Options

climateprediction.net

Today I read about climateprediction.net in the book "A Vast Machine", which is a project that uses idle computers to compute versions of GCMs for the timespan of the next 100 years, much like the SETI project did (with pattern matching for background radiation to find aliens, or something similar).

I always suspected that something like this existed, and it seems to be active still. But of course we should calculate how much CO2 is generated by computing a GCM instead letting a PC run idle :-)

Comments

  • 1.

    Cool!

    Comment Source:Cool!
  • 2.
    edited May 2011

    Is that your new management technique to provide fast supportive feedback?

    :-)

    Maybe I'll try to run one of the GCMs on my Laptop this weekend, if it works fine, you could blog about it :-) It is about a very important topic: Running the same GCM with slightly different parameters, to see how sensitive it is to changes. In simple contexts, it is possible to do this with pen and paper by calculating derivatives for example, but no longer for really complex software.

    Comment Source:Is that your new management technique to provide fast supportive feedback? :-) Maybe I'll try to run one of the GCMs on my Laptop this weekend, if it works fine, you could blog about it :-) It is about a very important topic: Running the same GCM with slightly different parameters, to see how sensitive it is to changes. In simple contexts, it is possible to do this with pen and paper by calculating derivatives for example, but no longer for really complex software.
  • 3.

    if it works fine, you could blog about it :-)

    You could write something about it, and I'd polish the English if necessary and stick it on the blog!

    It is about a very important topic: Running the same GCM with slightly different parameters, to see how sensitive it is to changes.

    Yes, that's important.

    Comment Source:> if it works fine, you could blog about it :-) You could write something about it, and I'd polish the English if necessary and stick it on the blog! > It is about a very important topic: Running the same GCM with slightly different parameters, to see how sensitive it is to changes. Yes, that's important.
  • 4.

    I could, should I start a dedicated page for it?

    It's hard to figure out what exactly I should write about. There are a lot of misconceptions about climate models that shouldn't go into such a blog post, I guess, like "nonlinear systems cannot be computed numerically", "climate is chaotic and cannot be modelled" etc. besides basic misconceptions about the physics (there is no greenhouse effect because QM is wrong or because it would violate the second law of thermodynamics).

    Instead, I could start by comparing climateprediction.net to the similar SETI project (which I don't know much about, ughhhh), and explain some basic concepts from software engineering like: What does it mean to test a model, how does it relate to stakeholder analysis, and what is verification and validation? Even the author of "A Vast Machine" gets that wrong. And why it is necessary to have ensemble runs to test the stability of model results with respect to parameter variation, the variations in model predictions (of temperature increase) from the IPPC report (which is between 1,5 and ca. 6, 7, or 8 degrees Celsius, I'd have to look that up).

    And why the IPPC estimate has a considerably lower bound than the model results have.

    Comment Source:I could, should I start a dedicated page for it? It's hard to figure out what exactly I should write about. There are a lot of misconceptions about climate models that shouldn't go into such a blog post, I guess, like "nonlinear systems cannot be computed numerically", "climate is chaotic and cannot be modelled" etc. besides basic misconceptions about the physics (there is no greenhouse effect because QM is wrong or because it would violate the second law of thermodynamics). Instead, I could start by comparing climateprediction.net to the similar SETI project (which I don't know much about, ughhhh), and explain some basic concepts from software engineering like: What does it mean to test a model, how does it relate to stakeholder analysis, and what is verification and validation? Even the author of "A Vast Machine" gets that wrong. And why it is necessary to have ensemble runs to test the stability of model results with respect to parameter variation, the variations in model predictions (of temperature increase) from the IPPC report (which is between 1,5 and ca. 6, 7, or 8 degrees Celsius, I'd have to look that up). And why the IPPC estimate has a considerably lower bound than the model results have.
  • 5.
    edited June 2011

    I could, should I start a dedicated page for it?

    Sure, start here:

    Blog articles in progress

    and do what everyone else there does.

    It's hard to figure out what exactly I should write about.

    Rule 1: don't worry about it too much.

    Rule 2: write about something you know about.

    Rule 3: write about something interesting.

    Rule 4: write about something that excites you right now, so you'll be motivated to get the job done.

    Rule 5: don't make it too complicated.

    Rule 6: but don't worry too much.

    Instead, I could start by comparing climateprediction.net to the similar SETI project (which I don't know much about, ughhhh)...

