Options

A new category for modeling?

In a conversation with Jim Stuttard, the idea came up of starting a new category here called "Climate Modeling" or something... where people could talk about climate models with more focus on climate and less focus on software than in "Azimuth Code Project". The idea is that people interested in climate models but not so much interested in arcane aspects of programming could read that.

I'm not sure this makes sense. Maybe it will only make sense later, when we get more people interested in climate modeling who aren't programming fiends.

Comments

  • 1.

    Here's a preprint of a paper for SIGGRAPH 2012 on a Laplacian eigenfunction basis approach to fluid dynamic simulation. The paper has a good introductory methods survey. Laplacian eigenfunctions have advantages and can be used with both grid and non-grid methods.

    There is a video and another interesting paper and video on efficient and accurate rotation of spherical harmonic expansions.

    Computer graphics as a field (apart from complexity and performance issues for architectures) makes no mention of computers eg. Foley and Van Dam's standard text on Interactive Computer Graphics.

    As visualisations have identical mathematics to simulations I wonder if the computer graphics community has managed to contribute any new maths or insights for climate models?

    Comment Source:Here's a preprint of a [paper](http://www.dgp.toronto.edu/~tyler/fluids/FluidDynamicsLaplacianEigenfunctions.pdf) for SIGGRAPH 2012 on a Laplacian eigenfunction basis approach to fluid dynamic simulation. The paper has a good introductory methods survey. Laplacian eigenfunctions have advantages and can be used with both grid and non-grid methods. There is a [video](http://www.dgp.toronto.edu/~tyler/fluids/) and another interesting [paper](http://www.dgp.toronto.edu/~tyler/) and video on efficient and accurate rotation of spherical harmonic expansions. Computer graphics as a field (apart from complexity and performance issues for architectures) makes no mention of computers eg. Foley and Van Dam's standard text on Interactive Computer Graphics. As visualisations have identical mathematics to simulations I wonder if the computer graphics community has managed to contribute any new maths or insights for climate models?
  • 2.

    Jim wrote

    As visualisations have identical mathematics to simulations I wonder if the computer graphics community has managed to contribute any new maths or insights for climate models?

    I don't know about climate science, but I believe some of the fields that use meshes of nodes in scientific simulations use "mesh simplification" algorithms that originated in graphics.

    Comment Source:Jim wrote > As visualisations have identical mathematics to simulations I wonder if the computer graphics community has managed to contribute any new maths or insights for climate models? I don't know about climate science, but I believe some of the fields that use meshes of nodes in scientific simulations use "mesh simplification" algorithms that originated in graphics.
  • 3.
    edited June 2012

    Jim wrote:

    Here's a preprint of a paper for SIGGRAPH 2012 on a Laplacian eigenfunction basis approach to fluid dynamic simulation. The paper has a good introductory methods survey. Laplacian eigenfunctions have advantages and can be used with both grid and non-grid methods.

    There is a video and another interesting paper and video on efficient and accurate rotation of spherical harmonic expansions.

    It would be great if you put such information and links onto the wiki. Don't be shy about creating new pages. It's suboptimal to put links to information about Laplacian eigenfunctions into a general strategy discussion of whether we should have a category called "Modelling" here on the Forum. Someday one of us will really want that information, and they probably won't look here.

    The first generation of Azimuthers (e.g. Graham Jones, David Tweed, Staffan Liljgeren and Frederik de Roo) put a lot of effort into the wiki, and then most of them slowed down, perhaps out of exhaustion. The second generation (e.g. Allan Erskine, Jim Stuttard and Glyn Adgie) seem to be writing lots of software, which is great... but sometimes I'm afraid that because they missed the literally hundreds of conversations about building the wiki, they don't really use it as much as they could, or should.

    I too got a bit exhausted writing wiki articles, so I'm not setting a good example these days. But I hope that when I teach some classes about this stuff I'll use the wiki to create my course notes, or even assign homework where I force students to help write wiki pages.

    As for the original topic: it seems like nobody is crying out for a new category called "Modelling"... but Jim's remark here would fit quite nicely in that category!

