Skip to content

Potts Model

Group (Subgroup)

DREAM3DReview (Coarsening)


This Filter simulates grain growth using the Potts model. The Potts model is a generalization of the Ising model to \f$ S \f$ states, or spins, on a regular lattice. This version of the Potts model functions in the context of Feature Ids on an Image Geometry, and thus will have the effect of coarsening Features; additionally, both 2D and 3D Image Geometries may be utilized. The Feature Ids are coarsened in place. The present implementation uses a Monte Carlo approach; the user may enter the desired number of Monte Carlo iterations to perform. For each Monte Carlo iteration, the algorithm proceeds as follows:

  1. Pick a Feature Id (spin) at random
  2. Compute the change in the Hamiltonian (\f$ \Delta E \f$) associated with flipping the chosen spin to another spin (candidate)
  3. If the energy change computed in step 2 is negative, then flip the spin to the candidate
  4. If the energy change is positive, then flip the spin if the following is true:

\f[ r < e^{\frac{-\Delta E}{kT}} \f]

where \f$ r \f$ is a random number on the interval \f$ [0, 1) \f$, \f$ \Delta E \f$ is the energy change computed in step 2, \f$ k \f$ is Boltzmann's constant (1.380 x 10-23 J/K), and \f$ T \f$ is a user-defined temperature value (in Kelvin).

  1. Increment the time step by \f$ 1 / N \f$, where \f$ N \f$ is the total number of lattice sites (i.e., the number of Cells in the Image Geometry, or the number of trues in the mask, if Use Mask is checked); this ensures that, on average, a spin flip is attempted once per lattice site for each iteration

The Hamiltonian used for this Potts model implementation is isotropic; thus, all boundaries are treated similarly. The major driving force that encourages flipping is the local neighborhood around each spin. Specifically, the energy change computed in step 2 is equivalent to the following:

\f[ \Delta E = \frac{1}{2} \sum_{n} \delta(s_{n}, s_{c}) - \delta(s_{n}, s_{i}) \f]

where \f$ s_{i} \f$ is the local spin attempting to flip, \f$ n \f$ are the valid neighbors of \f$ s_{i} \f$, \f$ s_{c} \f$ is the candidate spin to which \f$ s_{i} \f$ may flip, and \f$ \delta \f$ is the Kronecker delta function. The neighborhood used for each \f$ s_{i} \f$ is fully connected; in 2D, this equates to 8 neighbors, and in 3D this equates to 26 neighbors. If Periodic Boundaries is checked, then Cells on the surface will consider Cells on the opposite surface as valid neighbors. If Periodic Boundaries is not checked, the surface Cells will not consider Cells on the opposite surface as valid neighbors.

The user may specify a mask to ignore certain points from the simulation; lattice sites where the mask is false will never be selected to potentially flip, may not be selected as candidate spins for other lattice sites, and will not be considered valid neighbors when computing the energy change for the Hamiltonian. Masked points therefore act as sites where boundary motion is pinned; this phenomenon is known as Zener pinning. Note that spins with the Id value 0 (Feature Id = 0) also share this behavior (Feature Id 0 will act the same as if a mask value of false is at that position, even if no mask is being used).

This implementation of the Potts model uses several techniques to speed up the overall computation. First, after step 1, if the selected site's neighbors all have the same spin as the local site (i.e., the local site is completely within a grain), no spin flip will be attempted. Additionally, when selecting candidate spins, only spins that are among neighboring sites may be selected.


Name Type Description
Iterations int32_t Number of Monte Carlo time steps
Temperature double Temperature value to use when computing \f$ kT \f$, in Kelvin
Periodic Boundaries bool Whether to enforce periodic boundary conditions when computing neighbors
Use Mask bool Whether to use a boolean mask array to ignore certain points flagged as false from the algorithm

Required Geometry


Required Objects

Kind Default Name Type Component Dimensions Description
Cell Attribute Array FeatureIds int32_t (1) Specifies to which Cell each Feature belongs
Cell Attrubute Array Mask bool (1) Specifies if the point is to be counted in the algorithm, if Use Mask is checked

Created Objects


Example Pipelines

Please see the description file distributed with this plugin.

DREAM3D Mailing Lists

If you need more help with a filter, please consider asking your question on the DREAM3D Users mailing list: