaneously is assessed using logistic regression and the IBT parameter solution set is represented using a decision tree. In addition, the study analyses the IBT block switching characteristics of domestic households in Bangalore, during the time periods (2006-2009 & 2016-2018), using a steady state model and also reports the price elasticity of IBT. In contrast to previous studies that have assumed the positive association between household income levels and household consumption, a pre-requisite condition for effective implementation of an IBT, this study clearly demonstrates the positive influence of the household income on the monthly average household consumption by developing water demand models at aggregate municipal ward level and household level. First, the aggregate water demand model is developed for 198 municipal wards of Bangalore, using average monthly household water consumption as dependent variable and a set of demographic variables namely population density, household density etc. and infrastructural covariates such as average built-up area, property density, road density, per capita park area etc. Results suggest that average built up area and per capita park area have a significant positive influence on the average monthly household consumption. Owing to the availability of few extra covariates in the 39 municipal wards of South Bangalore, the study also develops an aggregate water demand model for the 39 municipal wards in South Bangalore. Analysis reveals that average household water consumption is higher in municipal wards, where the fraction of houses with more than 3 rooms is higher. Second, the study develops a household water demand model for Bangalore city, using survey data from 300 households in different regions of Bangalore. Analysis reveals that the household domestic consumption increases with respect to demographic variables such as household income, family size, educational level and infrastructural variables such as number of bedroom, bathrooms and presence of washing machine. The study also finds that the practice of sharing of water meters by multiple households increases the average household water consumption
Systems biology approach to dissect Salmonella-host interactions
No full textSystems understanding of the biological data is required to capture the complex biochemical information when a pathogen interacts with its host. Such interactions are dynamic and complex, especially because various components of the pathogen interact with the host at multiple levels. Understanding this complex data has been made somewhat feasible with the advent of systems biology. System biology involves studying the complex interactions which take place between genes, proteins, and other components in a biological network, compared to traditional biological research, where the focus is only on a small number of components. For instance, a systems biology study of host-pathogen interactions investigates the interaction between the components of two distinct organisms, a pathogen, and its animal host. The only way this is accomplished is through the integration and interpretation of the high-throughput data made available at various levels of detail. However, high-throughput techniques like sequencing and multi-omics not only generate a massive volume of data but also pose challenges to meaningfully extract and interpret the data.
Salmonella sp. is a Gram-negative, intracellular pathogen and causes severe complications contributing significantly to the global burden of foodborne illnesses. Salmonella Typhi is a human-restricted pathogen that causes a systemic disease called Typhoid (enteric fever). On the other hand, Salmonella Typhimurium, a zoonotic pathogen, causes a less severe form of the disease called gastroenteritis in humans but a systemic typhoid-like disease in susceptible mice strains (typhoid model). The genomes of most organisms, including Salmonella, as they exist today are the result of millions of years of evolution - the acquisition or elimination of certain key genetic determinants, genomic rearrangements, and the insertion of novel genes into existing genetic circuitries. Reverse genetics has proven extremely useful in deciphering the function of such key genes and their protein products required during Salmonella pathogenesis.
In the first part of this study, we sought to identify proteins that are exclusive to Salmonella sp. using comparative genome sequence analysis. Comparing the proteome of Salmonella with a non-redundant protein sequence database allowed us to not only detect sequences unique to the organism but also identify its conservation among the Salmonella serovars. We argue that such proteins with a unique sequence architecture might be acquired and evolutionarily retained in the genome of the organism only to serve some essential function. By careful genetic manipulation and phenotypic analysis, we were able to shed functional insights in a novel protein sequence, GtgF and demonstrate its role in S. Typhimurium adaptation in oxidative and genotoxic stress response mechanisms.
Using a mouse typhoid model, the second part of our study assessed and compared the virulence function of GtgF in WT S. Typhimurium 14028s. In a susceptible mouse strain, like BALB/c, WT S. Typhimurium 14028s administered through the orogastric route is known to cause lethal infection. However, we observed that S. Typhimurium ΔgtgF was significantly more lethal (approx. two times) than the WT in an oral typhoid model. A marginally higher bacterial burden in tissues with increased inflammatory markers in the blood, and exacerbated tissue inflammation explained why the mice were more susceptible to infection with ΔgtgF strain. In addition to grvA, we believe gtgF might be an additional anti-virulence genetic determinant encoded by the same phage, Gifsy-2. Although it is unclear why bacterial pathogens possess anti-virulence genes and what evolutionary advantage they provide to their host, it is believed that certain intracellular pathogens evolve toward a less virulent form in order to attain a chronic carrier state, a phenomenon well-established in the intracellular pathogen, Salmonella.
