Computational concerns

 

 

 

Sim. Simulation Nodes Elements Element CPU time

# name size

1 hex_Ia1m1 637 2160 3.33 nm 3d 11h 19m

2 hex_Ia1m2 2541 10000 2 nm 8d  2h 27m

3 hex_Ia2 2541 10000 2 nm 8d  3h 32m

4 hex_Ia3 2541 10000 2 nm 11d 13h 47m

5 hex_Hm1 1274 4320 3.33 nm 8d  2h 27m

6 hex_Hm2 5082 20000 2 nm 23d  3h 33m

7 hex_T 5082 20000 2 nm 34d 17h 20m

The coarse meshes of hex_Ia1m1 and hex_Hm1 produced results comparable to those of the fine meshes (remanence and coercivity differ only a few percent; cf. table 4 on page ). The fine meshes, however, have four times as many nodes and five times as many elements as the coarse mesh. Thus, the CPU time from saturation until demagnetization increases by a factor of three. To calculate remanence and coercivity it is therefore sufficient to choose an element size which is two times larger than the exchange length of the material (cf. table 3 on page ).

  

Figure 31: Simulation data of hex_Ia1m1

  

Figure 32: Simulation data of hex_Ia1m2

  

Figure 33: Simulation data of hex_Ia2

  

Figure 34: Simulation data of hex_Ia3

  

Figure 35: Simulation data of hex_Hm1

  

Figure 36: Simulation data of hex_Hm2

  

Figure 37: Simulation data of hex_T