Parallax Python SDK (strategic-mandate-parallax)

Numerical Insight Controls Execution — Official Python SDK for the Parallax quantitative valuation and Monte Carlo simulation platform by Strategic Mandate GmbH.

Installation

pip install strategic-mandate-parallax
# With full quantitative finance & visualization stack (pandas, SciPy, scikit-learn, Matplotlib):
pip install "strategic-mandate-parallax[quant]"

In Python, import the package as parallax:

import parallax
from parallax import Engine, Model, Config, Simulation, Valuation, Asset, Capex, Revenue
from parallax import viz

Architecture & Commercial Licensing (parallax-license)

Like commercial solver packages (gurobipy, blpapi), strategic-mandate-parallax on PyPI is a pure-Python thin SDK (py3-none-any.whl) and does not bundle the proprietary C++20 simulation engine binary.

• Without an Engine Runtime / License: You can freely install strategic-mandate-parallax in any environment (CI, cloud notebooks, developer laptops) to construct, inspect, type-check, and serialize models (parallax.model) and render institutional exhibits (parallax.viz).
• With a Licensed Engine Runtime (_parallax_native + license.pxl): parallax.Engine() dynamically loads the native C++20 runtime from ~/.parallax/runtime/ (or PARALLAX_RUNTIME_DIR / your Strategic Mandate installation) and unlocks sub-millisecond in-process Monte Carlo execution with zero-copy IEEE-754 binary memory views into numpy and pandas.

Activating Your License & Runtime

# 1. Inspect current license and native runtime status
parallax-license status

# 2. Install your cryptographic license file (.pxl)
parallax-license install-license ~/Downloads/license.pxl

# 3. Install the native engine runtime binary for your platform
parallax-license install-runtime ~/Downloads/_parallax_native.cpython-314-darwin.so

Key Capabilities

1. Schema-Driven Python Component Model (parallax.model): 92 strongly-typed component classes (Model, Config, Simulation, Valuation, Asset, Capex, Revenue, Expense, DebtFacility, EquityIssuance, GlobalVariable, DistConfig, Scenario, Trigger, TrpEvent, etc.) generated directly from COMPONENT_SCHEMA.json for programmatic and batch model generation.
2. In-Process C++20 Engine (parallax.Engine): Dispatches directly to parallax::InProcessEngine and parallax::MCResultsCache with the Python GIL released (Py_BEGIN_ALLOW_THREADS) during simulation.
3. Zero-JSON Binary Transfer to NumPy, pandas & scikit-learn: Reads raw IEEE-754 float64 scalar sample vectors (eng.scalar_array) and (n_paths, n_periods) time-series matrices (eng.timeseries_matrix) directly into numpy.ndarray, pandas.DataFrame, and scikit-learn (X, y) datasets (eng.sklearn_dataset).
4. Strategic Mandate Corporate Visualization Suite (parallax.viz): High-DPI quantitative exhibits (plot_fan_manifold, plot_npv_distribution, plot_scenario_ranges, plot_surrogate_attribution, plot_dscr_covenant, plot_pv_waterfall) and institutional pandas exhibit tables (sm_exhibit_table) with inline SVG sparklines and quantile range bars.

Building the Native Runtime from Source (Internal / CI)

# 1. Regenerate Python component classes from engine/COMPONENT_SCHEMA.json
node scripts/generate-python-models.cjs

# 2. Build the stripped native C++20 runtime (_parallax_native) into python/build/runtime/
cmake -S python -B python/build -DCMAKE_BUILD_TYPE=Release
cmake --build python/build -j8

# 3. Build the pure-Python thin wheel (dist/parallax_sdk-<version>-py3-none-any.whl)
cd python && python -s -c "import _build_backend; _build_backend.build_wheel('dist')"

License

Copyright © 2024–2026 Strategic Mandate GmbH. All rights reserved.
See LICENSE.txt for terms. Commercial licensing: https://nice.strategicmandate.com