Span programs as a linear-algebraic representation of functions

I feel like TheEGG has been a bit monotone in the sort of theoretical computer science that I’ve been writing about recently. In part, this has been due to time constraints and the pressure of the weekly posting schedule (it has now been over a year with a post every calendar week); and in part due to my mind being too fixated on algorithmic biology.

So for this week, I want to change things up a bit. I want to discuss some of the math behind a success of cstheory applied to nature: quantum computing. It’s been six years since I blogged about quantum query complexity and the negative adversary method for lower bounding it. And it has been close to 8 years since I’ve worked on the topic.

But I did promise to write about span programs — a technique used to reason about query complexity. So in this post, I want to shift gears to quantum computing and discuss span programs. I doubt this is useful to thinking about evolution, but it never hurts to discuss a cool linear-algebraic representation of functions.

I started writing this post for the CSTheory Community Blog. Unfortunately, that blog is largely defunct. So, after 6 years, I decided to post on TheEGG instead.

Please humour me, dear reader.

Advertisements

Quick introduction: the algorithmic lens

Computers are a ubiquitous tool in modern research. We use them for everything from running simulation experiments and controlling physical experiments to analyzing and visualizing data. For almost any field ‘X’ there is probably a subfield of ‘computational X’ that uses and refines these computational tools to further research in X. This is very important work and I think it should be an integral part of all modern research.

But this is not the algorithmic lens.

In this post, I will try to give a very brief description (or maybe just a set of pointers) for the algorithmic lens. And of what we should imagine when we see an ‘algorithmic X’ subfield of some field X.

From perpetual motion machines to the Entscheidungsproblem

There seems to be a tendency to use the newest technology of the day as a metaphor for making sense of our hardest scientific questions. These metaphors are often vague and inprecise. They tend to overly simplify the scientific question and also misrepresent the technology. This isn’t useful.

But the pull of this metaphor also tends to transform the technical disciplines that analyze our newest tech into fundamental disciplines that analyze our universe. This was the case for many aspects of physics, and I think it is currently happening with aspects of theoretical computer science. This is very useful.

So, let’s go back in time to the birth of modern machines. To the water wheel and the steam engine.

I will briefly sketch how the science of steam engines developed and how it dealt with perpetual motion machines. From here, we can jump to the analytic engine and the modern computer. I’ll suggest that the development of computer science has followed a similar path — with the Entscheidungsproblem and its variants serving as our perpetual motion machine.

The science of steam engines successfully universalized itself into thermodynamics and statistical mechanics. These are seen as universal disciplines that are used to inform our understanding across the sciences. Similarly, I think that we need to universalize theoretical computer science and make its techniques more common throughout the sciences.

Quick introduction: Problems and algorithms

For this week, I want to try a new type of post. A quick introduction to a standard topic that might not be familiar to all readers and that could be useful later on. The goal is to write a shorter post than usual and provide an launching point for future more details discussion on a topic. Let’s see if I can stick to 500 words — although this post is 933, so — in the future.

For our first topic, let’s turn to theoretical computer science.

There are many ways to subdivide theoretical computer science, but one of my favorite divisions is into the two battling factions of computational complexity and algorithm design. To sketch a caricature: the former focus on computational problems and lower bounds, and the latter focus on algorithms and upper bounds. The latter have counter-parts throughout science, but I think the former are much less frequently encountered outside theoretical computer science. I want to sketch the division between these two fields. In the future I’ll explain how it can be useful for reasoning about evolutionary biology.

So let’s start with some definitions, or at least intuitions.

Open-ended evolution on hard fitness landscapes from VCSPs

There is often interest among the public and in the media about evolution and its effects for contemporary humans. In this context, some argue that humans have stopped evolving, including persons who have a good degree of influence over the public opinion. Famous BBC Natural History Unit broadcaster David Attenborough, for example, argued a few years ago in an interview that humans are the only species who “put halt to natural selection of its own free will”. The first time I read this, I thought that it seemed plausible. The advances in medicine that we made in the last two centuries mean that almost all babies can reach adulthood and have children of their own, which appears to cancel natural selection. However, after more careful thought, I realized that these sort of arguments for the ‘end of evolution’ could not be true.

