mc_dagprop
mc_dagprop is a fast propagation toolkit for directed acyclic graphs (DAGs), with a C++ Monte Carlo engine and a Python analytic engine.
The package provides two propagation modes:
- Monte Carlo simulation based on sampled edge delays.
- Analytic propagation of full discrete probability mass functions (PMFs).
Both engines share the same event/activity DAG model and expose aligned naming:
MonteCarloPropagator and AnalyticPropagator.
Background
mc_dagprop was developed as part of the SORRI project at the Institute for Transport Planning and Systems (IVT), ETH Zurich. The SORRI project— Simulation-based Optimisation for Railway Robustness Improvement —focuses on learning real-life constraints and objectives to determine timetables optimized for robustness interactively. This research is supported by the SBB Research Fund, which promotes innovative studies in transport management and the future of mobility in Switzerland.
Features
- High-performance Monte Carlo core in C++ via pybind11.
- Deterministic analytic propagator for full event-time PMFs.
- Custom per-activity-type Monte Carlo delay distributions:
- Constant
- Exponential
- Gamma
- Empirical absolute/relative
- Single-run (
run(seed)) and batched (run_many(seeds)) Monte Carlo APIs. - Shared DAG concepts (
Event,Activity,DagContext) and unified naming. - Optional global
max_delaycap available in both engines.
Note: For Monte Carlo, configuring multiple distributions for the same
activity_typeoverrides previous settings. Keep exactly one distribution per type.
Installation
This library requires Python 3.12 or newer.
# with poetry
poetry add mc-dagprop
# or with pip
pip install mc-dagprop
Quickstart (Monte Carlo)
from mc_dagprop import (
Activity,
DagContext,
Event,
EventTimestamp,
GenericDelayGenerator,
MonteCarloPropagator,
)
# 1) Build your DAG timing context
events = [
Event("A", EventTimestamp(0.0, 100.0, 0.0)),
Event("B", EventTimestamp(10.0, 100.0, 0.0)),
]
activities = {
(0, 1): Activity(idx=0, minimal_duration=60.0, activity_type=1),
}
precedence = [
(1, [(0, 0)]),
]
ctx = DagContext(
events=events,
activities=activities,
precedence_list=precedence,
max_delay=1800.0,
)
# 2) Configure a delay generator (one distribution per activity_type)
gen = GenericDelayGenerator()
gen.add_constant(activity_type=1, factor=1.5)
# 3) Run propagation
sim = MonteCarloPropagator(ctx, gen)
result = sim.run(seed=42)
print("Realized times:", result.realized)
print("Edge durations:", result.durations)
print("Causal predecessors:", result.cause_event)
Simulator remains available as a compatibility alias of
MonteCarloPropagator.
Analytic Propagator
Use AnalyticPropagator to propagate discrete delay PMFs deterministically.
from mc_dagprop import (
AnalyticContext,
Event,
EventTimestamp,
OverflowRule,
UnderflowRule,
create_analytic_propagator,
)
from mc_dagprop.analytic import AnalyticActivity, exponential_pmf
step = 1.0
delay_pmf = exponential_pmf(scale=10.0, step=step, start=0.0, stop=300.0)
events = (
Event("A", EventTimestamp(0.0, 10.0, 0.0)),
Event("B", EventTimestamp(0.0, 20.0, 0.0)),
)
activities = {
(0, 1): (0, AnalyticActivity(idx=0, pmf=delay_pmf)),
}
precedence = (
(1, ((0, 0),)),
)
ctx = AnalyticContext(
events=events,
activities=activities,
precedence_list=precedence,
step=step,
underflow_rule=UnderflowRule.TRUNCATE,
overflow_rule=OverflowRule.TRUNCATE,
max_delay=None,
)
sim = create_analytic_propagator(ctx)
pmfs = sim.run()
print(pmfs[1].pmf.values)
print(pmfs[1].pmf.probabilities)
Notes:
stepis the shared PMF grid spacing for the analytic context.create_analytic_propagator(..., validate=True)validates PMF alignment, mass consistency, indices, and DAG acyclicity before running.- PMF binary operations (
convolveandmaximum) use numerically stable intermediates (np.longdouble) and a post-operation mass correction. This prevents tiny probabilities from being lost to cumulative floating-point drift in deep analytic propagation chains. max_delaymirrors Monte Carlo semantics by capping each event atmin(latest, earliest + max_delay).
Package structure
mc_dagprop.monte_carlo— compiled Monte Carlo core and delay generator.mc_dagprop.analytic— pure-Python PMF types, distributions, and propagator.mc_dagprop.types— shared typed aliases.demo/— runnable examples for analytic and Monte Carlo usage.
Install the distribution as mc-dagprop and import from mc_dagprop:
from mc_dagprop import Simulator
Building wheels for Windows, macOS, and Linux
Cross-platform wheel builds are configured through GitHub Actions in
.github/workflows/build-wheels.yml using cibuildwheel.
