forums.silverfrost.com

AsifArshid · Joined: 13 Mar 2013 Posts: 5

Dears,

Can anybody please confirm, if we can declare more then three dimensional arrays in Fortran 95 using Silver Frost (Plato) compiler ????

PaulLaidler · Posted: Fri Apr 05, 2013 8:24 am Post subject:

FTN95 conforms to the Fortran 95 Standard which (I think) allows up to 7 dimensions.

LitusSaxonicum · Posted: Fri Apr 05, 2013 12:46 pm Post subject:

Oh, that's a bit limiting. What about the 11-dimensions necessary for the M-theory version of string theory [url](http://en.wikipedia.org/wiki/String_theory). [/url]
Drat, we'll just have to use 11 separate arrays of the same length.

Eddie

davidb · Joined: 17 Jul 2009 Posts: 560 Location: UK

haha Eddie I think the cold in your shed must be getting to you!

Unfortunately 11 arrays each of size 100 (say) can't store all the elements (100 X 100 .... X 100) = 100**11.

If you have 100 elements in each of 7 dimensions and store double precision values thats about 8*100**7/10**9 Gigabytes or about 800 thousand Gigabytes. Rolling Eyes

You limit the size of problems you can "model" when you go anywhere near 7 dimensions (as Im sure you know Smile

). I think its a reasonable limit in the Standard. This going up to 15 in Fortran 2008 so we can worry about M theory then.

What I don't understand is why the OP needed to ask this when it would have been easier to compile two lines of code to find the limit.

LitusSaxonicum · Posted: Fri Apr 05, 2013 6:06 pm Post subject:

The shed is the only warm place - must be global warming run riot that has caused this winter.

I stand corrected - but it was not intended to be serious. So we could have around 15 REAL*8s in each of 7 dimensions with FTN95. That 64-bit version is going to be needed real soon now, with the 128-bit version soon after ...

Last time I was in PC World they had card readers, but my cards needed to be folded, spindled or mutilated to get them in the little slot, and I didn't think that was wise.

Eddie