Simulation and calculation

Despite their participation in the most ruthless and domineering academic journal cartel*, Studies In History and Philosophy of Science Part B: Studies In History and Philosophy of Modern Physics remains my favourite ‘philosophy of physics’ journal, though the ’t Hooft-edited Foundations of Physics is arguably more impactful. And their current issue, a bumper crop of works on modelling and simulation put together by Århusers** Århusianerer Matthias Heymann and Helge Kragh, is both timely and somewhat indulgent.

It is, alas, behind the Great Paywall of ScienceDirect, but if you skim the contents you might be able to find a preprint elsewhere for any articles that catch your eye. No such luck for the paper I want to spotlight here—are you reading this Sunderberg? No one will stop you putting a copy of the work on your website!—‘Cultures of simulations vs. cultures of calculations? The development of simulation practices in meteorology and astrophysics‘.

This is a sociological work that contains much of interest to those who read S. D. M. White’s 2007 essay on ‘Fundamentalist Physics’ and concluded that a complementary study from someone well outside the field might be useful too. Sunderberg’s article is somewhat drier in tone and focusses on the concrete development of simulation practices in astrophysics (and meteorology). My favourite paragraph:

The so-called Millenium simulation appears as somewhat of an astrophysical equivalent to [the climate modelling code] CMIP3 in terms of offering astro- physicists the opportunity to take output from a database, analyze, and publish papers on the basis of it, without having been involved in the simulations themselves (see http://www. virgo.dur.ac.uk/). The simulation is one of the ‘‘largest ever’’ of the formation of structure in the universe. It is an outcome of the Virgo consortium, which is an international collaboration among scientists from roughly the same countries as those that have important climate modeling centers (the UK, Germany, the Netherlands, Canada, the USA, and Japan). In terms of usage of the dataset, this effort also seems to have been a success. The Millenium simulation homepage currently lists 243 articles, written by astrophysicists all over the world, ‘‘that have directly used the Millennium Simulation data, and that we spotted on the astro-ph preprint server’’ (see http://www.mpa-garching.mpg.de/ millennium/#VISUAL_MATERIAL). However, the title of these publications and several abstracts show that Millenium output is used to contribute to (initialize, compare, etc.) some other calculation rather than analyze itself. Another major difference compared to CMIP is that the Millenium simulation was produced by using one particular simulation code and it had nothing to do with intercomparisons of codes. Compared to the use of data from CMIP, the Millenium simulation is not obviously a sign of the ‘‘superficiality’’ of a culture of simulation.

Good stuff throughout.

* No link for you, Elsevier!
** What is the preferred nomenclature for ‘people from Århus’?

Advertisement

Read Full Post »

Public Speaking For Scientists 7: Powerpoint

I can’t do a series on public speaking for scientists without laying down a few guidelines on Powerpoint. A quick note: the best lecturers I had during my undergraduate days consistently shunned Powerpoint, preferring the blackboard. (Overhead transparencies, thankfully, seem to be dying out). One of the reasons for this may be that equations are much less intimidating when they appear gradually, term by term, rather than all at once. Lecturers who used the blackboard also seemed to have more time to think about what they would present and how they would present it, rather than using time to prepare slides.

Powerpoint, as we all know, can be used well and can be used extremely poorly. Here are the laws of Powerpoint – if I were Emperor of astronomy, non-compliance would be punishable by firing squad.

Rule 1: Contrasting colours

If I see one more person put a yellow line on a white background …

The RGB colour model is part of the problem. The RGB model (as used by Matlab) assigns a colour by a fraction for Red, Green and Blue. For example: white (1, 1, 1), black (0, 0, 0), dark purple (0.5,0,0.5). While blue (0, 0, 1) and red (1, 0, 0) are clearly seen on a white background, green (0, 1, 0) is invisible. So you’ll need to use (0, 0.5, 0). Many “default” colours need to be darkened. If you’re using a black background, lighten the colours (especially red). If you’re using a background that is some other colour than black or white, stop it. Stop it right now. Or consult the nearest colour wheel and choose opposing colours, one dark and one light. Test your colours on a projector – they always look better on a screen or on paper.

And use thick lines on plots. And large labels. Or else.

Rule 2: Few words

It’s very difficult to read and listen at the same time. The words you put on screen should be keywords, and as few as possible. Don’t write a complete sentence unless you plan on saying it word for word. Don’t write:

Read Full Post »

Letter to the editor of the day

From the Sydney Morning Herald, to be read aloud:

As a member of the Australian Anti-Alliteration Association can I please ask Peter Pitt of Potts Point to move suburbs.

Ross Fyfe Lane Cove

Read Full Post »