Home on EPH

The PPT2 conjecture was a conjecture in quantum information theory which was recently disproven by OpenAI in a preprint they made public on October 6, 2026.

This is the privacy policy for the app Who’s That Feathered Friend?

Introduction Motivation Frankl’s conjecture History and progress 2016: a Polymath project The project Setup Shared tooling Methodology: very “unsupervised” Workstreams (or lines of attack) Many generalizations refuted Results: a lot of stuff… Observations…

Outline Deep learning as optimization Deep learning terminology Training Gradient descent Applying gradient descent to deep learning models Advantages of stochastic gradient descent (SGD) SGD variants Review: Jacobians Computing the gradient Backpropagati…

Fri, Sep 4 at 5:15 PM Me: The magpie hops, surely it does ChatGPT: Indeed it does.

What works and mostly, what doesn’t.

I visited family in Costa Rica and got to see tons of cool animals and birds.

Color quantization Usage Examples Are there clear breakpoints in file size as you quantize further?

In 2021, I had a project I called “birb stalker”.

Introduction Mixed-integer linear programs An analytical solution to a simpler problem Finding 676 3-letter package names satisfying our pairwise DL distance constraint Notes Introduction In my post yesterday Thoughts on being a package registry maintaine…

Introduction How did I become a maintainer & should I exercise that power?

I just keep vibe coding.

“Bus factor” is a somewhat morbid term used in software engineering (and possibly elsewhere), meaning the number of people who would have to get “hit by a bus” for a project to fall apart.

Note: the first half of the blog post documents my experience and takeways making valuetier.org, a webapp for helping users (or specifically, me) identify their values.

Deep learning as optimization, and then as human-in-the-loop meta-optimization.

Julia has a large ecosystem of over 10,000 registered packages in it’s “General” open-source package registry.

Why use deepcopy?

Eric P.

Joint work with Mosè Giordano, revisiting PackageAnalyzer.jl and what it finds across the Julia ecosystem.

Hannah Robertson, Eric P.

Jay Pathmanathan, Kim Laberinto, Hannah Robertson, Eric Hanson, Jacob Donoghue — Beacon Biosignals.

Joint work with Mosè Giordano, presenting results from PackageAnalyzer.jl, which inspects the packages in Julia’s General registry.

H.

My PhD thesis, supervised by Nilanjana Datta and Yan Pautrat.

VisualStringDistances.jl, which measures how close two strings look when printed, and an application of it: an automated check in the General registry’s auto-merge process to help prevent malicious lookalike registrations.

New features in Convex.jl, a few worked example problems, and where the package was headed next.

With Vishal Katariya, Nilanjana Datta and Mark M.

A talk for the Quantum Information Processing group meeting at the Cavendish Laboratory discussing the results of arXiv:1902.08173 placed in the physical context of repeated interaction systems.

With Koenraad Audenaert and Nilanjana Datta.

Three nice techniques In the past months, I’ve found myself really appreciating some nice techniques for constructing algorithms.

With Nilanjana Datta.

A talk at Beyond IID 2019 discussing the results of arXiv:1902.08173, which is work done in collaboration with Cambyse Rouzé and Daniel Stilck França.

A poster on my work with my supervisor Nilanjana Datta on uniform continuity bounds for the single-partite entropies, presented at the conference Fifth London Symposium on Information Theory in May 2019.

⊕ This post started as a talk for the CCIMI retreat, the slides of which are available here.

In a previous post, I discussed using type domain information to speed up generation of random density matrices with small dimension in Julia.

A brief presentation introducing partial Markovian evolution (classically), entanglement, and some of the results from arXiv:1902.08173, which was done in collaboration with Cambyse Rouzé and Daniel Stilck França.

In Lent 2018 and 2019, I gave the example classes and revision class for Cambridge’s Part III course “Quantum Information Theory”.

With Cambyse Rouzé and Daniel Stilck França.

A poster on my work with Hao-Chung Cheng, my supervisor Nilanjana Datta, and Min-Hsiu Hsieh on operational duality of some quantum information-theoretic protocols, presented at the conference Beyond IID 2018 in Cambridge, UK in July 2018, and at Quantum I…

Let \(\Lambda: \mathcal{B}(\mathcal{H})\to \mathcal{B}(\mathcal{H})\) be a quantum channelcompletely positive and trace-preserving map on a finite-dimensional Hilbert space \(\mathcal{H}\), and let \(d:= \dim \mathcal{H}\).

Update (19 January 2019): I looked at this code again, and realized I had made a few basic mistakes, such as constructing a statically sized normally-distributed random matrix via SMatrix{d,d,Float64, d*d}(randn(Float64, d,d)) which constructs a random ma…

With Hao-Chung Cheng, Nilanjana Datta and Min-Hsiu Hsieh.

I decided to redo my website using Hugo (before I was using Jekyll).

As I was rewriting my website, I found some visualizations I had stored on my old website to show a collaborator, and I figured it was worth writing a little to have a more proper place to put them; hence this post 😊.

I’m sure there are many ways to make an index in LaTeX, but I was asked this question recently and thought I’d put my response here, which is based on what I did in my combinatorics notes.

Update (26 January 10): The source code is now available under a MIT license at https://github.com/ericphanson/arxiv-search.

With Hao-Chung Cheng, Nilanjana Datta and Min-Hsiu Hsieh.

An essay submitted in the competition for a Smith-Knight or Rayleigh-Knight prize.

A poster on my work with my supervisor Nilanjana Datta on uniform continuity bounds for the single-partite entropies, presented at the conference Beyond IID 2017 in Singapore in August 2017, and the main conference of the thematic semester Analysis in Qua…

Winning entry for the 2017 CCIMI video contest.