QMC:Community Portal
From QMC
Contents |
[edit] Ideas for New Wiki Pages
Perhaps you have an idea for some Wiki pages but you want some feedback before going forward. Make a description of your concept below and invite comment from others. This page is NOT intended for QMCWiki policy discussions (e.g. technical things to improve the Wiki, who and how to invite participation); that discussion should go here.
[edit] A forum bulletin board
Although many online bulletin boards have problems with spam posts, I think they are worth the effort to maintain since it allows people to contribute without having to worry about anything wiki related. There certainly are a bunch of questions I would like to ask the QMC community, but haven't gone to any QMC specific events yet.
[edit] Brief History of Quantum Monte Carlo.
We all have to give talks or write papers where we provide some early reference to the main QMC ideas, to Fermi, Wigner, Feynman,... In these pages, we could collect and discuss the contribution of early works (pre 1975?), and scan in papers not available online.
[edit] Major Algorithmic Challenges and Review Article Nucleation
We could have discussion about some of the major challenges. I think these have the possibility of becoming traditional review articles. What is the advantage to having a review article written as a Wiki and not the traditional way?
- Perhaps it can nucleate a group activity that would otherwise not happen: no one individual may be able to do all the work for a review article.
- The community collectively has more expertise than a single author. This would keep a review article from being slanted toward the author's work.
- The wiki can stay current after the review article is put to bed.
- Review articles are a 'good thing' for the community. People outside the community can be drawn into the field and help us or at least know what we are up to.
- Review articles have much larger citation counts than individual research papers. (Note that when you say that you have a highly cited paper, you do not have to divide by the number of co-authors, e.g. the h-index.)
- Reviews of Modern Physics restricts articles to 50 pages. The full length version on the Wiki could be longer with supplementary data, programs, etc.
I see no advantage (that offsets the problems) to having a review article written on the wiki unless the authorship is potentially open at some stage to the wiki members. On the other hand, a large group cannot write a review article without some leadership. Attention to authorship issues is needed, hopefully before any given project gets too far. Rules for authorship should include: all those and only those that contribute significantly can be considered as authors; all authors have to take full responsibility for the final product.
Does anyone know of any scientific communities that have attempted this? Are there existing projects or discussions?
A few ideas are:
- The fermion sign problem. What is it? What approaches have been investigated? I have an old page (now offline) to contribute. As far as I know, there is no recent, general discussion in the literature.
- The order N problem in quantum Monte Carlo.
- Quantum Dynamics, in general, and within linear response.
[edit] Strengths and weaknesses of QMC with comparisons to other methods
I (Kevin) would like to see, and can help write, the scaling of various QMC methods for various problems and comparisons to competing methods. I think this would help newcomers to have a better idea of what problems QMC can likely attack and how big the problems need to be before QMC is the method of choice. What do you think?
DMC: One can look at the historical record, and find which QMC papers have had impact outside the QMC community. (for example a page that list recent QMC articles that have had an impact outside the QMC community) This is relatively objective, but admittedly backword looking. I think it is hard work to get relaiable scaling data, since it depends so much on computers, software AND most importantly, estimates of the real errors. Also scaling itself can be quite misleading; what usually is needed is which methods can solve a given problem within some confidence limits, with today's hardware and software. That aside, I think the big impacts of QMC are in areas where the alternative methods are not competitive. Of course massively parallel computing does alter the balance and will enlarge the problem space where QMC is competitive.
[edit] Crowdsourcing
(see article in Wired ) The idea of this page is that people would pose problems, hopefully not too difficult, and readers would send in solutions or suggestions. In the commercial world, you have InnoCentive--but they have money as a reward. I guess we could dangle the prospect of co-authorship of articles to whoever solves the problem, or maybe just goodfeeling.