    Comment Source:Jim wrote: >Here's a preprint of a paper for SIGGRAPH 2012 on a Laplacian eigenfunction basis approach to fluid dynamic simulation. The paper has a good introductory methods survey. Laplacian eigenfunctions have advantages and can be used with both grid and non-grid methods. > There is a video and another interesting paper and video on efficient and accurate rotation of spherical harmonic expansions. It would be great if you put such information and links [onto the wiki](http://www.azimuthproject.org/azimuth/show/How+to). Don't be shy about creating new pages. It's suboptimal to put links to information about Laplacian eigenfunctions into a general strategy discussion of whether we should have a category called "Modelling" here on the Forum. Someday one of us will really want that information, and they probably won't look here. The first generation of Azimuthers (e.g. Graham Jones, David Tweed, Staffan Liljgeren and Frederik de Roo) put a lot of effort into the wiki, and then most of them slowed down, perhaps out of exhaustion. The second generation (e.g. Allan Erskine, Jim Stuttard and Glyn Adgie) seem to be writing lots of software, which is *great*... but sometimes I'm afraid that because they missed the literally hundreds of conversations about building the wiki, they don't really use it as much as they could, or should. I too got a bit exhausted writing wiki articles, so I'm not setting a good example these days. But I hope that when I teach some classes about this stuff I'll use the wiki to create my course notes, or even assign homework where I force students to help write wiki pages. As for the original topic: it seems like nobody is crying out for a new category called "Modelling"... but Jim's remark here would fit quite nicely in that category!
  • 4.
    edited June 2012

    As for the original topic: it seems like nobody is crying out for a new category called "Modelling"... but Jim's remark here would fit quite nicely in that category!

    I'm undecided. Right now activity is so low that I don't have a problem with checking all active threads :-)

    Comment Source:<blockquote> <p> As for the original topic: it seems like nobody is crying out for a new category called "Modelling"... but Jim's remark here would fit quite nicely in that category! </p> </blockquote> I'm undecided. Right now activity is so low that I don't have a problem with checking all active threads :-)
  • 5.
    edited June 2012

    Going slightly off-topic, I haven't so much slowed down on the wiki out of exhaustion as just not had the time. I have the great inherent advantage over most of the other contributors that my first language is English (even though you might doubt that given what I write!) and even then writing a typical article takes a solid block of between one to two hours to even after I've done however many hours of "web-research" to know what to write. At the end of last year I also bought an "in need of severe work" house planning to live in it and fix it up over the course of years, and unfortunately it's turned out it may need to be in saleable condition within months, so I've been spending a large proportion of my time doing DIY. What other time I've got is being spent trying to improve some aspects of the Julia language to increase its suitability for "rapidly prototyping simulations/data analysis". As such it's looking at a potential tool that might be used within Azimuth work, but it's not directly part of the Azimuth goals.

    I really enjoyed writing articles and hope one day to have enough freedom and blocks of free time to start contributing them again.

    Comment Source:Going slightly off-topic, I haven't so much slowed down on the wiki out of exhaustion as just not had the time. I have the great inherent advantage over most of the other contributors that my first language is English (even though you might doubt that given what I write!) and even then writing a typical article takes a solid block of between one to two hours to even after I've done however many hours of "web-research" to know what to write. At the end of last year I also bought an "in need of severe work" house planning to live in it and fix it up over the course of years, and unfortunately it's turned out it may need to be in saleable condition within months, so I've been spending a large proportion of my time doing DIY. What other time I've got is being spent trying to improve some aspects of the [Julia language](http://julialang.org) to increase its suitability for "rapidly prototyping simulations/data analysis". As such it's looking at a potential tool that might be used within Azimuth work, but it's not directly part of the Azimuth goals. I really enjoyed writing articles and hope one day to have enough freedom and blocks of free time to start contributing them again.
  • 6.

    I didn't so much get exhausted as get a job (last April). It's a research job, modelling and programming in the area of evolutionary biology. As of this April, it is half-time, so I should now have spare time to contribute more, though that hasn't happened yet. I am more interested in biology than physics, and am disappointed Azimuth hasn't attracted many biologists. I would be more inclined to contribute if I thought biologists might read what I wrote. And personally, I am more interested in learning about network theory than climate science.

    John: I know you mean Graham Jones when you say Graham Young because you've done it a couple of times before. I wonder why?

