Hi Harikrishnan,

I recommend that you try using the 1.2a4 version of OOMMF with the module first. If the problems are still there, there is probably a compatibility issue with the module.

Best,

Kelvin FONG

LikeLike

]]>Hi Wen Zhao,

I recommend that you try using the 1.2a4 version of OOMMF with the module first. If the problems are still there, there is probably a compatibility issue with the module.

Best,

Kelvin FONG

LikeLike

]]>Dear Xuanyao,

I’m a student from China, who has been doing micromagnetic simulation projects based on Oommf. First of all, thanks so much for providing us with such a powerful extension module.

I’m writing to seek for assistance about this module, since I encountered the similar problem. The error message goes like “Unrecognized error thrown from inside “Oxs_Run” (Loop)”. Each time I have to reload the problem many times to get in. I’m not sure if there are any hidden problems here.

I would appreciate it if you could give me some support and advice! Looking forward to your reply!

Best regards,

Wen

LikeLike

]]>I am trying to do SOT simulation using this module.But actually, I am suffering from an error which given as follows.

Error thrown from inside “Oxs_Director::GetAllScalarOutputs” — Error evaluating output “Max dm/dt” of object “Oxs_SpinXferEvolve:” — Oxs_Ext ERROR in object Oxs_SpinXferEvolve:: Oxs_SpinXferEvolve::UpdateDerivedOutputs: Can’t derive Delta E from single state.

(Oxs_Output)

————–

LikeLike

]]>Sure.

kelvin.xy.fong@nus.edu.sg

LikeLike

]]>dHth is proportional to sqrt(dt), not Hth. Hth modeled using the stochastic Wiener process. Such that dHth=Hth-Hth0=N(0,1)*(sigma^2). Where N(0,1) is a random number with a normal distribution. Sigma^2 is proportional to sqrt(dt). In the Wiener process, increments are independent of the previous values. However, the motion itself is continuous.

If you don’t mind, can we discuss it in more detail via e-mail?

Thank you very much.

LikeLike

]]>Hth cannot be proportional to sqrt(dt)

If it is, integrating over time will yield the “noise power” of Hth having a time dependence. This cannot be true. The only possibility is for Hth to have Hth to be proportional to 1/sqrt(dt)

LikeLike

]]>Thank you very much for your kind reply.

Excuse me, but I meant thermal field Hth, not dm/dt. As far as I know, the thermal fluctuations are represented by field Hth. And Hth should be proportional to sqrt(dt), but not dm/dt.The term sqrt((2*alpha*kb*T*dt)/(mu0^2*gamma*Ms*V))=dHth; therefore, Hth=dHth+Hth0. After calculations of all Fields, they will be integrated.

In the absence of the external torques, the magnetization vector will move towards the new random position of the Heff=Hanis+Hdemag+Hexch+Hth.

LikeLike

]]>Hi Stanislav,

I think you might be confused between dm and dm/dt. The formulation says dm is proportional to sqrt(dt) and hence, dm/dt is proportional to 1/sqrt(dt). When the time integrator calculates m(t) from the initial value problem, it will multiply dm/dt with dt to solve for dm at every step, starting at m(t=0). The multiplication will result in dm being proportional to sqrt(dt) which is what you expect.

LikeLike

]]>Thank you very much for this module. Can I ask you a question regarding your examples?

I tested your examples, and I think your code seriously overestimates thermal fluctuations.

Even in the absence of external torque, it produces large deviations of the magnetization vector. I checked your source code, and it seems I know the problem.

In your code, you calculated the standard deviation of the distribution like sqrt((2*alpha*kb*T)/(mu0^2*gamma*Ms*V*dt)). In other words, it is proportional to 1/sqrt(dt). As far as I know, the conventional model of the Brownian motion (or Weiner process) tells that the standard deviation is proportional to sqrt(dt). Do you have any particular reason to add the dt into the denominator?

LikeLike

]]>