Repository navigation
Releases
sesmesh 1.3.2
Cites this version with its Zenodo DOI 10.5281/zenodo.24412360; 1.3.1 cited a wrong one. Code and output are unchanged from 1.3.0; see 1.3.0 for the CUDA back end.
Binaries: macOS arm64 and Linux x86-64 (glibc 2.34 or newer), built without GPU features. On Linux arm64 and Windows, or for a GPU back end, use cargo install sesmesh-cli; add --features hip (needs ROCm's hipcc) or --features cuda (needs CUDA's nvcc) at build time.
Full details: CHANGELOG.md.
sesmesh 1.3.1
Removes the CI badge from the README, which is the crates.io page of sesmesh and sesmesh-cli (nothing runs on GitHub). This release cites a wrong Zenodo DOI; use 1.3.2 (10.5281/zenodo.24412360). Code and output are unchanged from 1.3.0; see 1.3.0 for the CUDA back end.
Binaries: macOS arm64 and Linux x86-64 (glibc 2.34 or newer), built without GPU features. On Linux arm64 and Windows, or for a GPU back end, use cargo install sesmesh-cli; add --features hip (needs ROCm's hipcc) or --features cuda (needs CUDA's nvcc) at build time.
Full details: CHANGELOG.md.
sesmesh 1.3.0
A GPU back end for NVIDIA GPUs (CUDA), next to the AMD one from 1.2.0. Default meshes are unchanged from 1.2.0 (identical apart from the version in the OBJ header).
Added
- GPU back end, CUDA (#25):
Backend::Cuda(CudaOptions)/Backend::cuda(i)inSurfaceParams,--backend cudain the CLI (featurecuda; defaultCpu). It runs the same kernels as the HIP back end. They are compiled withnvccinto a fat binary for compute capability 7.5 (Turing) to 12.0, plus PTX. Only the NVIDIA driver is needed at run time; it is loaded when a device is opened, so a binary built with the feature still runs on the CPU where the driver is missing. Exact mode writes the CPU back end's files byte for byte, and--refine fastworks as with HIP. Batches live in device memory (discrete GPUs).Backend::Autotries HIP, then CUDA, then the CPU. - Tested on a laptop with a GeForce RTX 2070 Max-Q (sm_75, 8 GB) and a 6-core i7-9750H. The CPU/GPU gate found no difference in 774 SES and 212 SAS/vdW cases (3J3Q at 0.25 Å, 800 M triangles, included). 1AON at 0.5 Å takes 3.04 s exact and 1.70 s fast, against 5.48 s on the laptop's CPU, which bounds both modes (
docs/06-gpu-cuda.md). tools/gpu/cuda-build.Containerfilebuilds the CUDA back end on a machine without CUDA.tools/gpu/gate.shtakesBACKEND=cuda, and can make its CPU references on another machine (MODE=cpu|gpu).
Changed
sesmesh-gpu: the batch descriptions, encodings and self-test moved to a shared modulebatch, with traits both back ends implement. The kernel source iskernels/sesmesh.cuh, shared by HIP and CUDA.--gpu-deviceapplies to--backend hipandcuda.- The GPU gate uses no Python. It streams the OBJ files (tens of GB at 0.25 Å) through
sha256sumandgrep, compares statistics withjq, and prints the totals withsummary.sh. This release passed it on both back ends.
Binaries: macOS arm64 and Linux x86-64 (glibc 2.34 or newer), built without GPU features. On Linux arm64 and Windows, or for a GPU back end, use cargo install sesmesh-cli; add --features hip (needs ROCm's hipcc) or --features cuda (needs CUDA's nvcc) at build time.
Full details: CHANGELOG.md.
sesmesh 1.2.0
A GPU back end for AMD GPUs (HIP / ROCm) and an opt-in fast mode for previews. Default meshes are byte-identical to 1.1.2.
Added
- GPU back end, HIP (#24): new crate
sesmesh-gpu,Backend::{Cpu, Auto, Hip}inSurfaceParams,--backend/--gpu-devicein the CLI (featurehip; defaultCpu). The device classifies, gathers and samples the grid in f32 and integer arithmetic only; refinement stays on the CPU in f64, so exact mode writes the CPU back end's meshes byte for byte. On a Ryzen AI Max+ 395 (16 Zen 5 cores, Radeon 8060S), 1AON at 0.25 Å takes 2.37 s instead of 3.40 s and 3J3Q at 1 Å takes 23.6 s instead of 28.9 s. - Fast mode (
RefineMode::Fast,--refine fast): the device also refines the crossing edges, in f32 with brick-local double-float coordinates; the CPU finishes in f64 the edges the device did not converge on. It gives the same triangles as exact mode, and the vertices are as close to the surface, but positions differ in the last bits. 1AON at 0.25 Å takes 1.41 s, 3J3Q at 1 Å 12.3 s (1.98–2.47× the 16 cores). - GPU benchmarks and validation:
hip-exact/hip-fastcases inbenches/surface.rsand the CPU/GPU gatetools/gpu/gate.sh(980 cases, 0 differences for this release).
