Really Big Questions
As John Archibald Wheeler would have it, these RBQs fuel my scientific quest.
- Why is the Universe computable? - The extraordinary effectiveness of AIT techniques like Solomonoff induction and UAGI like AIXI is based on just one assumption: the Universe is computable by a universal Turing machine. Why so? Is this a self-evident and necessary axiom/assumption as formalizing such a question within language is related to computation? Why does the Church-Turing thesis hold? What does that tell us about the nature of our physical laws, scientific method, or their limits thereof? Why do simple programs give rise to complex behavior? Is comprehension by compression and uncomputability of universal automata two sides of the same coin? Is everything relative to a partial trace or course-graining? Is truth relational? The ubiquity of universality and the computational reducibility self-referential? Does a Markov blanket automatically give rise to a probability distribution of Bayesian beliefs and information geometry? Is a non-anthropocentrism viewpoint possible?
- Is there a computational action? - The action in Lagrangian in physics is fundamental to many core theories and looks very similar to computational and algorithmic complexity - Are they? What roles do the computational resources play? What is the relation between computational, algorithmic complexity and thermodynamics? Can locality and causality be violated?
A dozen favorite problems
Wheeler’s illustrious student, Richard Feynman suggested to have a few favorite problems handy to brood on. Here are mine:
- Is there any structure in the universe, or do structures reflect the incapabilities/symmetries of our laws/instruments? Can all symmetry of the universe be broken, or do the last few symmetries (e.g., CPT symmetry and Lorentz invariance) reflect the limitations of the math we use to do physics?
- Why don’t we see any super Turing processes in Nature, or is the quantum measurement collapse (or any conversion from continuous to discrete) a super Turing process?
- Why Occam’s razor works in general, especially in science.
- How to define art? How do we formalize intrinsic motivation without turning it into a cost function?
- How do we formalize the blurry spectrum of non-living/living and intelligent/dumb? Integrated information theory?
- What is more foundational? Shannon entropy or Kolmogorov entropy? Circuit complexity or algorithmic complexity?
- How bad are the approximate estimates of AIT metrics? How are these estimates so successful (e.g., compression or LLM program synthesis) if it’s semi-computable?
- What properties of quantum circuits make them difficult to simulate classically?
- Can direct computer interfaces (e.g., Neuralink) enable different programming models without discrete languages, rather by more fluid thought?
- Which ontological standing is more sound - the world is at its finest scale, continuous or discrete? Quantum-classical, or NumberTheory. If the world is discrete, we have to give an ontic explanation of the specific value of Plank’s constant, or speed of light.
- Can we do away with all fundamental constants of nature or explain why they have certain values?
- Can there be any experimental evidence of consciousness/qualia as a post-hoc explanatory brain process?
Research Themes
Research themes evolve over time. My older interests are listed with strikethrough. Some of my core interests are:
- Artificial General Intelligence - library learning; universal artificial intelligence; recursive self-improvement; universal constructors; explainable neural networks via code distillation; neuro-evolution; neuro-symbolic;
cellular automata rule learning;algorithmic perturbation analysis;do-calculus; experimental algorithmic information theory;computational complexity phase change behavior; relations between expressibility-reachability-learnability-universality; intrinsic motivation and autotelesis …read more - Quantum Computation - quantum computer architecture; high-level programming and design automation;
quantum Hamiltonian complexity;quantum learning theory; quantum resource theory; quantum complexity geometry; quantum information; quantum reference frames;quantum swarm intelligence;tensor networks;holographic quantum circuit complexity;Feynman checkerboard QFT…read more - Bioinformatics - in silico design space exploration for xenobiology; artificial life; synthetic biology; causal models of gene regulatory network; self-replicating RNA medicines
- Computational Applied Category Theory - computational category theory; universal algebras; categorical intelligence
- Swarm intelligence -
multi-agent collaboration;swarm robotics; emergence; coarse-graining - Others -
fractals;game theory;RtOS;computer vision;digital steganography
Here’s a visual that captures the most important topics:
Research Journey
During high school, my fascination was primarily centered around computer science, which accounted for about three-quarters of my motivation. This field introduced me to topics like Boolean algebra, computer organization, and object-oriented programming. Physics and mathematics also captivated my interest. In physics, wave-particle duality, transistors, and thermodynamics stood out as particularly intriguing, while in mathematics, I enjoyed exploring permutations, combinations, and probability.
