The Supercomputer MACH-2: Use Cases

[Back to Use Cases]

Use Case: Diagonalization of Very Large Sparse Matrices

Involved Scientists Description of the Application

Quantum spin liquids (QSLs) are exotic phases of matter that have been extensively studied in the last decades due to their interesting magnetic properties and possible technological applications, e.g. in quantum computers. The identification of QSLs is, however, a very challenging problem since exact solutions for strongly correlated condensed matter systems rarely exist. Numerical methods are often the only option left.

Figure: Ground-state energies as a function of the coupling constants and phase borders of a SU(3) symmetric spin model on the triangular lattice. (CC BY 4.0 attributed to [1])
We studied SU(N) symmetric spin systems on regular lattices. To obtain the lowest lying energy eigenstates and eigenvalues we used an iterative eigenvalue solver algorithm. The Hamiltonians for such systems are huge sparse matrices. The size of this matrices as well as the method used require a huge amount of memory. Furthermore, the structure of the problem under investigation leads to massive communication between the threads, which makes an implementation on a distributed memory system not feasible. Using the Mach2 Supercomputer with its extensive amount of shared memory allowed us to identify a QSL phase in a SU(3) symmetric spin model on the triangular lattice.

References

  1. Boos, C., Ganahl, C. J., Lajkó, M., Nataf, P., Läuchli, A. M., Penc, K., ... & Mila, F. (2020). Time-reversal symmetry breaking Abelian chiral spin liquid in Mott phases of three-component fermions on the triangular lattice. Physical Review Research, 2(2), 023098. DOI: 10.1103/PhysRevResearch.2.023098


JKU Scientific Computing Administration