Changed
- Releases are built, tested and uploaded from local machines; no workflow runs on GitHub (
tools/release/package.sh,docs/release.md). - The paper and website give the time on the 16-core Zen 5 machine wherever they give an M4 time.
Binaries: macOS arm64 and Linux x86-64 (glibc 2.34 or newer), built without hip. On Linux arm64 and Windows, or for the GPU back end: cargo install sesmesh-cli (add --features hip; it needs ROCm's hipcc at build time).
Full details: CHANGELOG.md.
sesmesh 1.1.2
A fix for a panic in the extended-XYZ reader found by the 24 CPU-hour fuzzing campaign, plus the 1 000-entry robustness sweep and the campaign tooling. Default meshes are byte-identical to 1.1.1.
Fixed
- Extended XYZ: a
Properties=descriptor whose column counts add up pastusize::MAXis now an input error. Before, the sum overflowed (a panic in debug builds; in release builds the wrapped count let short lines through and indexing panicked). Found by the fuzzing campaign (#22) after 276 s; regression test inio::xyz.
Added
- Robustness sweep of 1 000 further random PDB entries (#23): 1 000/1 000 ok (
docs/validation/sweep1000.csv).xtask sample-sweepandxtask sweeptake--set. fuzz/campaign.shfor long fuzzing campaigns. Every one of the 7 fuzz targets ran for ≥ 24.9 CPU-hours (174.5 in total, 494 M inputs); after the XYZ fix no crash, timeout or out-of-memory input (docs/validation/fuzz.csv).
Changed
- Cross-tool comparison runs repeated with 1.1.1 (no vertex more than 0.01 Å off the SES); paper draft updated.
Full details: CHANGELOG.md. Install: cargo install sesmesh-cli, or download a binary below.
sesmesh 1.1.1
A correctness fix for exact vertex refinement, a comparison with six other SES programs, a paper draft, and metadata for search engines and citation.
Fixed
- Exact vertex refinement could leave a rare vertex well off the SES where the field is not monotone along its grid edge: Newton/Illinois stalled and 1.1.0 placed the vertex at an unevaluated iterate (worst seen: 0.22 Å on 1RCL at 0.5 Å; 1AON 0.094 Å; 6VXX 0.083 Å at 0.25 Å), with normals up to 134° off. Such edges are now bisected on the sign-change bracket to within 10⁻⁵ Å. Default meshes change on those vertices only (0.001–0.011 % of vertices; triangles unchanged); every other vertex is bit-identical to 1.1.0. The semver policy now allows correctness fixes like this in a patch release (
docs/api-review.md).
Added
- Cross-tool comparison (#5, #6) with MSMS, NanoShaper, ESES, EDTSurf, UCSF ChimeraX and sesa on the same spheres (
docs/comparison.md,tools/compare/): NanoShaper and ESES agree with sesmesh to ≤ 0.03 % in area and ≤ 0.006 % in volume at 0.25 Å over the 53-entry validation corpus, and on 2PTN's 14 cavities. - Stored reference numbers (
tests/data/reference/) and a test comparing sesmesh with them. - Paper draft in
paper/.
Changed
- Crates link to the website and are listed under
science::bioinformatics;CITATION.cffadded ("Cite this repository").
Full details: CHANGELOG.md. Install: cargo install sesmesh-cli, or download a binary below.
sesmesh 1.1.0
Opt-in removal of sub-grid fragments, a corrected memory guard with lower peak memory, missing output directories created by the CLI, and the project website. Default meshes are byte-identical to 1.0.0.
Added
--drop-fragments/SurfaceParams::with_remove_fragments: removes sub-grid fragments, tiny solid components that Marching Cubes cuts off from reentrant sheets thinner than the grid spacing (4HHB at 0.25 Å: one, 0.037 Ų). Opt-in; counted inSurfaceStats::removed_fragmentsand the statistics JSON.- The memory guard's estimate is reported:
SurfaceStats::estimated_triangles/estimated_memory, a summary line with--max-memory,engine.estimated_*in the statistics JSON. - Project website: https://uvar-apps.github.io/sesmesh/ (guides, tutorials, gallery, live 3D viewer).
Changed
- Every source file carries a header with
Copyright (c) 2026 Ugo VarettoandSPDX-License-Identifier: MIT OR Apache-2.0; Cargo authors, licence files, README, crate docs, man page and website name the author. - Lower peak memory with identical output, e.g. the 1K4R capsid at 1 Å 1.68 → 1.54 GB, decimated to 1.5 M triangles 4.04 → 1.97 GB; 1AON at 0.25 Å decimated to 2 M triangles 5.23 → 2.85 GB.