In my undergraduate years, I pursued avionics (a portmanteau of aviation electronics), a field closely related to electronics engineering. This journey broadened my understanding of microcontrollers, robotics, computer vision, and swarm intelligence, all while deepening my passion for programming and computer architecture. My curriculum also provided insights into modern physics, relativity, quantum mechanics, and statistical mechanics. Supplementary online and guest lectures introduced me to exciting fields like genetic programming, machine learning, artificial intelligence, and quantum computation. Interestingly, there were areas I struggled to appreciate or wished had been part of my studies. Signal processing, control systems, and functional analysis didn’t resonate with me as much, while topics like automata theory, field theory, and learning theory sparked curiosity but weren’t included in my curriculum.
My professional journey began at ISRO, where I honed skills in low-level assembly programming and VHDL-based FPGA programming. These experiences allowed me to delve into system-level design and fault analysis, providing a complete hands-on understanding—from high-level logic to the timed electrical signals executing that logic. This foundational knowledge later became indispensable in my exploration of quantum computing.
For my master’s, I studied at TU Delft, where I was part of the computer engineering department. The program’s interdisciplinary nature exposed me to electrical engineering, embedded systems, and computer science. My specialization in quantum computation introduced me to quantum information, quantum communication, and quantum computer architecture. These studies enabled me to map my classical full-stack perspective into the realm of quantum computation. My fascination with programming further motivated my thesis, which focused on quantum algorithms. Inspired by genetic programming where physical systems are encoded as genes, I applied this perspective to develop quantum algorithms.
With this perspective, I started my doctoral research. I bore witness to the shift towards variational approaches, NISQ and QML. However, at that stage, I was a staunch symbolist trying to advocate the principled design of quantum algorithms with provable computational resource guarantees.
I discovered algorithmic information theory during my second year. The synergy between experimental algorithmic information theory (EAIT) and quantum computation (QC) quickly became my favorite intellectual perspective. This fusion felt like the culmination of my academic interest right from the start, the perfect balance of computer engineering, computer science, and quantum information that piqued my curiosity. QC enables the exploration of rich, exotic computation, while AIT serves as the gatekeeper of algorithms, resources, and explainability. Both fields have profound implications for artificial intelligence and automation, an area increasingly becoming integral across science. With QC, I ventured into quantum machine learning (QML) and automation in quantum computation (AutoQC), including algorithm design, compilation, and control. Meanwhile, AIT led me to explore universal reinforcement learning, artificial general intelligence (AGI), and the thermodynamics of computation. These intertwined fields form the foundation of my research and professional identity.
Today, I would best describe myself as a quantum computer architect, integrating my computer science and quantum mechanics expertise within a computer engineering framework for full-stack quantum computation.
Mind Map
… of all things I love to keep myself updated about and the associated contributors to those ideas…
The blue boxes and their 1-hop neighbours are my core interests.
I tried finding my core interests in the ‘Domain of Science’ charts. I eventually figured that most of my core interests lie in the field of computer science. The interests within physics, mathematics, and biology reflect the topics in computer science. Here’s the result:
Curated Lists
Favourite Sources
Favourite Researchers
Scientists with similar research interest
Some researchers whom I currently follow closely for my research as they have overlapping interests with me.
- Automation in quantum computing: Florian Marquardt, Leopoldo Sarra, Mario Krenn, Robert Wille
- Program synthesis / symbolic regression / genetic programming: Kevin Ellis, Armando Solar-Lezama, Max Tegmark, Lee Spector
- Algorithmic information theory: Hector Zenil, Cristian S. Calude, Markus Müller, Noson Yanofsky
- Digital physics: Stephen Wolfram
- Artificial general intelligence: Jürgen Schmidhuber, Ben Goertzel, Marcus Hutter
- Quantum information theory: David Wolpert, John Baez, Charles Bennett
Favourite formulae/hypotheses
List of formulae or hypotheses that I find most impactful
- Cantor’s Continuum hypothesis
- Church-Turing-Deutsch principle, and Solovay-Kitaev theorem
- Curry-Howard-Lambek correspondence
- Mass-Energy-Information equivalence
- Choi-Jamiołkowski isomorphism
- Holevo’s bound, Quantum Speed Limits, and Limits to Computation
- Thermodynamic Kolmogorov complexity, and Uncomplexity
- Solomonoff induction, AIXI, and Quantum Knowledge Seeking Agent
- Lorentz invariance, CPT symmetry, Quantum Reference Frames, and Noether’s theorem
- Area law of entanglement
- Fractals: Mandelbrot set, Barsley fern fractal, Glynn Julia set