    Comment Source:I didn't so much get exhausted as get a job (last April). It's a research job, modelling and programming in the area of evolutionary biology. As of this April, it is half-time, so I should now have spare time to contribute more, though that hasn't happened yet. I am more interested in biology than physics, and am disappointed Azimuth hasn't attracted many biologists. I would be more inclined to contribute if I thought biologists might read what I wrote. And personally, I am more interested in learning about network theory than climate science. John: I know you mean Graham Jones when you say Graham Young because you've done it a couple of times before. I wonder why?
  • 7.
    edited June 2012

    In my case, it's a combination of lack of (finding) time, and also that I'm not an expert in many topics to write decent articles. But it seems I'll soon have to start collecting everything I do know something about:

    CERN Press Release

    Comment Source:In my case, it's a combination of lack of (finding) time, and also that I'm not an expert in many topics to write decent articles. But it seems I'll soon have to start collecting everything I do know something about: [CERN Press Release](http://press.web.cern.ch/press/PressReleases/Releases2011/PR19.11E.html)
  • 8.

    John wrote:

    As for the original topic: it seems like nobody is crying out for a new category called "Modelling"... but Jim's remark here would fit quite nicely in that category!

    I didn't notice the title of this discussion was 'Policies and Conventions/A new category for modeling?' rather than 'Climate Modelling'. So I posted an example of the sort of non-computational paper I had in mind. It's mathematical with what might be called abstracted domain dependencies ie. not much domain model knowledge (engineering?).

    Without bikeshedding I think this suggests maybe having separate categories for detailed, technical discussions of maths, computation and domain (engineering) models.

    Learners might be able to use these separate areas as different road maps towards the front line. For example, learning from a survey paper that there have been ~3000 papers published on stochastic modelling over the past 20 years. After I had recovered from turning white, I resolved to ask which ones do I need? What's controversial? There would have to be answers from both the domain and from the maths ie. the engineering and the science.

    Given the low traffic it doesn't seem necessary to separate domain and maths discussions until we get to a point where we have to have separate versions for beginners like me.

    Another case might be that, if I've read it correctly, in Isaac Held's survey article he points out that (is it Jerry or Gerry? (sic)) North's model has some sort of direct slab to atmosphere coupling which differs from the models Isaac and +Tim Merlis design. It seems obvious to ask for +Nathan Urban to explain the domain assumptions and possibly different mathematics involved in his and the other two approaches (if such a distincition makes any sense).

    So we already have the Azimuth Code Project pages. I need somewhere to ask detailed science or engineering questions where the people who know the answers will read them.

    John also wrote:

    I'm afraid that because they missed the literally hundreds of conversations about building the wiki, they don't really use it as much as they could, or should.

    There must be getting on for 1000 wiki pages? Is it time for some refactoring? I think the project needs the help of a librarian. I've still only read a fraction of them.

    Only us geeks need to chat about tech and that's nearly all already documented (but massive) somewhere else. My alibi is I don't have the critical domain or maths knowledge evern to write paper and book reviews. I do keep detailed study notes as documented code sketches, written and not yet returned to. I will add a slab ocean page to the wiki when I've got some worthwhile correct and documented code. If I finish any of my maths and climate literature surveys I'll just make a discussion page until they've been reviewed. Iiuc the wiki was specified to be of the same standard (and this case, presumably crowd-source reviewed) as the ncat lab with non of the gaps and errors of wikipedia.

    PS. I didn't receive any notification of posts #4-#7. There is also the mysterious "if you do not visit...you will not receive any further notifications" message?

    Comment Source:John wrote: > As for the original topic: it seems like nobody is crying out for a new category called "Modelling"... but Jim's remark here would fit quite nicely in that category! I didn't notice the title of this discussion was 'Policies and Conventions/A new category for modeling?' rather than 'Climate Modelling'. So I posted an example of the sort of non-computational paper I had in mind. It's mathematical with what might be called abstracted domain dependencies ie. not much domain model knowledge (engineering?). Without [bikeshedding](http://en.wikipedia.org/wiki/Parkinson%27s_Law_of_Triviality) I think this suggests maybe having separate categories for detailed, technical discussions of maths, computation and domain (engineering) models. Learners might be able to use these separate areas as different road maps towards the front line. For example, learning from a survey paper that there have been ~3000 papers published on stochastic modelling over the past 20 years. After I had recovered from turning white, I resolved to ask which ones do I need? What's controversial? There would have to be answers from both the domain and from the maths ie. the engineering and the science. Given the low traffic it doesn't seem necessary to separate domain and maths discussions until we get to a point where we have to have separate versions for beginners like me. Another case might be that, if I've read it correctly, in Isaac Held's survey article he points out that (is it Jerry or Gerry? (sic)) North's model has some sort of direct slab to atmosphere coupling which differs from the models Isaac and +Tim Merlis design. It seems obvious to ask for +Nathan Urban to explain the domain assumptions and possibly different mathematics involved in his and the other two approaches (if such a distincition makes any sense). So we already have the Azimuth Code Project pages. I need somewhere to ask detailed science or engineering questions where the people who know the answers will read them. John also wrote: > I'm afraid that because they missed the literally hundreds of conversations about building the wiki, they don't really use it as much as they could, or should. There must be getting on for 1000 wiki pages? Is it time for some refactoring? I think the project needs the help of a librarian. I've still only read a fraction of them. Only us geeks need to chat about tech and that's nearly all already documented (but massive) somewhere else. My alibi is I don't have the critical domain or maths knowledge evern to write paper and book reviews. I do keep detailed study notes as documented code sketches, written and not yet returned to. I will add a slab ocean page to the wiki when I've got some worthwhile correct and documented code. If I finish any of my maths and climate literature surveys I'll just make a discussion page until they've been reviewed. Iiuc the wiki was specified to be of the same standard (and this case, presumably crowd-source reviewed) as the ncat lab with non of the gaps and errors of wikipedia. PS. I didn't receive any notification of posts #4-#7. There is also the mysterious "if you do not visit...you will not receive any further notifications" message?
  • 9.
    edited June 2012

    Jim wrote:

    My alibi is I don't have the critical domain or maths knowledge evern to write paper and book reviews. I do keep detailed study notes as documented code sketches, written and not yet returned to. I will add a slab ocean page to the wiki when I've got some worthwhile correct and documented code.

    It's important that you all put links to interesting resources on the Wiki, or else we won't find them later. You can also use the Wiki as a notebook, if you think there may be people with a similar interest as yours. That's how I use the page on C++: Of course all of this stuff is much better documente elsewhere, the page is simply a comprehensive list of tools and links and references for scientists who know that they would like to use C++, but don't know where and how to start.

    You don't have to write perfect encyclopedic articles. It's a Wiki, it is always work in progress, the main point is that it is useful. Primarily for you, then also for others.

    Comment Source:Jim wrote: <blockquote> <p> My alibi is I don't have the critical domain or maths knowledge evern to write paper and book reviews. I do keep detailed study notes as documented code sketches, written and not yet returned to. I will add a slab ocean page to the wiki when I've got some worthwhile correct and documented code. </p> </blockquote> It's important that you all put links to interesting resources on the Wiki, or else we won't find them later. You can also use the Wiki as a notebook, if you think there may be people with a similar interest as yours. That's how I use the page on C++: Of course all of this stuff is much better documente elsewhere, the page is simply a comprehensive list of tools and links and references for scientists who know that they would like to use C++, but don't know where and how to start. You don't have to write perfect encyclopedic articles. It's a Wiki, it is always work in progress, the main point is that it is useful. Primarily for you, then also for others.
  • 10.

    Thanks for all that Tim. I didn't have a feel for how to use the wiki. I'm trying to concentrate on producing literate code which explains what it's doing (which I'm sure all professional programmers now do) as that seems to be the greatest deficit. I will try and bundle up some of the references I've been learning from and start with just some pages of links.

    I've talked to Glyn Adgie about some possible pages, specifically on describing the various climate filters in the North book in signal processing circuit terms.

    Comment Source:Thanks for all that Tim. I didn't have a feel for how to use the wiki. I'm trying to concentrate on producing literate code which explains what it's doing (which I'm sure all professional programmers now do) as that seems to be the greatest deficit. I will try and bundle up some of the references I've been learning from and start with just some pages of links. I've talked to Glyn Adgie about some possible pages, specifically on describing the various climate filters in the North book in signal processing circuit terms.
  • 11.
    edited June 2012

    I've talked to Glyn Adgie about some possible pages, specifically on describing the various climate filters in the North book in signal processing circuit terms.

    The best way is to start with a stub. Just use the Template page and remember to start a thread here on the forum so that others who are also interested in the topic can join the discussion.

    Comment Source:<blockquote> <p> I've talked to Glyn Adgie about some possible pages, specifically on describing the various climate filters in the North book in signal processing circuit terms. </p> </blockquote> The best way is to start with a stub. Just use the [[Template page]] and remember to start a thread here on the forum so that others who are also interested in the topic can join the discussion.
  • 12.