    Don't write about something you'd need to learn a lot about... unless you want to use the writing as a way to learn more! I often write This Week's Finds while I'm learning stuff; I enjoy that approach. But I'm sure you know lots of interesting things already. You listed a bunch. I suggest picking one and writing about that. Then, if it's fun, try another.

    If you let me, I can use my patented tricks to make your writing more enjoyable to read.

    Comment Source:> I could, should I start a dedicated page for it? Sure, start here: [[Blog articles in progress]] and do what everyone else there does. > It's hard to figure out what exactly I should write about. Rule 1: don't worry about it too much. Rule 2: write about something you know about. Rule 3: write about something interesting. Rule 4: write about something that excites you right now, so you'll be motivated to get the job done. Rule 5: don't make it too complicated. Rule 6: but don't worry too much. > Instead, I could start by comparing climateprediction.net to the similar SETI project (which I don't know much about, ughhhh)... Don't write about something you'd need to learn a lot about... unless you want to use the writing as a way to learn more! I often write _This Week's Finds_ while I'm learning stuff; I enjoy that approach. But I'm sure you know lots of interesting things already. You listed a bunch. I suggest picking one and writing about that. Then, if it's fun, try another. If you let me, I can use my patented tricks to make your writing more enjoyable to read.
  • 6.

    Ok, I'll give it a try. But I'll make climateprediction.net only one part of the whole story.

    Comment Source:Ok, I'll give it a try. But I'll make climateprediction.net only one part of the whole story.
  • 7.

    Installed the BOINC-software and registered project climateprediction.net, but wasn't assigned any task to compute, for reasons unknown ...

    Comment Source:Installed the BOINC-software and registered project climateprediction.net, but wasn't assigned any task to compute, for reasons unknown ...
  • 8.

    I used the predecessor to look for Mersenne primes in the GIMP project (no not that one), and my LSA used to look for aliens in SETI@home with the same program. I also used climateprediction.net for a while while. But its good that they are still running. I didnt have any problems running it but are you firewalled ? I think this is handled by BOINC so it shouldn't be a problem

    Comment Source:I used the predecessor to look for Mersenne primes in the GIMP project (no not that one), and my LSA used to look for aliens in SETI@home with the same program. I also used climateprediction.net for a while while. But its good that they are still running. I didnt have any problems running it but are you firewalled ? I think this is handled by BOINC so it shouldn't be a problem
  • 9.

    Meanwhile my local BOINC installation has downloaded a model and is crunching away. I could record a video of the online visualization - is it possible to post a video on the blog? How about the wiki?

    Comment Source:Meanwhile my local BOINC installation has downloaded a model and is crunching away. I could record a video of the online visualization - is it possible to post a video on the blog? How about the wiki?
  • 10.
    I used to run that model via BOINC. On one hand it only incurred a ~10W penalty versus idle, but on the other hand the processor (VIA C7) in my box was so slow it took over half a year for it to finish a task.
    Comment Source:I used to run that model via BOINC. On one hand it only incurred a ~10W penalty versus idle, but on the other hand the processor (VIA C7) in my box was so slow it took over half a year for it to finish a task.
  • 11.
    edited June 2011

    Well, it seems that everybody and his cat has known about this, except me who discovered it only recently :-)

    My laptop is running a coupled atmospheric-ocean GCM in a timespan from 1940 to 1980 with a timestep somewhere around 1 hour, which makes me wonder if it will ever finish.

    Comment Source:Well, it seems that everybody and his cat has known about this, except me who discovered it only recently :-) My laptop is running a coupled atmospheric-ocean GCM in a timespan from 1940 to 1980 with a timestep somewhere around 1 hour, which makes me wonder if it will ever finish.
  • 12.
    edited July 2011

    Tim wrote:

    is it possible to post a video on the blog? How about the wiki?

    I succeeded in linking to a YouTube video on my Tsunami blog entry.

    I can put all types of files on UC Riverside website, including video files in various formats. As long as standard browsers are smart enough to play your video file, you could just give me the file, I can put it on my website, and we can link to it from anywhere.

    Comment Source:Tim wrote: > is it possible to post a video on the blog? How about the wiki? I succeeded in linking to a YouTube video on my [Tsunami](http://johncarlosbaez.wordpress.com/2011/03/12/tsunami/) blog entry. I can put all types of files on UC Riverside website, including video files in various formats. As long as standard browsers are smart enough to play your video file, you could just give me the file, I can put it on my website, and we can link to it from anywhere.
Sign In or Register to comment.