james@certora.com
jrw12@cs.washington.edu
I am chief blockchain skeptic and formal verification shill at Certora.
This is my academic homepage. Before joining Certora I was a graduate student at the University of Washington. My research interests are in programming languages and applications of PL techniques to systems, and I continue to be active in research both in my day job and beyond. I'm also a sucker for math, music, and puzzles.
October 1, 2020.
Our paper "Learning the Boundary of Inductive Invariants" has been
conditionally accepted to POPL 2021 (submitted version).
We combine a little bit of Hamming geometry with deep results from exact learning theory
to identify a class of systems and invariants that can be inferred using
only a polynomial number of calls to a SAT solver. These results identify
a relatively easy class of verification problems and partially explain why
state-of-the-art algorithms perform so well in practice.
September 30, 2020.
I am teaching UW's undergraduate programming languages course, CSE341, this quarter.
Check out the course website!
April 6, 2020.
Our paper about Armada will appear in PLDI 2020 (preprint).
Armada is a verification system for shared-memory multithreaded programs
that allows programmers to write verified systems whose performance matches
unverified counterparts, by giving them control over memory layout and choice
of synchronization primitives. The key insight is that high-level reasoning
principles from the literature can be proved sound against a low-level
semantic model that is expressive enough to support these programmer choices.
April 6, 2020.
Our paper about
Szalinski will appear in PLDI 2020 (preprint).
Szalinski is a tool for decompiling low-level CAD models into higher-level
CAD programs that use higher-order functions like map and fold. Szalinski
uses E-graphs to efficiently reason about the space of programs that are
equivalent to the input model.
March 30, 2020.
In spring 2020, I am teaching CSE 490P (Advanced Programming Languages and Verification).
Check out the course webpage!
I am the Chief Technology Officer of Certora, a startup based in Tel Aviv that builds formal verification tools for blockchain smart contracts. I live and work in Seattle.
This is my academic homepage. Before joining Certora I was a graduate student at the University of Washington, where I was advised by Zach Tatlock in the PLSE group. My research interests are in programming languages, systems, and formal methods. My thesis work is on compositional techniques for verifying distributed systems implementations. I generally enjoy working with proof assistants and SMT solvers on applications to all kinds of concurrent programming. I also dabble in floating point, compilers, and 3D printing.
Before grad school, I did my undergraduate at Williams College, graduating in 2013, where I worked with Steve Freund on dynamic race detection. Since then Steve and I have continued to collaborate, including on an "our powers combined" paper on verified dynamic race detection with Cormac Flanagan.
Outside computer science, I enjoy good coffee, choral music, distance running, and small planes. I do not enjoy cars of any size.
I sing baritone in the St. Mark's Cathedral Choir, Evensong Choir, and Compline Choir. The Compline Choir performs each Sunday night at 9:30pm at St. Mark's. The Compline service a 30 minute chanted/sung service that tends to draw hundreds of people every week and thousands via a live radio broadcast and the podcast. It's a classic Seattle experience. You should check it out! You can listen live on King FM or get the podcast.
I occasionally play handbells.
Finally, I like to ride my bike (a Trek 520): in 2009 I biked the TransAm. I'm always thinking about my next tour.
February 21, 2017.
Exercises on Generalizing the Induction Hypothesis.
This post collects several Coq exercises on generalizing the
induction hypothesis.
January 9, 2017.
A Port of the Proof of Peterson's Algorithm to Dafny.
This code-only post is a port of the proof of Peterson's Algorithm to Dafny.
It also serves as a good example of how to reason about concurrent systems
in Dafny, essentially by writing a thread scheduler.
April 24, 2016.
How to build a simple system in Verdi.
In this long-awaited post, we'll show how to implement and verify
a simple distributed system using network semantics.
May 8, 2015
A Proof of Peterson's Algorithm.
In this post, we take a break from distributed systems to look at shared
memory systems. As a case study, we give a proof that Peterson's algorithm
provides mutual exclusion.
April 16, 2015
Network Semantics for Verifying Distributed Systems.
This is the first post in a series
on Verdi. In this post,
we'll get our feet wet by defining a formal model of how
distributed systems execute on the network.
October 20, 2014
Reasoning about Cardinalities of Sums and Products.
In this short, code-heavy post, we extend some of the work from
a previous post to reason about
the cardinalities of sums and products.
September 14, 2014
Dependent Case Analysis in Coq without Axioms.
This post shows how to get around the limitations of
the destruct
tactic when doing case analysis on dependent
types, without resorting to the dependent destruction
tactic,
which relies on additional axioms.
September 4, 2014
"run" + "time" = ???.
This brief post records Mike's description of the three ways of
combining the words "run" and "time" in computer science
writing.
June 12, 2014
"More Sums than Differences" Sets, Part 2: Counting MSTD Sets.
This is the (much delayed) second post in a series on More Sums
than Difference Sets. In this post, we'll take a first crack at the
question, "How many MSTD sets are there?" To do so, we'll write a
straightforward C program that counts MSTD sets. Then we'll run it to
count MSTD sets and benchmark its performance.
April 10, 2014
Tail Recursion Modulo cons.
Tail recursion has come up in a few conversations this
week. This post explores a generalization of tail call
optimization that I wasn't aware of
until Doug described it to me.
March 3, 2014
"More Sums than Differences" Sets, Part 1: A puzzle.
This is the first post in a series on "More Sums than
Differences" Sets. In this post, we'll get our terminology
straight and ask a lot of questions.
December 31, 2013
Easy access to the off-campus proxy.
I use the UW proxy to access the ACM digital library from off campus, but it's annoying to
type the proxy URL every time I click a link to a new paper. Here are two ways to make life
easier.
In autumn 2020, I am teaching UW's undergraduate programming languages course, CSE341. Check out the course webpage!
In spring 2020, I am teaching CSE 490P (Advanced Programming Languages and Verification). Check out the course webpage!
In winter 2017, I taught CSE 341 (Programming Languages). Check out the course webpage!
I like books. I am guilty of being somewhat more of a collector than a reader, but I don't let that get me down.
I currently have around 120 technical books in my office (listed behind that link). I have another 150 or so technical books at home that are not-so-related to my research, and another 50 or so non-technical books as well. Someday, I'll list them all here.
I am happy to lend books to anyone interested. Just let me know!
I am also very happy to receive recommendations of good books that I'm missing!