• To: mathgroup at smc.vnet.net
• Subject: [mg93496] Re: Storing and Loading Definitions / Emulating Associative Arrays
• From: David Bailey <dave at removedbailey.co.uk>
• Date: Wed, 12 Nov 2008 06:44:55 -0500 (EST)
• References: <gf3mfg\$ekk\$1@smc.vnet.net>

```Nikolaus Rath wrote:
> Hi,
>
> I have a routine that calculates some sequence of numbers. The
> calculation takes very long, so I would like to store the results in a
> file.
>
> What I would like to do is to pack the data, together with the
> parameters of its generation, into some sort of associative array.
> Since I couldn't find such a thing in Mathematica, I came up with the
> idea of using definitions instead:
>
>
> f[seed_, p_, n_] = (* Complicated computation *) {2,4,4};
> data[source] = "simulation";
> data[seed] = 42;
> data[parameters] = {12, 0};
> data[values] = f[data[seed], Sequence @@ data[parameters]]
>
> Now I would like to (somehow) store this set of definitions:
>
> Write[stream, something[data]]
>
> and later load it again, possibly with a different name
>
>
>
> But what is the function something[] that I have to use for this
> purpose? If I simply store the expression "data", then it is written
> literally, without the associated definitions. If I use
> Definition[data] then the output includes the name "data", so that I
> cannot easily associate it with "data2" when loading. Is there a
> simple way to achieve what I want?
>
> I guess a way would be to store definitions as an anonymous function:
>
>     data = Function[{x}, Switch[x,
>        source, "simulation",
>        seed, 42,
>        parameters, {12, 0},
>        values, f[42, 12, 0]]];
>
>     Write[stream, data]
>
> but then I still need a way to convert between the anonymous function
> and the definitions when saving and loading. I do not want to work
> with the anonymous function all the time, since it seems very
> difficult to modify any parameters after the initial definition:
>
>    (* do some stuff *)
>    data[comment] = "Analysis on 11/7 showed that this situation is unli=
> kely";
>
>    (* Write back the data with the comment *)
>
>
>
> Best,
>
>    -Nikolaus
>
> --
>  =C2=BBIt is not worth an intelligent man's time to be in the majority.
>   By definition, there are already enough people to do that.=C2=AB
>                                                          -J.H. Hardy
>
>   PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6  02CF A9AD B7F8 AE4E 425C
>
Before committing to a particular scheme, I think you need to be a bit
more specific about the size of your problem. For example, are you
talking about thousands of numbers, or hundreds of millions. In the
latter case, you may want to think of schemes that avoid loading all the
data into memory at once.

Save or DumpSave may well be all you need (DumpSave creates a binary
file which is much faster to read back in).

Remember that your final program may work fast enough without any
special tricks, or the bottleneck may turn out to other than where you
anticipate it to be.

David Bailey
http://www.dbaileyconsultancy.co.uk

```

• Prev by Date: Re: Re: Re: 6.0.3 Windows front end won't connect
• Next by Date: PDE problem with Wentzell-Robin boundary condition: wrong solution without any warning