Reloading OVFs ("load" method for A, K's, Msat..)

[Fixtandantilux]

Is there any way to reload a previous state in order to re-launch a new simulation with new parameters from a previously saved starting point?

I'm trying to set a granular system, generated by ext_makegrains and I want to modify several aspects of the simulation: In order to compare results I would need to use the same starting point but, as far as I see in the API, you can only feed mumax with regions and m (besides excitation), but there is no "loadfile" method for A, Msat or K1 (magnitude nor vector). Could it be easily possible to add the load method in order to load these parameters?

When you define a grain-dependent material it's important to be able to fully save the initial material configuration (Msat, anisotropy, exchange... not only the magnetization state). In the same way, it would be useful in order to restart/repeat a simulation in the same way that OOMMF does, by adding a lot of parameters in the same OVF file (instead of saving only one magnitude) or by adding parameters to the simulation by loading several OVF files.

[barnex]

A "save all"/"load all" function might be nice, but for the moment I don't have the time to implement and (especially) maintain that.

But you choose the material parameters yourself, so I don't really see the need to save/restore them. An easy way to load the same list of parameters in many files is with include("somefile"). Then in somefile you type the code to define the parameters and you can use it over and over. Would that mostly solve your problem?

On Apr 30, 2014 4:53 PM, "Gabriel Rodriguez-Rodriguez" < notifications@github.comjavascript:_e(%7B%7D,'cvml','notifications@github.com');> wrote:

Is there any way to reload a previous state in order to re-launch a new simulation with new parameters from a previously saved starting point?

I'm trying to set a granular system, generated by ext_makegrains and I want to modify several aspects of the simulation: In order to compare results I would need to use the same starting point but, as far as I see in the API, you can only feed mumax with regions and m (besides excitation), but there is no "loadfile" method for A, Msat or K1 (magnitude nor vector). Could it be easily possible to add the load method in order to load these parameters?

When you define a grain-dependent material it's important to be able to fully save the initial material configuration (Msat, anisotropy, exchange... not only the magnetization state). In the same way, it would be useful in order to restart/repeat a simulation in the same way that OOMMF does, by adding a lot of parameters in the same OVF file (instead of saving only one magnitude) or by adding parameters to the simulation by loading several OVF files.

— Reply to this email directly or view it on GitHubhttps://github.com/mumax/3/issues/15 .

[Fixtandantilux]

The idea is to get a set of randomized grains with (also) randomly distributed parameters (msat, anisU, K1,...) and get this arrangement for several simulations. Maybe we could get a list of grain-msat-k1-anisU values and import them for each new run so, in this case, it won't be a code to define the parameters, but a list of grain-parameters list.

[barnex]

If you use the same code to generate the grains+random parameters, you will get exactly the same result in each run. All randomness is perfectly reproducible and can be controlled by setting the respective seeds. So there should be no need to save/load. -Arne;

On Mon, May 5, 2014 at 9:54 AM, Gabriel Rodriguez-Rodriguez notifications@github.com wrote:

The idea is to get a set of randomized grains with (also) randomly distributed parameters (msat, anisU, K1,...) and get this arrangement for several simulations. Maybe we could get a list of grain-msat-k1-anisU values and import them for each new run so, in this case, it won't be a code to define the parameters, but a list of grain-parameters list.

— Reply to this email directly or view it on GitHub.

[Fixtandantilux]

Obviously! I'm used to get the random seed from timeclock just in order to avoid exactly this issue! Thanks a lot for the tip. G.- P.S.: Anyway, It would be awesome to be able to relaunch simulations from a previously saved state by loading an all-parameter OVFs