Fixed
--max-memory/--auto-coarsenunderestimated the peak (a 1 GiB auto-coarsened capsid run peaked at 1.46 GB). The estimate is now an upper bound of the measured peak (1.15–1.7× on the test runs); the same run now peaks at 0.65 GB. Inputs close to the limit may now be refused or meshed at a coarser spacing than with 1.0.0.- CLI: missing output directories (
--output-dir, batch side-file directories, parents of-o) are created instead of failing every input.
Install
- CLI:
cargo install sesmesh-cli, or download an archive below and check it againstSHA256SUMS. - Library:
sesmesh = "1.1"— docs at https://docs.rs/sesmesh
Full changelog: CHANGELOG.md.
sesmesh 1.0.0
First stable release. The public API now follows semantic versioning.
Added
- Mesh decimation (
--decimate R,--max-triangles N,--decimate-error E;mesh::decimate): quadric-error half-edge collapses in pure Rust. Retained vertices keep their exact surface position, analytic normal and atom; the mesh stays closed and manifold; deterministic. --auto-coarsenwith--max-memory: grow the grid spacing until the memory estimate fits instead of refusing (SurfaceParams::auto_coarsen).- gzip-compressed PLY and MSMS output (
-o mesh.ply.gz,-o mesh.vert.gz). sesmesh completions <shell>andsesmesh man; the archives below includesesmesh.1and bash, zsh, fish and PowerShell completions.- Ctrl-C cancels cleanly (exit code 130).
- Fuzz target for biological assemblies.
Changed
- Lower memory, identical meshes: field data ≈ 1 150 → 560 bytes per atom; peak memory on the 1K4R capsid at 1 Å 2.42 → 1.74 GB, on 3J3Q at 1 Å 7.89 → 7.16 GB.
- Specification revised for 1.0: memory budget 150 B per triangle + 800 B per atom; analytic accuracy bound area ≤ 0.7 % / volume ≤ 1.5 % at 0.25 Å for atom-sized spheres (proteins: −0.20 % / −0.02 %); performance targets set to the measured times. Cross-tool references and a 24 CPU-hour fuzzing campaign follow after 1.0 (#19).
Fixed
- Assembly operator expressions that would expand to more than 100 000 copies or 500 million atoms are refused (exit code 4) instead of exhausting memory.
Install
- CLI:
cargo install sesmesh-cli, or download an archive below and check it againstSHA256SUMS. - Library:
sesmesh = "1"— docs at https://docs.rs/sesmesh
Full changelog: CHANGELOG.md.
sesmesh 0.9.1
Bug-fix release.
Fixed
- CLI: a reader closing standard output early (
sesmesh x.cif -o - | head,--stats - | head, pipedinfo/radii) no longer reports "error: standard output: Broken pipe" with exit code 1; the CLI stops quietly with exit code 0.--stats -no longer panics in that case.
Install
- CLI:
cargo install sesmesh-cli, or download an archive below and check it againstSHA256SUMS. - Library:
sesmesh = "0.9"— docs at https://docs.rs/sesmesh
Full changelog: CHANGELOG.md.
sesmesh 0.9.0
First public release of sesmesh, a fast and accurate Solvent Excluded Surface (Connolly surface) mesher — release candidate for 1.0.
Highlights
- Exact implicit SES field (Quan & Stamm) on a sparse narrow-band grid; Marching Cubes with every vertex placed on the exact surface and analytic normals. SAS and vdW surfaces too.
- Closed, consistently oriented, edge-manifold meshes; components and internal cavities (
--cavities keep|remove|only); per-vertex patch type and nearest atom; per-atom areas. - Inputs: PDB, PDBx/mmCIF, XYZ / extended XYZ, PQR (gzip), biological assemblies. Outputs: OBJ (with colours and groups), binary PLY, MSMS
.vert/.face, CSV side files, JSON statistics. - Multi-threaded and SIMD-accelerated (AVX-512 / AVX2 / SSE4.2 / NEON, runtime dispatch); byte-identical output for any thread count.
- Validation: on 53 PDB entries, area −0.20 % and volume −0.02 % (mean) at 0.25 Å relative to the h → 0 limit; 200/200 random PDB entries meshed without error; HIV-1 capsid (2.44 M atoms) at 1 Å in ≈ 70 s. See
docs/validation-report.md.
Install
- CLI:
cargo install sesmesh-cli(binarysesmesh), or download an archive below and check it againstSHA256SUMS. - Library:
sesmesh = "0.9"— docs at https://docs.rs/sesmesh
Before 1.0 (tracked in #19): decimation (#1), performance targets on the reference machines (#20), the long fuzzing campaign (#22), cross-tool reference comparisons (#5).
Full changelog: CHANGELOG.md. Manual: docs/cli.md.