In the third and concluding part of the study, we developed a drug-repurposing strategy to design antibiotic adjuvants that can target the MsgA-like family of proteins in Salmonella. As these proteins are mostly conserved in pathogenic organisms, targeting this family of proteins has the selective advantage of having negligible and unintentional anti-commensal activity. It is well-known that targeting conditionally essential proteins can impede the rate of anti-microbial resistance. While discovering novel drugs or even new classes of antimicrobials might be an extremely time-consuming and resource-intensive affair, orthogonal approaches such as designing antibiotic potentiators can be implemented simultaneously to preserve our existing arsenal of antimicrobials. Through high-throughput virtual screening with an FDA-approved drug subset, we identified potential drugs which can be repurposed as antibiotic adjuvants against the intended targets. The computational findings were validated to determine whether these drugs can indeed potentiate the action of the co-administered antibiotic. Some true validations obtained in this study clearly demonstrate that the proposed methodology holds promise; however, it remains to be seen whether these observations hold true for drug-resistant isolates. Consequently, the mechanism of action of the drugs must also be established. However, as a proposed methodology, these observations has significant ramifications in the antibiotic resistance problem. In summary, the entire workflow presented here is extremely generic and can be adapted for any pathogenic microorganism to tackle the rising concern of anti-microbial resistance
Top-k Spatial Aware Ads
No full textConsider an app on a smartphone which displays local business ads. When a user opens the
app, then k local business ads need to displayed (where k would typically be 3 or 5) such that
the profit made by the app is maximized. The pricing model needs to take into account that
(a) each business is willing to bid a different price, and (b) farther the distance of the user on
whose smartphone the ad is displayed, the lesser is the price paid by to the app.
Motivated by such applications, in this work, we design fast algorithms to retrieve top-k
objects using the provided spatial and non-spatial attributes. We refer them as Top-k Spatial
Aware Ads Queries (SAA). In Top-k-Saa, the query is user location and we return top-k
objects that have the best score. The scoring function is based on the distance between the
object and query point (spatial attribute) and non-spatial attributes. We propose algorithms
that efficiently preprocess the data using appropriate data structures and aid in fast query
processing. A simple O(n log k) algorithm returns the top-k ads based on the scoring function
value. We obtain the following results.
1. Our first algorithm uses O(n log n) space and answers the Top-k-Saa query in O(k log2 n)
time. The fast query time is obtained by leveraging the properties of additively weighted
Voronoi diagram, along with other supporting data structures.
2. Our second algorithm improves upon the first algorithm by improving the query time to
O(k log n) in expectation, while using the same space. This is achieved via an interesting
combination of randomization with a “top-2” structure
Numerical Study of Accelerating Turbulent Boundary Layers
No full textThis thesis focuses on the prediction of favourable pressure gradient turbulent boundary layer flows. Particularly, the focus is on mild favourable pressure gradient turbulent boundary layers and relaminarising turbulent boundary layers (where the favourable pressure gradient is relatively large.) The calculations were performed on ANSYS FLUENT software, and the turbulence models used are the Spalart-Allamaras model and k-ω SST model. Furthermore, an integral method called Green's Lag Entrainment method is also used, which was originally designed for the prediction of turbulent flows. A MATLAB code was developed to simulate the test cases using Green's Lag Entrainment method (1977). A limited set of mild favourable pressure gradient experiments on a low-speed wind tunnel and the measurements were made using Particle Image Velocimetry. It should be noted that in the PIV experiment, the flow acceleration was mild (owing to mild FPG), and hence it stayed turbulent and did not relaminarise. In the present study, three re-laminarisation experiments/DNS computations were considered as test cases. 1)An initial zero-pressure turbulent boundary layer of Re_θ=1120=, subjected to a strong favourable pressure gradient over a region in the wind tunnel relaminarised at x = 73cm from the leading edge of the flat plate. This was originally studied by Patwardhan using experiments, and he also re-created in direct numerical simulations, and the agreement between them was good. We will refer to this case as the Low Re case of Patwardhan (2014). 2) Similarly, the initial zero pressure gradient turbulent boundary layer of Re_θ=1900, subjected to strong favourable pressure gradient different in magnitude (whilst keeping the same non-dimensional free-stream velocity distribution) over a much larger region than the Low Re case of Patwardhan re-laminarises at x = 87cm from the leading edge of the flat plate. Again, this also was originally studied by Patwardhan using experiments and direct numerical simulations, and the agreement between them was good. We will refer to this case as the High Re case of Patwardhan. 3)An experiment published in literature by Bourassa & Thomas (2009) where the initial Reynolds is much higher, perhaps the highest in the whole literature concerning re-laminarising flows of Re_θ=4590. Studies using RANS turbulence models and Green's Lag Entrainment method were performed on all the above cases. The overall conclusion is that while the prediction upstream of the onset of relaminarisation is fairly good. But it is interesting to see that the integral method agrees more closely with the experiments and DNS computations than the RANS turbulence models for all three cases upstream of relaminarisation. After relaminarisation, for all three cases, both the RANS turbulence models and the integral method fared poorly. Since the model didn't fare well downstream of the relaminarisation, another set of FLUENT calculations was devised where the flow was calculated without any turbulence model, i.e., a laminar model is solved. The results from these laminar calculations showed a good agreement in the relaminarisation zone for Patwarhdan's Low and High Re cases. This is consistent with the Ranjan & Narasimha (2018) improved version of the quasi-laminar theory, which was developed to explain the later stages of flow relaminarisation.
References:
[1] Green J. E., Weeks D. J., Brooman J. W. F., “Prediction of Turbulent Boundary Layers and Wakes in Compressible Flow by Lag Entrainment Method, A. R. C., R & M. No. 3791, 1977.
[2] Patwardhan S. S., “Effect of favourable pressure gradients on turbulence in boundary layers”, A Ph.D. Thesis, Indian Institute of Science, Bangalore, 2014.
[3] Bourassa, C. & Thomas, F. O., An experimental investigation of a highly accelerated turbulent boundary layer. J. Fluid Mech. 634, 359–404. 2009.
[4] Rajesh Rajan & Roddam Narasimha, "An assessment of the two-layer quasi-laminar theory of relaminarization through recent high-Re accelerated TBL experiments, arXiv preprint arXiv:1611.09746, 2018