Re: question about MDiG

[Joel Pitt]

Hi Miguel,

There is no inherent limit in MDiG, it doesn't store entire
distribution maps in memory, but rather saves them to disk, so memory
limits shouldn't be a huge issue. The r.mdig.kernel module generates
dispersal events for an entire map in memory, so that might be the
main memory consideration and depends very much on your simulation
parameters for generating events.

I've run 500m and 150m resolution simulations for the entirety of New
Zealand, and to do that for 100 or so replications would take a
weekend when I was doing my PhD back around 2005-2006. No doubt it's
faster now.

Having said that, I think other packages that simulate everything in
memory would be a lot faster. MDiG loads/saves a map for every step in
the simulation, because it's built to make use of GRASS modules for
simulation steps.

Note, there are few people who have used MDiG for various research
projects, but I have to warn you that I don't have huge amounts of
time to fix or change things anymore (and it's been a while since I
looked at the code!)

I hope that helps!

Joel

On 29 July 2014 04:20, Miguel Lurgi Rivera
<miguel.lu...@adelaide.edu.au> wrote:
>
> Hello Joel,
>
> My name is Miguel Lurgi. I am postdoc at the university of Adelaide. I am currently working on species distributions and range dynamics and I am trying to figure out the computational capacities of several packages that are able to simulate processes like this.
>
> I was wondering if you could give me some information about this regarding MDiG. For example, how many populations is the software able to simulate at the same time? How big is the landscape that can be simulated? Or any other information of this kind that you think would allow me to assess its capabilities would be greatly appreciated.
>
> Thanks in advance!
>
> Miguel.-

[Joel Pitt]

That's 100x100 cells, and MDIG should handle that easily!

Please report any issues you have here, I might be able to help:
https://github.com/ferrouswheel/mdig/issues

I use github for development, and if you get seriously into working
with mdig it might be worth making a fork of it and then you can merge
in your changes if they fix bugs or improve the software.

Cheers,

On 29 July 2014 07:38, Miguel Lurgi Rivera
<miguel.lu...@adelaide.edu.au> wrote:
>
> Hello Joel,
>
> Thanks very much for your reply. It seems like you developed quite a robust software. Saving stuff to disk is a clever way of saving memory space. I have used python in the past, so I would probably be ok using MDiG on my own without too much help.
>
> I am sorry I am not very familiar with the extension of NZ, so, I don’t have an idea of how many grid cells would there be with those resolutions. Would you say it would be around 10,000 populations?
>
> Thanks!
>
> Miguel.-