- Windows (
windows-latest) - macOS (
macos-latest,x86_64andarm64) - Linux (
ubuntu-latest)
To run a local source build without the CI workflow:
python -m pip install --upgrade build
python -m build
References
Metadata
Release files for mc-dagprop 0.9.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| mc_dagprop-0.9.1.tar.gz | 47.6 kB | With provenance | Details for mc_dagprop-0.9.1.tar.gz |
Built distributions (wheels)
Total release size: 6.0 MB
Release files / mc_dagprop-0.9.1.tar.gz
| Download URL | mc_dagprop-0.9.1.tar.gz |
|---|---|
| Size | 47.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8d4be21316ed2cbb221d8147756c5dfd56995a6a3a372085013aba6992e4c40e
|
|
BLAKE2b-256 checksum How to use checksums |
a4440fc82f12e49c9bbafe8a07511adbecfcac24e63e7f43915fad874dd378cc
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp313-cp313-win_amd64.whl
| Download URL | mc_dagprop-0.9.1-cp313-cp313-win_amd64.whl |
|---|---|
| Size | 181.5 kB |
| Tags | CPython 3.13 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
8c89a16d6a1f73610aa07ab8c23cac97dc396043382c66b26f1ba9b351ab99ee
|
|
BLAKE2b-256 checksum How to use checksums |
12e91ce2a78b945799ea3a7a2e6fd40279fdf5395375642393ede74c6bc48ea1
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp313-cp313-win32.whl
| Download URL | mc_dagprop-0.9.1-cp313-cp313-win32.whl |
|---|---|
| Size | 161.4 kB |
| Tags | CPython 3.13 Windows x86-32 |
|
SHA-256 checksum How to use checksums |
409c5b4a38a74c35422db2c94a718e71db59307984be76b7fca832d19e5e13cd
|
|
BLAKE2b-256 checksum How to use checksums |
ad641744aa30facd169a00145b16ad1078ba3d7e3fc97939a63330df1ecb8d35
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | mc_dagprop-0.9.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 2.4 MB |
| Tags | CPython 3.13 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
98ec599ec8d6abf42829cd3050619270b8bba74a2e57f4a78aa015d1713986ca
|
|
BLAKE2b-256 checksum How to use checksums |
bdf51d335d77d8d33818c681a79e48a6a3dcb7db1405079ede236afea61a05f3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp313-cp313-macosx_11_0_arm64.whl
| Download URL | mc_dagprop-0.9.1-cp313-cp313-macosx_11_0_arm64.whl |
|---|---|
| Size | 202.3 kB |
| Tags | CPython 3.13 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
ff58509689b7f0141a6bda22526e92b3c157228adc611bd4b197a77564dff154
|
|
BLAKE2b-256 checksum How to use checksums |
2ab904a138d531e8534def8a181ab32af06ff2d3126d7b1c257bf732b4a3b765
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp312-cp312-win_amd64.whl
| Download URL | mc_dagprop-0.9.1-cp312-cp312-win_amd64.whl |
|---|---|
| Size | 181.5 kB |
| Tags | CPython 3.12 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
698bbdbecd30474beb177ab42a72b40d0216e9c635bb08ed1880e0b6df6e05cb
|
|
BLAKE2b-256 checksum How to use checksums |
dbb04c3524c64164de18936c3989ca9b925badd52835819f74a8970773f357b9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp312-cp312-win32.whl
| Download URL | mc_dagprop-0.9.1-cp312-cp312-win32.whl |
|---|---|
| Size | 161.4 kB |
| Tags | CPython 3.12 Windows x86-32 |
|
SHA-256 checksum How to use checksums |
d5a8a5d957eb00ba06b3c5455a72dd347b45990da2e741a1cccc742d39782ac9
|
|
BLAKE2b-256 checksum How to use checksums |
67bcc16d674fbdef7988a6595700d5597006d684ffa7ac888f1acd97266f08f3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl
| Download URL | mc_dagprop-0.9.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 2.4 MB |
| Tags | CPython 3.12 Linux glibc 2.27+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
c428dd3aa6a5cdd8b27e68c034633f32efd813363437751869ca82e3172cfe92
|
|
BLAKE2b-256 checksum How to use checksums |
2987c28c40ca09e80b3fc28f9cb7eb14bc6632deadab8daf3eb53967ae3d4b9d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logRelease files / mc_dagprop-0.9.1-cp312-cp312-macosx_11_0_arm64.whl
| Download URL | mc_dagprop-0.9.1-cp312-cp312-macosx_11_0_arm64.whl |
|---|---|
| Size | 202.3 kB |
| Tags | CPython 3.12 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
35271641abcb2953e18db1659b7dda80ab9cefe2a2ed501221b60b933dfb6ab7
|
|
BLAKE2b-256 checksum How to use checksums |
89133dc71487a98be6fef05c820fce012d6fd295a2b33cbb2cf45465b2241753
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logSecurity for mc-dagprop 0.9.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 9 files in this release have been uploaded with provenance metadata.
Attestation for mc_dagprop-0.9.1.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Feb 22, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.