HYDERABAD EECS THEORY SEMINARS (HyTS)

• Venue: TBD, IIIT Hyderabad

Title: Introductory Tutorial on Quantum Noise Models

Abstract: Quantum noise or decoherence is what makes quantum computing and processing of quantum information very difficult. Hence noise and error correction are active areas of research, in both, theoretical, as well as, applied settings. Qubits interact and entangle with the environment, and this interaction is modelled unitarily. This is what is known as a noisy quantum channel. In this tutorial, I will start with the basics of quantum operations representation and discuss various aspects of quantum noise models. The goal is to shed light on some of the “famous” models: bit-flip, phase-flip, depolarisation, amplitude damping. With this understanding one can then independently pursue other channels of current research interest.

Bio: D. B. Dhokarh was until this month a member of the cybersecurity centre in the CS department at IIT Hyd. His previous stints include pursuing research in computational genomics at Mayo Clinic (Minnesota); and quantum optics at UW-Madison. He is interested in quantum computing and information theory; software development; and pedagogy. D. B. Dhokarh has a PhD in theoretical physics from UW Madison. He also worked for several years as a software engineer for Mentor and Cadence, after completing his B.Tech. in EE from IIT Delhi. Besides these professional interests, he is also a very serious student of tabla.

Title: Extreme Points of the \((0,\delta)\)-LDP Polytope with Small Input Size and Arbitrary Output Sizes

Abstract: The structure of locally differentially private (LDP) mechanisms can be understood through the geometry of the corresponding privacy polytope. While the extreme points of the \( (\epsilon,0)\)-LDP polytope are well characterized (Kairouz et al., 2014; Holohan et al., 2017; Pensia et al., 2017), comparatively little is known for the \((\epsilon,\delta)\)-LDP polytope with \(\delta>0\). Recent work (Elangovan and Jog, 2024) has shown that even in the special case \(\epsilon=0\), the \((0,\delta)\)-LDP privacy polytope exhibits fundamentally different behaviour. In this work, we provide complete characterizations of the extreme points for the low-input-alphabet regime \(k=2\) and \(k=3\) and with arbitrary output alphabet size \(m\). We also identify new extreme mechanisms for larger input alphabet sizes \(k\), of the star configuration type, as introduced by Elangovan and Jog (2024).
This is joint work with Supriya Rawat, Gowtham R Kurri and Anand Sarwate.

Bio: Myna Vajha is an Assistant Professor at the EE department, IIT Hyderabad. She received BTech. (2011) and M.Tech. (2013) degrees from IIT Kharagpur and University of Southern California respectively, and the Ph.D. degree from IISc Bangalore in 2020. Post PhD, Myna worked on physical layer algorithms for WiFi at Qualcomm Wireless R&D, Bangalore. Her current research interests are in the areas of coding theory with applications to streaming, privacy, storage, quantum error correction; differential privacy and design of wireless systems.

Title: Fast Fourier Transforms for structured signals: Generalizations of the traditional radix-2 FFT

Abstract: Given an \(N\)-length complex vector whose frequency support \(J\) is known a priori, how fast can we compute its DFT? A full FFT takes \(O(NlogN)\) arithmetic operations; standard techniques exploiting the support cost \(O(|J|^2)\). Can the structure of \(J\) be used to do better? In this talk, I describe a generalization of the radix-2 FFT that achieves \(O(|J|\log^2|J|)\) complexity whenever \(J\) is additively structured - that is, whenever its sumset satisfies \(|J+J|\le c|J|\) for an absolute constant \(c\). The algorithm relies on structured non-uniform downsampling patterns in the time domain, derived from connections between tiling sets, convolution idempotents, and Freiman's theorem.

Bio: Aditya Siripuram is a faculty member with the Department of Electrical Engineering (affiliated with the Department of Artificial Intelligence) at the Indian Institute of Technology Hyderabad. He is interested in the theory and analysis of algorithms for signal processing and machine learning.

Title: Function-Correcting Codes: Partitions and Multi-Level Protection

Abstract:In many communication and storage systems, the receiver is not interested in recovering the entire message, but only in computing a specific function of it. To address this, a new class of codes, called function-correcting codes (FCCs), has been developed. FCCs, introduced by Lenz et al., exploit this requirement by imposing distance constraints only between codewords whose function values differ, thereby reducing redundancy compared to classical error-correcting codes. These codes are especially useful when the message is large but the function's output is small.

In this talk, we present three extensions of the FCC framework. First, we introduce function-correcting codes with data protection, where the encoding provides one level of protection for the data and a stronger level for the function value. Second, we introduce function-correcting partition codes (FCPCs), defined directly on a partition of the message space rather than on a specific function, which allows a single code to protect multiple functions simultaneously through the join of their domain partitions. Third, we introduce generalized FCPCs, which simultaneously protect multiple partitions of the message space, each with its own distance requirement. This framework unifies and subsumes the previous two as special cases. For each setting, we discuss the corresponding construction procedure, derive upper and lower bounds on the optimal redundancy, and illustrate the results through examples.

