Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Technical milestone reached: global earth system simulations with 1.2 km resoln

mpimet.mpg.de

Post reply on HN
Print view View original

71–80 of 105 posts

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#71

Post by sp332 »

counters wrote: post #17
OrvalWintermute wrote: post #11

This sounds alot less than Nvidia's FourCastNet? https://resources.nvidia.com/en-us-fleet-command/watch-27?xs... How do these two compare?

They're complete different classes of models. The modeling system in the linked article is a high-fidelity numerical simulation of the coupled Earth system. It's a giant PDE solver for Navier-Stokes applied to the Earth's atmospheres and oceans, coupled together with a great deal of additional physics simulation. The intent is to reproduce, in simulation, the Earth's atmospheric and oceans with the highest fidelity.…

You realize "facsimile" means "copy", right? You didn't explain how they are different.

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#72

Post by jokabrink »

adamius wrote: post #40

How does such a simulation work? Eg how do hou predict the temperature at x,y? Is it ground type, water, sand?Altitude? Neighboring values thereof? What are the inputs? Do you give it a starting point and apply it to a bunch of elements like some giant automata like game of life? Some kind of finite element analysis thing? So many questions.

Such simulations usually consist of three major systems (atmosphere, land, and ocean) that are coupled together at their geometric boundaries by a coupler that 'communicates' values like temperature from one domain to another. The coupler is needed because of different grid geometries, time step size differences and other aspects.

You initialize the system at some known state (I.e. set the temperature, pressure, etc. at all grid points to real world measurements) and then integrate a complex differential equation for the next time step and so forth. So it is not like a automaton. Finite elements analysis comes closer, but I think they use a different scheme like finite volume methods.

A lot of insight can be gained by [this](https://pure.mpg.de/rest/items/item_3379183/component/file_3...) paper. The first 10 pages should give you a rough overview.

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#74

Post by ordu »

adamius wrote: post #40

How does such a simulation work? Eg how do hou predict the temperature at x,y? Is it ground type, water, sand?Altitude? Neighboring values thereof? What are the inputs? Do you give it a starting point and apply it to a bunch of elements like some giant automata like game of life? Some kind of finite element analysis thing? So many questions.

> Do you give it a starting point and apply it to a bunch of elements like some giant automata like game of life?

Roughly speaking yes. Divide all into a grid of cells. Model a cell state with a bunch of numbers, apply some rules to update cell state with neighbors. The trick is to figure out rules of updating state. One needs to write differential equations first, incorporating all relevant physical processes into them, and to transform equations into those rules of updating, which will be a way more complex than with Game of Life.

Though it may be even more complex, like different time steps at different time-points, or changing a grid of cells to increase details in some areas where much is going on by slowing down simulation. Most of complications are due to a limited abilities of our computers: the idea to get more precision by calculating less.

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#77

Post by rongopo »

dunefox wrote: post #70
rongopo wrote: post #67

Earlier quoted context omitted.

If the margins are so tight that 1 day can only produce 2.5 days of simulation, why to lose even a small margin with Julia?

How do you know they would lose a significant margin with Julia?

There were talks about it in the Climate modelling arena, and around a 10% slower was mentioned in some discussions.

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#78

Post by adrianN »

Maursault wrote: post #68
adrianN wrote: post #63

Earlier quoted context omitted.

Maybe read https://arxiv.org/pdf/quant-ph/9908043.pdf before you think about computers operating on 2*140 objects and try to be less aggressive in your comments.

Are numbers not objects? Is scientific notation not a way of expressing numbers that are too large or too small to be conceived or expressed in decimal form? Does a large database keep every object and every bit of data stored in it in RAM? How is it possible a fractal can be rendered in parts of the whole? Even older computers ordinarily can operate on 2^140 objects. user@decadeoldcomputer:~$ echo 2^140 |bc 13937965…

No computer can operate on 2**140 objects in any meaningful way, because no computer can even remember whether it's already done with one of the objects or not. Your example operates on a single object, a number.

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#79

Post by origin_path »

rongopo wrote: post #66
hackandthink wrote: post #54

Being curious about the implementation language. Viewing the code is not easy. If you guess Fortran, you might be right: (different ICON Project) "The infrastructure, ICON-Land, for this ICON-A land component has been newly designed in a Fortran2008 object-oriented, modular, and flexible way." https://mpimet.mpg.de/fileadmin/publikationen/Reports/WEB_Bz... Fortran alive and kicking: https://developer.nvidia.com/cuda-…

Most of what they do in the Climate simulation arena uses fortran, this case not being an exception, reason being other language would be less efficient --- and thus more climate endangering emissions would be produced. Plotting and data wrangling is another story.

They don't use FORTRAN because of efficiency or emissions reductions lol, they use FORTRAN because they always did and the sort of scientists who do modelling rarely see any reason to upgrade their skills or tools.

Programming is something scientists tend to study only as far as needed to get results that look right. This is how the most influential COVID model ended up being a 15,000 line student-quality C program with hundreds of single-letter name global variables.

Re: Technical milestone reached: global earth system simulations with 1.2 km resoln

#80

Post by marshallward »

origin_path wrote: post #79
rongopo wrote: post #66

Earlier quoted context omitted.

Most of what they do in the Climate simulation arena uses fortran, this case not being an exception, reason being other language would be less efficient --- and thus more climate endangering emissions would be produced. Plotting and data wrangling is another story.

They don't use FORTRAN because of efficiency or emissions reductions lol, they use FORTRAN because they always did and the sort of scientists who do modelling rarely see any reason to upgrade their skills or tools. Programming is something scientists tend to study only as far as needed to get results that look right. This is how the most influential COVID model ended up being a 15,000 line student-quality C program w…

Fortran is no longer in ALL CAPS.
Post reply on HN