Hi,
If I have closed and open the software file and again run the optimization step the particle positions get a bit changed even slightly the eigen value also changes. Can you please let us know why it is happening?
Generally speaking, re-running the optimization on the same machine results in the exact same particle positions. The random number generators are seeded deterministically. As far as I’ve tested, this has been the case, but perhaps there is a combination of parameters that I have not checked.
Are you re-running it on the same machine?
Yes I am running on same machine.
If I close and open again the project file and re run the optimization step the eigen values get changed a bit.
Also, does the particle initialization is only dependent on the machine?
As I run the same data on software and python the eigen values were again a bit different.
As far as I’ve tested, I always get the exact same particle positions when I re-run a project on the same system. Perhaps if you could share a small sample that reproduces this, I can track down the source of the change.
The platform/machine differences arise from different floating point behavior from different compilers/cpu architectures. Some computations can be reordered by the compiler or use SIMD instructions with differing precision. Over iterations, the differences can start to add up. Initially in the first couple of iterations, the results are generally identical across all platforms.