intensify 0.2.0__tar.gz

intensify-0.2.0/.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,62 @@
+name: Bug report
+description: Something doesn't work as documented.
+labels: ["bug"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to file a bug report. A minimal,
+        reproducible example is the fastest path to a fix.
+
+  - type: textarea
+    id: reproducer
+    attributes:
+      label: Minimal reproducer
+      description: Self-contained Python snippet that demonstrates the bug.
+      render: python
+    validations:
+      required: true
+
+  - type: textarea
+    id: expected
+    attributes:
+      label: Expected vs observed
+      description: What you expected to happen, and what actually happened.
+    validations:
+      required: true
+
+  - type: textarea
+    id: traceback
+    attributes:
+      label: Full traceback (if any)
+      render: text
+
+  - type: input
+    id: version
+    attributes:
+      label: intensify version
+      placeholder: "`python -c 'import intensify; print(intensify.__version__)'`"
+    validations:
+      required: true
+
+  - type: input
+    id: python
+    attributes:
+      label: Python version
+      placeholder: "3.12.2"
+    validations:
+      required: true
+
+  - type: dropdown
+    id: backend
+    attributes:
+      label: Active backend
+      options: [jax, numpy]
+    validations:
+      required: true
+
+  - type: input
+    id: os
+    attributes:
+      label: OS
+      placeholder: "Ubuntu 22.04 / macOS 14 / Windows 11"

intensify-0.2.0/.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,22 @@
+name: Feature request
+description: Suggest a new feature or capability.
+labels: ["enhancement"]
+body:
+  - type: textarea
+    id: problem
+    attributes:
+      label: What problem does this solve?
+      description: Describe the use case, not just the desired API.
+    validations:
+      required: true
+
+  - type: textarea
+    id: proposal
+    attributes:
+      label: Proposed solution
+      description: A sketch of the API or implementation.
+
+  - type: textarea
+    id: alternatives
+    attributes:
+      label: Alternatives considered

intensify-0.2.0/.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,19 @@
+## Summary
+
+<!-- What does this PR do and why? 1-3 sentences. -->
+
+## Changes
+
+<!-- Bulleted list of the concrete changes. -->
+
+## Checklist
+
+- [ ] Tests added or updated (and pass locally)
+- [ ] `ruff check intensify tests` clean
+- [ ] Public API changes documented in `CHANGELOG.md` under `## [Unreleased]`
+- [ ] Math changes cite a paper or textbook in docstrings / comments
+- [ ] Breaking change? Call it out clearly above
+
+## Related issues
+
+Closes #

intensify-0.2.0/.github/workflows/ci.yml

@@ -0,0 +1,129 @@
+name: CI
+
+on:
+  push:
+    branches: [main, master]
+  pull_request:
+    branches: [main, master]
+
+jobs:
+  test:
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        python-version: ["3.10", "3.11", "3.12"]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+          cache: 'pip'
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install hatch
+          hatch env create
+
+      - name: Lint with ruff
+        run: hatch run lint:check
+
+      - name: Test with pytest
+        run: hatch run pytest:test
+
+  typecheck:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -e ".[dev]" mypy
+
+      - name: mypy
+        run: mypy intensify --ignore-missing-imports --no-strict-optional || true
+
+  docs:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install hatch
+          pip install -e ".[docs]"
+
+      - name: Build docs
+        run: sphinx-build -W -b html docs docs/_build/html
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: docs-html
+          path: docs/_build/html
+
+  readme-doctest:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+      - name: Install
+        run: |
+          python -m pip install --upgrade pip
+          pip install -e ".[dev]"
+      - name: Doctest README
+        env:
+          MPLBACKEND: Agg
+        run: |
+          python -c "
+          import os, re, sys, subprocess
+          src = open('README.md').read()
+          blocks = re.findall(r'\`\`\`python\n(.*?)\n\`\`\`', src, flags=re.S)
+          if not blocks:
+              print('No python blocks found in README'); sys.exit(0)
+          code = blocks[0]
+          print('Running README quickstart block...')
+          env = dict(os.environ, MPLBACKEND='Agg')
+          subprocess.check_call([sys.executable, '-c', code], env=env)
+          "
+
+  build:
+    runs-on: ubuntu-latest
+    needs: [test, typecheck, docs]
+    if: github.ref == 'refs/heads/main'
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Build distribution
+        run: |
+          pip install hatch
+          hatch build
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: dist
+          path: dist/*

