POEMA learning week 2: Day 1

Today we had a half day but productive one. The speaker Benoît Legat lead us through the jungle of code relating to polynomial optimization using Julia before jumping into examples. In particular we did a rather deep dive into SumOfSquares.jl a sub package of  JuMP.

At first glance there is staggering amount of packages for polynomial in the relatively young language of Julia. Upon some reflection, it is clear why. Just as new cities use the ruins of their predecessors as foundation so too code is written on the failings of previous code. In the case of software the cycle of innovation and improvement is much faster. Keeping up-to-date is hard work. Navigating the jungle of old and new code is no small feat for the new and uninitiated. Fortunately we now have this lecture to guide use a bit in understanding how polynomial optimization fits into the larger Julia/Jump ecosystem.

In code, and perhaps math in general it is good to have a minimal viable example when introducing new information. However, care has to be taken as the wrong example can just as easily confuse as elucidate. A way around this is just to have many examples. Today we were given a cornucopia of examples in the SumOfSquares.jl Documentation.

To concluded, I now have a better understanding of the polynomial optimization packages out there in the wild and I have a whole bag of new code examples to try.


Comments

Post a Comment

Popular posts from this blog

POEMA learning week 2: Day 2

Image
Today was a full day. Starting with a triple whammy of POEMA ESR talks we got off to a running start. I will try my best to recall the highlights (based on my own limited understanding and memory) of the day: ESR2 Andrew  : Attacking polynomial optimization problems via the critical values route. This means computing the Jacobian of the system and then solving a determental system using a very subject specific algorithm Sparse FGLM algorithm . ESR4 Alejandro  : A verbose intro into real zero (RZ) polynomials from theory down to numerics. I had hoped to see a clearer link between RZ polynomials and the Generalized Lax Conjecture but hey-ho. ESR12 Arefeh : Leveraging both primal and dual formulations in solving SDP's. Also extensions to sparse large SDP's and preconditioners for low rank SDPs. Considering how much I rely on SDP solvers I am painfully unaware of all the intricacies of their operation. I am glad Arefeh is an expert I can refer to should the need arise. Prof. Dr. Th
Read more

Purpose of this blog: To log my mathematical endeavours.

Why does this blog exist? I am required to maintain a blog of my research activities as mandated by my contract with POEMA. Secondly this blog will serve as record for evaluating my progress (both for me and my superiors) as well as a platform to showcase my interests (to possible collaborators). Thirdly, it serve as a source of information to anyone interested in me or the topics I study. I must warn the reader that I am a novice to the practice of blogging. Hence, constructive criticism is welcomed. What is the content of this blog? In this blog I will record my mathematical endeavours, indexed by the week in which they take place. The mathematical endeavours are directed towards the topic of :  Approximation hierarchies for (non-)commutative polynomial optimisation. This topic falls under general polynomial optimisation techniques and is know to use tools from (but not limited to): real algebra moment theory operator theory semidefinite optimisation. In addition to f
Read more