Speaker Bios: Charul Rajput is currently a Post-Doctoral Fellow with the Signal Processing and Communication Research Centre (SPCRC), International Institute of Information Technology, Hyderabad, India. Previously, she was with the Department of Mathematics and Systems Analysis, Aalto University, Finland, from April 2024 to January 2026. She was a Post-Doctoral Fellow with the Department of Electrical Communication Engineering, Indian Institute of Science, India, from March 2022 to April 2024, where she received the C. V. Raman Post-Doctoral Fellowship. She received her Ph.D. degree from the Indian Institute of Technology Roorkee, India, in 2022. Her primary research interests include algebraic coding theory, function-correcting codes, coded caching, and federated learning.

Mahak received her B.Sc. and M.Sc. degrees in Mathematics from the University of Delhi in 2016 and 2018, respectively, and her Ph.D. degree in Mathematics from the Indian Institute of Technology Roorkee in 2025. Her doctoral research, conducted under the supervision of Prof. Maheshanand Bhanitwal, focused on codes in projective spaces. Since September 2025, she has been working as a postdoctoral researcher under Prof. Lalitha at the International Institute of Information Technology Hyderabad.

Title: Towards Achieving Linkage-Robust Anonymization via \((\ell,\delta)\)-Diversity

Abstract: In this talk, we shall revisit the basic question of dataset anonymization, by asking: How much does anonymity degrade upon the linkage (or composition) of datasets? We shall confine our interest to the popular anonymity notion that is “\(\ell\)-diversity,” and argue that this notion degrades rather quickly upon linkage, in the worst case. This will then motivate our (natural) definition of “\((\ell,\delta)\)-diversity,” which we then show degrades gracefully and can be achieved quite simply for i.i.d. data samples, with the assistance of a central anonymizer. We shall also discuss an efficient, “optimal” generalization algorithm for achieving \((\ell,\delta)\)-diversity when the equivalence classes obtained by binning quasi-identifiers are restricted to be contiguous. This more forgiving notion of anonymity, we hence believe, is somewhat better suited as a privacy measure, in contrast to standard, hard diversity requirements.

The talk is based on joint work with Anshoo Tandon (Centre of Data for Public Good, IISc).

Bio: Arvind is an Assistant Professor at the Department of Electrical Engineering, IIT Madras. He received the B.E. (Hons.) degree in Electronics and Communication Engineering from BITS Pilani University, India, in 2018, and the Ph.D. degree in Electrical Communication Engineering from IISc, Bengaluru, in 2023, specializing in error-control coding. He then spent a couple of years as a Research Fellow at the Centre of Data for Public Good (CDPG), IISc, where he worked on differential privacy.

Arvind was a recipient of the 2023 IEEE Jack Keil Wolf ISIT Student Paper Award. His papers have also won paper awards at the ACM IKDD CODS, NCC, and the IEEE SPCOM. Arvind's Ph.D. thesis was awarded the Seshagiri Kaikini Medal for the best Ph.D. thesis from the ECE Department at IISc, for the year 2023-24.

Title: Quantum LDPC Codes

Abstract: Quantum computing has a potential advantage over classical computing in terms of exponential speedup, owing to the principles of superposition and entanglement. However, qubits and quantum operations are inherently faulty and quantum error correction plays a central role in enabling fault tolerant quantum computing. The fault models vary based on technologies and the current quantum technologies available are based on superconducting qubits, ion traps, neutral atoms and photonic/bosonic systems. The area of designing codes and decoders for various fault models and also to implement gates and circuits in a fault tolerant way is central to enabling quantum computing. In this tutorial-style talk, we will cover the basics of quantum error-correcting codes (QECC) starting with the 9-qubit Shor code. We will develop the general framework of stabilizer codes and CSS (Calderbank-Shor-Steane) codes, touching upon examples of Steane code and toric code. We will study the logical operators of the stabilizer and CSS codes. Then, we will discuss two important classes of codes known as hypergraph product and lifted product codes, both of which have asymptotically good parameters. We will conclude with a discussion on open problems in the area.

Bio: Lalitha Vadlamani received her B.E. degree in Electronics and Communication Engineering from the Osmania University, Hyderabad, in 2003 and her M.E.and Ph.D. degrees in Electrical and Communication Engineering from the Indian Institute of Science (IISc), Bangalore,in 2005 and 2015 respectively. From May 2015, she has been at IIIT Hyderabad, where she is affiliated to Signal Processing and Communications Research Center and she is currently an Associate Professor. Her research interests include coding for distributed storage and computing, quantum error correcting codes, index coding, polar codes and Reed-Muller codes. She was a recipient of Prof. I.S.N. Murthy medal from IISc, 2005 and the TCS Research Scholarship for the year 2011. She was a visiting Scientist at the Simons Institute of the Theory of Computing, Berkeley and also a visiting professor at the Indian Institute of Science, Bangalore, in 2024. She is currently serving as the Newsletter Editor of the IEEE Information Theory Society.

