HERMES++: Advancing Modular Research
- HERMES++ is a polysemous research label representing diverse, modular frameworks that extend baseline HERMES systems with enhanced automation and interoperability.
- It spans multiple domains by introducing innovations such as AI-driven orchestration, hybrid Monte Carlo-analytical workflows, and enriched metadata publication processes.
- The evolving architectures emphasize improved performance, security, and scalability while addressing domain-specific challenges and open research problems.
Searching arXiv for the cited HERMES and HERMES++-related papers to ground the article in current records. HERMES++ denotes, in current arXiv usage, not a single canonical system but a cluster of advanced frameworks, hypothetical extensions, and C++-based evolutions associated with projects named HERMES. Across the supplied literature, the label is used in at least five distinct research settings: a cloud–edge–IoT computing continuum architecture; astroparticle and detector simulation; Galactic multi-messenger sky modeling; multi-component plasma simulation on BOUT++; and automated research-software publication workflows. This suggests that “HERMES++” functions less as a standardized artifact name than as a recurring marker for a more extensible, automated, or technically enriched successor state of an existing HERMES system (Dehury et al., 9 Dec 2025).
1. Terminological scope and domain distribution
The term appears with different semantics depending on field, software lineage, and implementation context. In some cases it names a hypothetical next step; in others it is an explicit analogy to a C++ reimplementation; in still others it is a convenient designation for an advanced toolkit layered on an existing HERMES project.
| Domain | HERMES basis | HERMES++ characterization |
|---|---|---|
| Computing continuum | HERMES for monetization, orchestration, and semantics | hypothetical evolution with more automation, richer economics, stronger privacy, and more expressive semantics |
| UHECR propagation | Monte Carlo code for ultra-high-energy nuclei | hypothetical extension with updated backgrounds, magnetic fields, interactions, and computational tooling |
| Satellite activation background | HERMES-SP detector background toolkit | hybrid Monte Carlo plus analytical framework described as an advanced toolkit |
| Galactic emissions | public multi-messenger sky-map code | modular, extensible C++ engine with Python interface, framed as what HERMES++ should do |
| Plasma simulation | Hermes-3 on BOUT++ | often written as HERMES++ in analogy with BOUT++ |
| Software publication | HERMES metadata-publication workflow | conceptual next-generation workflow with richer metadata intelligence and curation |
A common misconception is that HERMES++ refers to one research platform spanning these domains. The supplied literature does not support that reading. Instead, it documents several unrelated systems that share the HERMES name and, in multiple cases, use “++” to indicate extension, modularization, or a more capable implementation (Domenico et al., 2013).
2. Computing-continuum HERMES++ as an architectural evolution
In the computing-continuum literature, HERMES is defined as “Heterogeneous Computing Continuum with Resource Monetization, Orchestration, and Semantic,” a conceptual framework for the cloud–edge–IoT continuum that tightly couples a monetization bridge, an orchestration bridge, and a semantic interconnect. A hypothetical HERMES++ is explicitly described as preserving these pillars while deepening them with more automation, richer economics, stronger privacy, and more expressive semantics (Dehury et al., 9 Dec 2025).
The underlying HERMES architecture treats the continuum as an open, multi-stakeholder ecosystem rather than a single-vendor stack. Its monetization bridge is a decentralized marketplace in which compute resources, storage, network resources, data streams, AI models, analytics, and higher-level services can be advertised, discovered, priced, and billed through blockchain and smart contracts. Its orchestration bridge is heterogeneity-aware and WebAssembly-based, with hierarchical orchestrators and lightweight agents spanning cloud, fog, edge, and IoT tiers. Its semantic interconnect provides shared ontologies, a semantic registry, a knowledge base, and semantic brokers for interoperability across devices, services, and administrative domains.
A central construct is the lens: a pluggable data-processing module inserted into data paths, for example as a privacy lens, AI inference lens, or aggregation lens. Lenses are cataloged in a lens library, can be offered by third parties, and are deployed, reconfigured, removed, or migrated dynamically by orchestrators. In HERMES++, this mechanism is projected forward toward AI-driven orchestration, richer game-theoretic pricing and bidding, privacy-preserving monetization, automated ontology alignment and evolution, self-sovereign identity for devices, and privacy-preserving analytics such as federated learning and secure multi-party computation.