    Hadn't heard of a Template Page. Glyn and I will start a stub.

    Comment Source:Hadn't heard of a Template Page. Glyn and I will start a stub.
  • 13.

    I think it's helpful to think about the wiki in terms of "notes". When you're reading a book/article, you often have slightly different goals than the author and hence you try to write notes that record what you care about and miss out the stuff that's less important for your goal. For example, in scholastic work it's considered very important to acknowledge any work in the area that's related and track the provenance of ideas (even in their initial flawed forms) and quite a chunk is devoted to that. Likewise it's not (currently) acceptable to simply provide a hyper-link to a description of the existing understanding. While in both those cases Azimuth aims to avoid "stating stuff that's wrong", AIUI it's perfectly permissible for that kind of info to simply not be there, and to focus on what may be useful to a researcher/worker coming to these issues for the first time.

    So I think a helpful attitude is "if I've made notes (literal or underlinings/marginalia in the original) that to my current understanding are correct, then I may as well clean the sentences a little and put them on the wiki". (Ie, if you think you genuinely don't have a clue what's going on then clearly don't do it, but otherwise put them up there.) I think John has often said in the past that it's easier to check, possibly correct and polish info that's gone up than to put it up in the first place so don't feel intimidated about putting "work in progress" up. (I know I haven't been!)

    Regarding "refactoring", I don't think it's that kind of problem as much as that we've accumulated lots of stuff without also buildign a hierarchy of detail so a new guy/gal can see which pages to read in what order to get a feeling for the stuff, and which pages are details only an active worker really needs to read. A while ago Curtis Faith was talking about how we could provide "overview pages" and additional structure on top of the basic wiki data.

    Comment Source:I think it's helpful to think about the wiki in terms of "notes". When you're reading a book/article, you often have slightly different goals than the author and hence you try to write notes that record what you care about and miss out the stuff that's less important for your goal. For example, in scholastic work it's considered very important to acknowledge any work in the area that's related and track the provenance of ideas (even in their initial flawed forms) and quite a chunk is devoted to that. Likewise it's not (currently) acceptable to simply provide a hyper-link to a description of the existing understanding. While in both those cases Azimuth aims to avoid "stating stuff that's wrong", AIUI it's perfectly permissible for that kind of info to simply not be there, and to focus on what may be useful to a researcher/worker coming to these issues for the first time. So I think a helpful attitude is "if I've made notes (literal or underlinings/marginalia in the original) that _to my current understanding_ are correct, then I may as well clean the sentences a little and put them on the wiki". (Ie, if you think you genuinely don't have a clue what's going on then clearly don't do it, but otherwise put them up there.) I think John has often said in the past that it's easier to check, possibly correct and polish info that's gone up than to put it up in the first place so don't feel intimidated about putting "work in progress" up. (I know I haven't been!) Regarding "refactoring", I don't think it's that kind of problem as much as that we've accumulated lots of stuff without also buildign a hierarchy of detail so a new guy/gal can see which pages to read in what order to get a feeling for the stuff, and which pages are details only an active worker really needs to read. A while ago [[Curtis Faith]] was talking about how we could provide "overview pages" and additional structure on top of the basic wiki data.
  • 14.

    Thanks David, Very helpful tips (is there a wiki posting protocol page?). I'm still looking for an appropriate place to repost (or link to if I can find out how to link to a post number) my questions from post #9?

    Comment Source:Thanks David, Very helpful tips (is there a wiki posting protocol page?). I'm still looking for an appropriate place to repost (or link to if I can find out how to link to a post number) my questions from post #9?
  • 15.

    Have you read the Help edit this wiki page linked to from the home page?

    Comment Source:Have you read the [[Help edit this wiki]] page linked to from the home page?
  • 16.

    It might be good later if you talk about the wiki pages. I think that when the Azimuth wiki grows, we will see when and how to make it. But I am also a fan of people trying new ideas, so add it and see how it works!

    contradicting replies but i would go for the latter.

    Comment Source:It might be good later if you talk about the wiki pages. I think that when the Azimuth wiki grows, we will see when and how to make it. But I am also a fan of people trying new ideas, so add it and see how it works! contradicting replies but i would go for the latter.
Sign In or Register to comment.