Exploratory landscape analysis with orivex¶
Compute the landscape features you need, with explicit costs and reproducible definitions.
orivex computes exploratory landscape analysis (ELA) features from sampled decision vectors and their objective values. It is a successor to pflacco, with a NumPy/SciPy core and an optional differentiable PyTorch backend.
Pre-release
Install from a source checkout. orivex implements selected feature families and is not yet a drop-in replacement for pflacco. See the feature overview for coverage.
Why orivex¶
- Selective computation: request individual outputs; shared intermediates run once per call.
- Explicit costs: inspect declared CPU, memory, and additional objective-evaluation costs.
- Versioned definitions: trace feature semantics, invariance claims, and source literature.
- Verification: analytical, metamorphic, and R flacco differential tests check the implementation.
- Differentiable features: use the supported Torch profile with device and autograd awareness.
Start with a sample¶
import numpy as np
from orivex import LandscapeSample, compute
rng = np.random.default_rng(42)
x = rng.uniform(-5.0, 5.0, size=(200, 2))
y = np.sum(x**2, axis=1)
sample = LandscapeSample(x, y, lower=[-5.0, -5.0], upper=[5.0, 5.0])
result = compute(sample, "ela_distr.*")
for name, item in result.values.items():
print(name, item.value, item.status.value)
Objective values are min-max normalized by default. Choose preprocessing explicitly when reproducing reference results; see objective normalization.
Find your next step¶
| Task | Documentation |
|---|---|
| Install and compute your first features | Getting started |
| Choose outputs and control execution | Feature selection |
| Understand definitions and costs | Feature catalogue |
| Differentiate features in a model | PyTorch and autograd |
| Interpret invalid values and record provenance | Results |
| Look up signatures and data models | API reference |
| Contribute and reproduce benchmarks | Development |
orivex is distributed under the MIT License.