Math-Hack Day
Registration is open for the math-hack day of INTEGRATE ASG on 25 April
– Published 15 April 2016
INTEGRATE ASG is organising a math-hack day at the Pufendorf Institute on 25 April where three topics will be discussed:
- Dots on a sphere
- Monte-Carlo populations
- Modelling protein chains
All three will be presented at the beginning of the day (see programme below). People
will then split up into groups (part choice, part random, part seasonal shuffle) and discuss/work
on the topics over the day. We will all meet up again mid-afternoon when the groups will tell
us what they got up to, ideas they had and how it went in general.
We have the main lecture hall but also other rooms in the Pufendorf booked, so there
will be plenty of space to spread out.
Bring your laptops. Note that the Pufendorf has eduroam as its wifi network.
The day will be fueled by coffee, tea, and a sandwich lunch.
This will be a day for those who want to work together to code and solve problems.
Those wishing to attend what promises to be a fun and interesting day should REGISTER by sending an email with subject "Registration for 25 April INTEGRATE" to Melvyn B. Davies (mbd@astro.lu.se).
Note that as places are limited, you are encouraged to register early to avoid disappointment. In any case please register by 12.00 on Tuesday 19 April.
INTEGRATE MATH-HACK DAY -- Monday, 25 April
Location: Pufendorf Institute, Lund University
09.30-09.45 Tea/coffee
09.45-09.50 Melvyn B. Davies -- Welcome and introduction to INTEGRATE Math-hack Day
Introduction to the topics:
09.50-09.55 Dots on a sphere -- Alexey Bobrick
09.55-10.00 Monte-Carlo populations -- Alex Mustill
10.00-10.05 Modelling protein chains -- Anders Irbäck
10.05-12.30 group discussions/work
12.30- 13.00 Lunch sandwiches at Pufendorf
13.00-14.30 further group discussions and groups prepare presentations
Progress reports:
14.30-14.50 Dots on a sphere-- group member(s)
14.50-15.10 Monte-Carlo populations -- group member(s)
15.10-15.30 Modelling protein chains -- group member(s)
15 minutes max for the group presentations + at least 5 minutes for discussion
15.30-16.00 Tea/coffee
MATH-HACK DAY TOPICS
Topic A) Dots on a sphere -- Alexey Brobrick
What is the best way to put points as uniformly as possible onto a spherical shell? This question arises in a number of settings.
For example, in using a hydrodynamics method known as smooth particle hydrodynamics, a ball of gas (which could
be a star or planet) is modelled as a spherical distribution of particles. One therefore wishes to have an algorithm for
placing particles on the surface of a sphere as evenly as possible. A related question concerns the locations of
vertices in grid-based fluid dynamics methods.
Topic B) Monte-Carlo populations -- Alex Mustill
In theoretical astrophysics one may wish to generate synthetic populations of objects whose properties are consistent with an observed population: for example, generate a synthetic population of planets whose distributions of mass and orbital radius are consistent with discovered planets. This can be done by Monte-Carlo sampling from a distribution for the properties consistent with the observed one, but what is the best way of doing this?
- Prescribing a functional form for the distribution and finding the best-fit parameters?
- Interpolation on the observed cumulative distribution?
- Using kernel density estimates or similar non-parametric methods?
How does the suitability of each method depend on the number of objects in the observed population and on the dimensionality of the parameter space?
Topic C) Modelling protein chains -- Anders Irbäck
In modeling protein chains, the main degrees of freedom are torsion angles (bond lengths and bond angles vary much less).
In many problems, it is useful to be able to modify the structure of a small segment of the chain, while leaving the rest of the chain
unchanged. Solutions to this problem exist, some of which are inspired by methods for similar problems in robotics. However, in
simulations of proteins focusing on the main degrees of freedom, this problem is to be solved over and over again. Much could
potentially be gained by finding methods with improved speed and robustness.