Title: Some advances in complexity of constant depth quantum circuits

Abstract: \(\mathsf{AC}^0\) is the class of constant-depth, polynomial-size circuits consisting of NOT gates and unbounded fan-in AND and OR gates. A fundamental result in theoretical computer science is that the PARITY function on \(n\) bits cannot be computed by an \(\mathsf{AC}^0\) circuit (FSS’84, Ajtai’83, Yao’85, Hastad ’86). The study of \(\mathsf{AC}^0\) and related complexity classes has led to major advancements in several other areas, such as learning theory, cryptography and pseudorandomness. Thus, it is natural to study the quantum analogue of \(\mathsf{AC}^0\). The first challenge one faces in the quantum world is how to formally define such a class. In the classical setup, \(\mathsf{AC}^0\) circuits implicitly allow wires to branch (fan-out). However, allowing analogous quantum circuits to have a dedicated FANOUT gate which copies basis states to multiple target qubits gives them unexpected power. Specifically, a quantum circuit with FANOUT can compute PARITY in constant depth. This dichotomy leads to two distinct quantum circuit classes: \(\mathsf{QAC}^0\) (without FANOUT) and \(\mathsf{QAC}^0_f\) (with FANOUT).
This talk is an introduction to these two (and related) classes. In the first part of the talk, we will understand the power of the FANOUT gate. The FANOUT gate enables a form of parallelization in quantum circuits that provides them with surprising computational abilities. The second part of the talk will compare the power of \(\mathsf{QAC}^0\) versus \(\mathsf{QAC}^0_f\), and we will explore several recent results in this direction.

Bio:Manaswi Paraashar is a faculty member in the Department of Computer Science and Engineering, IIT Hyderabad. Prior to joining IITH, he was postdoc at the University of Copenhagen and Aarhus University, Denmark. He completed my PhD at Indian Statistical Institute, India.

Title: Robust and Numerically Stable Analog Coded Computing for Distributed Computation

Abstract: Distributed coded computing provides an effective framework for mitigating stragglers and preserving privacy in large-scale distributed systems. Along this line, analog coded computing has received significant attention due to its suitability for real-time distributed computation while providing resilience against stragglers and privacy guarantees. However, numerical instability arising from floating-point computations and robustness against Byzantine workers remain as major challenges. This talk presents recent developments in analog coded computing frameworks addressing these 1challenges through techniques from coding theory, approximation theory, and numerical analysis. In particular, we will discuss methods for improving numerical stability, providing robustness against Byzantine workers, and studying the tradeoff between privacy and robustness in distributed analog computation. We will then extend the discussion to approximation-based coded computing frameworks based on Berrut interpolation for computing non-polynomial functions arising in machine learning applications, together with recent spline-based coded computing methods that achieve improved approximation accuracy.

Bio: Rimpi Borah is currently a Ph.D. scholar in the Department of Electrical Engineering at the Indian Institute of Technology Delhi, under the supervision of Dr. Harshan Jagadeesh. She received her M.Tech degree from the Indian Institute of Information Technology, Guwahati, with specialization in Communication and Signal Processing in 2022, and her B.Tech degree from Dibrugarh University, Assam, in 2020. Her research interests include coding for computation, particularly analog coded computing, privacy and security in coded computing, and information theory. Her recent work focuses on developing robust coded computing frameworks resilient to stragglers and Byzantine workers.

Title: Sun-Jafar-type weakly private information retrieval

Abstract: Consider a client who wants to retrieve a desired file from a distributed database without revealing the identity of the desired file. This problem is known as the private information retrieval (PIR) problem and is addressed using a PIR scheme that guarantees the retrieval of the client's demand with absolute privacy. In designing PIR schemes, one aims to maximize the rate, defined for a given scheme as the ratio of the desired file size to the total number of downloaded bits. In 2017, Sun and Jafar presented a capacity-achieving PIR scheme (that is, a scheme with the highest possible rate) for the replication setting without server collusion. This was further extended to other settings, including that of \(T\)-colluding servers and MDS-coded storage. However, perfect privacy is costly; consequently, a weakly PIR (WPIR) scheme is used to relax the privacy constraint in a controlled and quantifiable manner, with a corresponding decrease in rate. Thus, WPIR captures a rate-privacy trade-off. In this talk, we will focus on the PIR and WPIR settings, privacy metrics, and our Sun-Jafar-type WPIR scheme. This scheme is a natural extension of the Sun-Jafar-type PIR scheme and its variants, and this framework helps us describe a wide new class of WPIR schemes. Then, we will provide an explicit way to calculate the rate and privacy metrics, present our trade-off results, and describe their comparative advantage over existing results. Finally, we will discuss converse results and briefly mention our ongoing research work.

Bio: Chandan Anand is a PhD scholar at the Signal Processing and Communication Research Center (SPCRC), IIIT Hyderabad, under the supervision of Dr. Prasad Krishnan. He completed his B.Tech. from MMMUT, Gorakhpur. His research interests include information-theoretic privacy, private information retrieval, information theory, and coding theory.