TTDFT: A GPU accelerated Tucker tensor DFT code for large-scale Kohn-Sham DFT calculations

We present the Tucker tensor DFT (TTDFT) code which uses a tensor-structured algorithm with graphic processing unit (GPU) acceleration for conducting ground-state calculations on large-scale systems. The localized basis computed from an additive separable approximation to Kohn-Sham Hamiltonian. discrete problem is solved using Chebyshev filtered subspace iteration method that relies matrix-matrix multiplications of sparse symmetric Hamiltonian matrix and dense wavefunction matrix, expressed in basis. These multiplication operations, constitute most computationally intensive step solution procedure, are GPU accelerated providing ∼8-fold GPU-CPU speedup these operations largest systems studied. computational performance TTDFT presented benchmark studies aluminum nano-particles silicon quantum dots system sizes ranging up ∼7,000 atoms. Program Title: TTDFT: density functional theory CPC Library link program files: https://doi.org/10.17632/8dgmcs8ys2.1 Licensing provisions: LGPL Programming language: C/C++ External routines/libraries: TuckerMPI ( https://gitlab.com/tensors/TuckerMPI ), cuBLAS https://docs.nvidia.com/cuda/cublas/index.html cuSparse https://docs.nvidia.com/cuda/cusparse/index.html ALGLIB http://www.alglib.net/ Boost https://www.boost.org/ BLAS http://www.netlib.org/blas/ LAPACK http://www.netlib.org/lapack/ PETSc https://www.mcs.anl.gov/petsc SLEPc http://slepc.upv.es ) Nature problem: Real-space Solution method: real-space based techniques acceleration. Tensor-structured adopted computing basis, representing eigenfunctions further L 1 regularization improve sparsity efficiency parallel scalability proposed algorithm. (ChFSI) method. Additional comments including restrictions unusual features: works Troullier-Martin (TM) pseudopotentials Kleinman-Bylander form. current release supports only non-periodic local (LDA) exchange-correlation functional. This project GitHub via Git, free distributed version control software. archived at time submission this work can be found library through files DOI provided above. repository https://github.com/ttdftdev/ttdft_public .