Upon more reflection, there just seem to be better arguments for open-ended evolution.

One way of seeing that we’re still evolving is by observing that we actually created a new environment, with very different struggles than the ones that we encountered in the past. This is what Adam Benton (2013) suggests in his discussion of Attenborough. Living in cities with millions of people is very different from having to survive in a prehistoric jungle, so evolutionary pressures have shifted in this new environment. Success and fitness are measured differently. The continuing pace of changes and evolution in various fields such as technology, medicine, sciences is a clear example that humans continue to evolve. Even from a physical point of view, research shows that we are now becoming taller, after the effects of the last ice age faded out (Yang et al., 2010), while our brain seems to get smaller, for various reasons with the most amusing being that we don’t need that much “central heating”. Take that Aristotle! Furthermore, the shape of our teeth and jaws changed as we changed our diet, with different populations having a different structure based on the local diet (von Cramon-Taubadel, 2011).

But we don’t even need to resort to dynamically changing selection pressures. We can argue that evolution is ongoing even in a static environment. More importantly, we can make this argument in the laboratory. Although we do have to switch from humans to a more prolific species. A good example of this would be Richard Lenski’s long-term E-coli evolution experiment (Lenski et al., 1991) which shows that evolution is still ongoing after 50000 generations in the E-coli bacteria (Wiser et al., 2013). The fitness of the E. coli keeps increasing! This certainly seems like open-ended evolution.

But how do we make theoretical sense of these experimental observations? Artem Kaznatcheev (2018) has one suggestion: ‘hard’ landscapes due to the constraints of computational complexity. He suggests that evolution can be seen as a computational problem, in which the organisms try to maximize their fitness over successive generations. This problem would still be constrained by the theory of computational complexity, which tells us that some problems are too hard to be solved in a reasonable amount of time. Unfortunately, Artem’s work is far too theoretical. This is where my third-year project at the University of Oxford comes in. I will be working together with Artem on actually simulating open-ended evolution on specific examples of hard fitness landscapes that arise from valued constraint satisfaction problems (VCSPs).

Why VCSPs? They are an elegant generalization of the weighted 2SAT problem that Artem used in his work on hard landscapes. I’ll use this blog post to introduce CSPs, VCSPs, explain how they generalize weighted 2 SAT (and thus the NK fitness landscape model), and provide a way to translate between the language of computer science and that of biology.

Minimal models for explaining unbounded increase in fitness

On a prior version of my paper on computational complexity as an ultimate constraint, Hemachander Subramanian made a good comment and question:

Nice analysis Artem! If we think of the fitness as a function of genes, interactions between two genes, and interactions between three genes and so on, your analysis using epistasis takes into account only the interactions (second order and more). The presence or absence of the genes themselves (first order) can change the landscape itself, though. Evolution might be able to play the game of standing still as the landscape around it changes until a species is “stabilized” by finding itself in a peak. The question is would traversing these time-dependent landscapes for optima is still uncomputable?

And although I responded to his comment in the bioRxiv Disqus thread, it seems that comments are version locked and so you cannot see Hema’s comment anymore on the newest version. As such, I wanted to share my response on the blog and expand a bit on it.

Mostly this will be an incomplete argument for why biologists should care about worst-case analysis. I’ll have to expand on it more in the future.

Proximal vs ultimate constraints on evolution

For a mathematician — like John D. Cook, for example — objectives and constraints are duals of each other. But sometimes the objectives are easier to see than the constraints. This is certainly the case for evolution. Here, most students would point you to fitness as the objective to be maximized. And at least at a heuristic level — under a sufficiently nuanced definition of fitness — biologists would agree. So let’s take the objective as known.

This leaves us with the harder to see constraints.

Ever since the microscope, biologists have been expert at studying the hard to see. So, of course — as an editor at Proceedings of the Royal Society: B reminded me — they have looked at constraints on evolution. In particular, departures from an expected evolutionary equilibrium is where biologists see constraints on evolution. An evolutionary constraint is anything that prevents a population from being at a fitness peak.

In this post, I want to follow a bit of a winding path. First, I’ll appeal to Mayr’s ultimate-proximate distinction as a motivation for why biologists care about evolutionary constraints. Second, I’ll introduce the constraints on evolution that have been already studied, and argue that these are mostly proximal constraints. Third, I’ll introduce the notion of ultimate constraints and interpret my work on the computational complexity of evolutionary equilibria as an ultimate constraint. Finally, I’ll point at a particularly important consequence of the computational constraint of evolution: the possibility of open-ended evolution.

In a way, this post can be read as an overview of the change in focus between Kaznatcheev (2013) and (2018).

Algorithmic lens as Alan Turing’s wider impact

Today is Alan Turing’s birthday. He would have turned 106.

It has been too long since I last wrote about him on TheEGG. Today, I want to provide an overview of some of his most important work based on my and other’s answers on this old cstheory question. This will build slightly on a post I wrote two years ago for the Heidelberg Laureate Forum, but it will share a lot of text in common.

Turing is far from obscure. Every computer scientist and programmer has heard his name. The Nobel prize of Computer Science is named after him. He has even joined the ranks of mathematicians with feature-length films. Although a film that misrepresents much history. But even outside of film, I feel that our perceptions and representations of Turing are shaped too heavily by the current boundaries and constraints of computer science. Or at least how computer science is popularly (mis)understood.

Also, it is just easier to film the building a giant machine than about proving theorems and revolutionizing how we think about the world.

As the great breadth of his work shows, Turing would not recognize the disciplinary boundaries that confine computer science to technology. Like Abel Molina, he would see many motivations for computer science, from Science and Technology to Mathematics and Philosophy to Society. Turing viewed the whole world through the algorithmic lens. A wide ambition that is sometimes lacking in modern computer science.

In this post, I want to highlight some of the aspects of the world that Turing looked at.

Pairing tools and problems: a lesson from the methods of mathematics and the Entscheidungsproblem

Three weeks ago it was my lot to present at the weekly integrated mathematical oncology department meeting. Given the informal setting, I decided to grab one gimmick and run with it. I titled my talk: ‘2’. It was an overview of two recent projects that I’ve been working on: double public goods for acid mediated tumour invasion, and edge
effects in game theoretic dynamics of solid tumours. For the former, I considered two approximations: the limit as the number n of interaction partners is large and the limit as n = 1 — so there are two interacting parties. But the numerology didn’t stop there, my real goal was to highlight a duality between tools or techniques and the problems we apply them to or domains we use them in. As is popular at the IMO, the talk was live-tweeted with many unflattering photos and this great paraphrase (or was it a quote?) by David Basanta from my presentation’s opening:

Artem: I don't follow the "pick the best tool for the job" mentality 1/2 @mathonco

— David Basanta (@dbasanta) February 23, 2015

Artem: I rather pick the best goal with the best tool 2/2 @mathonco

— David Basanta (@dbasanta) February 23, 2015

Since I was rather sleep deprived from preparing my slides, I am not sure what I said exactly but I meant to say something like the following:

I don’t subscribe to the perspective that we should pick the best tool for the job. Instead, I try to pick the best tuple of job and tool given my personal tastes, competences, and intuitions. In doing so, I aim to push the tool slightly beyond its prior borders — usually with an incremental technical improvement — while also exploring a variant perspective — but hopefully still grounded in the local language — on some domain of interest. The job and tool march hand in hand.

In this post, I want to unpack this principle and follow it a little deeper into the philosophy of science. In the process, I will touch on the differences between endogenous and exogenous questions. I will draw some examples from my own work, by will rely primarily on methodological inspiration from pure math and the early days of theoretical computer science.

Five motivations for theoretical computer science

There are some situations, perhaps lucky ones, where it is felt that an activity needs no external motivation or justification.  For the rest, it can be helpful to think of what the task at hand can be useful for. This of course doesn’t answer the larger question of what is worth doing, since it just distributes the burden somewhere else, but establishing these connections seems like a natural part of an answer to the larger question.

Along those lines, the following are five intellectual areas for whose study theoretical computer science concepts and their development can be useful – therefore, a curiosity about these areas can provide some motivation for learning about those cstheory concepts or developing them. They are arranged from the likely more obvious to most people to the less so: technology, mathematics, science, society, and philosophy. This post could also serve as an homage to delayed gratification (perhaps with some procrastination mixed in), having been finally written up more than three years after first discussing it with Artem.