Security and governance are first-class concerns. The framework is guided by the DIRECTS threat model—Denial-of-Service, Integrity, Repudiation, Elevation of Privilege, Confidentiality, Trust, and Spoofing—and combines ledger-based auditability, cryptographic verification, signed WASM modules, access-control policies, and smart-contract enforcement. The paper is architectural rather than benchmark-driven: it does not report numerical latency or throughput measurements, and the proposed HERMES++ remains a conceptual extension rather than a published implementation (Dehury et al., 9 Dec 2025).
3. Astroparticle and detector-simulation usages
In astroparticle physics, HERMES is a Monte Carlo code for the propagation of ultra-high-energy nuclei through an expanding and magnetized Universe. It covers the energy range , models adiabatic losses, pair production, photo-pion production, and photo-disintegration, and tracks interactions with the CMB, the cosmic infrared/optical background, and the universal radio background. Its propagation formalism is written in redshift as
with tunable cosmological parameters in a general FRW framework (Domenico et al., 2013).
That HERMES is modular, with tunable cosmology, radiation fields, magnetic-field models, and source distributions. It produces spectra at Earth, GZK horizons, arrival-direction skymaps including Galactic and extragalactic magnetic deflections, and secondary photons and neutrinos, with electromagnetic cascades delegated to EleCa. The corresponding HERMES++ is not an implemented release but a natural extension identified in the details: updated EBL/CIOB and URB models, more sophisticated GMF and structured EGMF prescriptions, improved photo-disintegration cross sections, tighter coupling to air-shower and detector simulations, enhanced Monte Carlo efficiency and parallelization, and broader benchmarking against CRPropa and related codes (Domenico et al., 2013).
A different HERMES++ appears in high-energy detector background simulation for the HERMES-SP CubeSat program. There, the software toolkit combines direct Monte Carlo and analytical calculations to simulate proton-induced activation background and is reported to be two orders of magnitude faster and more precise than a direct Monte Carlo simulation. Its three-stage workflow is: Monte Carlo production of primary unstable isotopes via Geant4, analytical computation of decay chains and time-dependent activities through Bateman-equation treatment, and detector-response simulation for the decays of activated isotopes (Galgoczi et al., 2021).
The framework uses 16 monochromatic proton energies from 4 MeV to 700 MeV, AP9 proton spectra for a worst-case 600 km, inclination orbit, and Geant4 radioactive-decay data with explicit treatment of excitation level and floating levels. More than 1.5 million decay chains are built across energies and volumes, while decay branches with branching ratio below 5% are discarded and isotopes with are removed to avoid numerical issues. For half a year in orbit, the total activity of the entire satellite is given as 851 Bq, and for the GAGG scintillator with a 20 keV threshold the expected activation-induced background is counts per second after six months, while the SDD contribution is negligible (Galgoczi et al., 2021).
4. Galactic multi-messenger and plasma-simulation interpretations
In Galactic astrophysics, HERMES—“High-Energy Radiative MESsengers”—is a public numerical framework for simulating diffuse, non-thermal, multi-messenger and multi-wavelength emission of the Milky Way. It computes radio–microwave synchrotron and free–free emission, gamma rays from production, inverse Compton scattering and bremsstrahlung, neutrinos from charged-pion decay, Faraday rotation measures, dispersion measures, and dark-matter-induced prompt emission. Its core observable is the line-of-sight integral
with taken as energy or frequency depending on channel (Dundovic et al., 2021).
That code is written in C++11 with a Python interface via pybind11 and is strongly modular. Skymap classes hold HEALPix maps, integrator classes implement process-specific line-of-sight calculations, astrophysical component classes represent CR distributions, magnetic fields, gas, ISRF, electron density, cross sections, and dark-matter profiles, and caching plus std::thread parallelization reduce the cost of nested integrals. The details explicitly state that, in spirit, “HERMES++” is precisely such a modular, extensible, high-performance C++ engine with a rich Python interface (Dundovic et al., 2021).
In plasma physics, HERMES++ is most directly identified with Hermes-3, often written as HERMES++ in analogy with BOUT++. Hermes-3 is a drift-reduced, multi-fluid plasma code built on the BOUT++ framework for 1D, 2D, or 3D simulations, in steady transport or time-dependent turbulence regimes, with an arbitrary number of ion and neutral species. It uses a component–state architecture in which a nested state dictionary stores physical fields and modular components implement continuity, momentum, pressure evolution, collisions, reactions, sheath boundary conditions, recycling, electron force balance, anomalous diffusion, and other closures (Dudson et al., 2023).
