OMEN

Official Resources

• Homepage: https://engineering.purdue.edu/gekcogrp/software-projects/omen/
• Repository: https://github.com/spcl/dace-omen (DaCe variant)
• License: Academic / Open Source variants exist

Overview

OMEN is a massively parallel, multidimensional quantum transport simulator optimized for high-performance computing (HPC) environments. It focuses on the simulation of post-CMOS nanodevices (nanowires, TFETs) using the semi-empirical tight-binding method ($sp^3d^5s^*$) coupled with the Non-Equilibrium Green's Function (NEGF) or Wave Function (WF) formalisms. Known for its extreme scalability, OMEN has been used to simulate realistic devices with tens of thousands of atoms including electron-phonon scattering.

Scientific domain: Quantum Transport, Supercomputing, Device Physics Target user community: HPC specialists and device physicists studying dissipative transport

Theoretical Methods

• NEGF: Full calculating of retarded/lesser Green's functions with self-energies for scattering.
• Wave Function: Faster "transmitting boundary" method for ballistic transport.
• Scattering: Self-consistent Born approximation for acoustic/optical phonons and surface roughness.
• Basis: Localized tight-binding orbitals.

Capabilities

• Device Simulation:
  • Nanowire Field-Effect Transistors (GAA-NWFETs).
  • Tunnel FETs (TFETs) with band-to-band tunneling.
  • 2D material logic devices.
• Physics:
  • Dissipative transport (Joule heating).
  • Band structure effects in confined geometries.

Key Strengths

• HPC Performance: A Gordon Bell Prize finalist code, capable of scaling to >200,000 cores on machines like Titan/Summit.
• Scattering: One of the few codes that can handle full 3D atomistic NEGF with scattering (albeit at high cost).
• Data Centric: Recent versions (DaCe-OMEN) explore new programming paradigms for efficiency.

Inputs & Outputs

• Inputs: Atomistic structure maps, material parameter files, bias commands.
• Outputs: Current, charge density, energy-resolved current spectra.

Interfaces & Ecosystem

• CP2K: Integration allows for DFT-based transport (Hamiltonian blocks from CP2K fed into OMEN).
• libNEGF: Can use external solver libraries.

Performance Characteristics

• Cost: Extremely high for NEGF+Scattering.
• Optimization: Uses SSE/AVX intrinsics, GPU acceleration, and advanced MPI/OpenMP tiling.

Comparison with Other Codes

• vs. NEMO5: OMEN is the specialized transport engine; NEMO5 wraps it in a larger multiphysics framework. OMEN is often used for pure transport performance benchmarks.
• vs. Kwant: OMEN is a fully-fledged device simulator with physics-specific scattering models; Kwant is a Hamiltonian solver toolkit.

Community and Support

• Development: ETH Zurich (Mathieu Luisier) and Purdue.
• Source: GitHub (DaCe-OMEN) / NanoHub.

Verification & Sources

• Primary Publication: M. Luisier et al., SC10 (2010).
• Verification status: ✅ VERIFIED
  • Validated against experimental data for nanowire transistors.