Dima Pasechnik’s visit to Birmingham to collaborate with Sergey Shpectorov took place from 22/07/2015 to 24/07/2015. This is a quote from their report on this visit: “The focus of this collaboration is on the improvements to Sergey’s GAP package for computation of axial algebras. These are commutative non-associative algebras generated by idempotents satisfying certain fusion rules. The motivating example is the Griess-Norton algebra for the Monster sporadic simple group. The available programs for computations with axial algebras rely on the linear algebra functions available in GAP. This means that the program cannot handle partial algebras (intermediate stages) over the rationals whose dimension exceeds two-three thousand. At the same time, the algebras that we construct have significant groups of symmetries and, in particular, every partial algebra is a G-module for a certain group G. It is hoped that an efficient implementation of the basic module operations using the known character table of G will allow to increase the limit on the dimension by several orders. It is also hoped that, due to the fundamental nature of the module operations, the new routines will also find applications in many other places.”

Nick Loughlin (Newcastle) visited Olexandr Konovalov in St Andrews on September 7th-9th, 2015. We acknowledge very much his feedback, which we are reproducing below:

“Two principal aims of my visit to St Andrews were to get familiar with package design and developing a parallel implementation of existing sequential code. Secondary objectives included discussing the upcoming software carpentry workshop and GAP days in Manchester in November, GAP’s built-in and user-contributed profiling features and creating new objects and methods for them.

We focused on the *philosophy* of the package design good-practice as is usually practiced in the GAP community, and Olexandr said this will appear shortly as a blog post on his website; I’ve volunteered to help develop a hands-on tutorial in creating a package using the “example” package as a template. As far as parallel computing goes, we studied the shared-memory model implemented in HPC-GAP and the distributed model that SCSCP package is built around, and implemented some of my code in both so I have now a working example in each.

My secondary objectives were all happily satisfied at least in part. The one thing that we didn’t cover fully was profiling, and Olexandr suggested that perhaps this was a sensible subject for a tutorial at the upcoming GAP days.”