The demonstrated physics spans 1D divertor plasmas with deuterium and all charge states of neon, 2D axisymmetric single-null X-point transport, and 2D blob dynamics with a vorticity–potential formulation and sheath closure. The software is configured entirely through text input files rather than separate executables. Verification includes Method of Manufactured Solutions with second-order convergence, Sod shock-tube tests, and global power and particle balance studies. The limitations are also clear: cross-field operators currently assume orthogonal poloidal grids, more advanced multi-ion closures are not yet implemented, kinetic neutrals are not included, and extension to large 3D problems depends on further optimization and preconditioner development (Dudson et al., 2023).
5. HERMES++ as an automated software-publication workflow
In research-software engineering, HERMES stands for HELmholtz Rich MEtadata Software Publication. It is both a concept and a Python package implementing automated software publication with rich metadata in CI pipelines. The workflow is phase-based: harvest, process, curate, deposit, and post-process. Harvest plugins collect metadata from sources such as CITATION.cff, CodeMeta files, and Git history; process collates them into a single extended CodeMeta JSON-LD document with provenance annotations; curate supports review; deposit targets systems such as InvenioRDM and Zenodo; and post-process feeds DOI and record information back into repository files (Kernchen et al., 2024).
The implementation is plugin-based, with phase-specific base classes such as HermesHarvestPlugin, HermesDepositPlugin, and HermesPostprocessPlugin, registration via Python entry points, configuration through hermes.toml, and settings validation through pydantic-settings. CLI subcommands mirror the workflow phases: hermes harvest, hermes process, hermes curate, hermes deposit, and hermes postprocess. Built-in plugins include Git, CFF, CodeMeta, Invenio/InvenioRDM, file output, DOI feedback into CITATION.cff, and record-ID feedback into configuration (Kernchen et al., 2024).
The projected HERMES++ in this domain is a next-generation version that retains the five-phase model, CodeMeta JSON-LD core, plugin architecture, and CI-centric operation, while enhancing metadata intelligence, provenance-aware merging, policy-aware curation, richer plugin coverage, improved diagnostics, hermes init-style guided configuration, and stronger CI templating for branch- and tag-triggered publication workflows. The case studies reported in the paper show feasibility but also identify limitations in logging, configuration UX, template flexibility, process and curate extension points, and metadata-model tooling (Kernchen et al., 2024).
6. Shared design motifs, divergences, and open problems
Across these otherwise unrelated uses, a family resemblance is visible. This suggests that “++” repeatedly marks a move toward modularity, extensibility, and integration rather than a specific theoretical formalism. In the computing-continuum architecture, that appears as bridges linking monetization, orchestration, semantics, and trust; in the activation toolkit, as a hybrid decomposition between Geant4 event simulation and analytical decay-chain evolution; in the Galactic code, as separation between integrators, skymaps, and astrophysical components; in Hermes-3, as a component–state architecture over BOUT++; and in the software-publication workflow, as a phase-based plugin system (Dehury et al., 9 Dec 2025).
The divergences are equally important. Some HERMES++ instances are purely conceptual extensions rather than released software, notably the computing-continuum framework and the UHECR propagation code’s projected successor. Others are already concrete implementations with validation and performance characterization, such as the HERMES-SP activation-background toolkit, the Galactic multi-messenger simulator, Hermes-3, and the automated software-publication workflow (Domenico et al., 2013).
Open problems also remain domain-specific. The computing-continuum framework identifies open standards, scalability and performance across millions of devices, privacy-preserving monetization, semantic scalability, and decentralized governance as major challenges. The UHECR code points toward improved background models, magnetic-field prescriptions, interaction modeling, and benchmarking. Hermes-3 identifies non-orthogonal metric terms, more advanced collisional closures, kinetic neutral models, and larger-scale solver optimization. The metadata-publication workflow highlights richer curation, better metadata-model libraries, and more robust CI integration (Kernchen et al., 2024).
HERMES++ is therefore best understood as a polysemous research label. In one setting it denotes a hypothetical continuum-native platform joining WebAssembly orchestration, semantic interoperability, decentralized markets, and DIRECTS-driven trust; in another it refers to efficient hybrid activation-background simulation; elsewhere it names an extensible Galactic emission engine, a BOUT++-based multi-fluid plasma code, or a next-generation FAIR software-publication workflow. The shared implication is not common subject matter, but a recurrent design trajectory: from a named HERMES baseline toward a more composable, automated, and technically expressive successor.