intensify-0.2.0/.github/workflows/publish.yml

@@ -0,0 +1,37 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags: ["v*.*.*"]
+
+jobs:
+  build-and-publish:
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/project/intensify/
+    permissions:
+      id-token: write  # OIDC trusted publishing
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Build sdist + wheel
+        run: |
+          python -m pip install --upgrade pip hatch
+          hatch build
+
+      - name: Sanity-check artifacts
+        run: |
+          ls -la dist/
+          python -m pip install twine
+          python -m twine check dist/*
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          packages-dir: dist/

intensify-0.2.0/.gitignore

@@ -0,0 +1,47 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+env/
+venv/
+.venv/
+env.bak/
+venv.bak/
+
+# IDEs
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+
+# Distribution / packaging
+dist/
+build/
+*.egg-info/
+*.egg
+manifest.in
+wheels/
+
+# Testing
+.pytest_cache/
+.coverage
+htmlcov/
+nosetests.xml
+coverage.xml
+*.cover
+
+# Project-specific
+*.db
+*.sqlite
+data/
+results/
+.DS_Store
+
+# JAX
+.jax/
+
+# Sphinx build output
+docs/_build/

intensify-0.2.0/.pre-commit-config.yaml

@@ -0,0 +1,24 @@
+repos:
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.11.2
+    hooks:
+      - id: ruff
+        args: [--fix]
+      - id: ruff-format
+
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.6.0
+    hooks:
+      - id: check-yaml
+      - id: check-toml
+      - id: check-added-large-files
+        args: ["--maxkb=1000"]
+      - id: end-of-file-fixer
+      - id: trailing-whitespace
+      - id: mixed-line-ending
+
+  - repo: https://github.com/codespell-project/codespell
+    rev: v2.3.0
+    hooks:
+      - id: codespell
+        args: ["--skip=*.ipynb,coverage.xml,.venv,docs/_build"]

intensify-0.2.0/CHANGELOG.md

@@ -0,0 +1,177 @@
+# Changelog
+
+All notable changes to `intensify` are documented here.
+The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/) and
+this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+### Added
+- `MLEInference.fit(..., fit_decay=False)` — locks every β slot in the
+  flat parameter vector to its initial value via a zero-width L-BFGS-B
+  bound. Reduces the active parameter count, speeds up the fit, and
+  matches the problem `tick.HawkesExpKern` solves (β supplied as
+  input). Supported for `ExponentialKernel` and `SumExponentialKernel`
+  in univariate, marked, nonlinear, and multivariate paths. Raises
+  `TypeError` for kernels without an identifiable decay-rate axis
+  (`PowerLawKernel.beta` is a tail exponent, not a decay rate).
+- **Scaling-study benchmarks** in `docs/scaling.md` showing
+  `mv_exp_5d` fit time vs N from 501 to 91,249 total events. Both
+  intensify and tick scale linearly; the ~10× wall-clock ratio is
+  stable across the whole range (it does not grow at scale);
+  parameter-recovery RMSE is statistically indistinguishable by
+  N≈10k. Reproducible via `benchmarks/run_scaling.py`.
+- **O(N·M) recursive multivariate neg-log-likelihood** for the
+  shared-β exp-Hawkes case (every β_{m,k} the same scalar), cached
+  per `M`. Auto-selected by `_fit_multivariate_numpy` when
+  `fit_decay=False` and the kernel matrix has a shared β. Replaces the
+  dense O(N²) lag matrix with a `jax.lax.scan` carrying an M-vector
+  state per source dim. For `mv_exp_5d` (N=1099, M=5): ~220× fewer
+  compute steps, **17 ms fits vs 200 ms** on the same problem.
+  Per-cell β still falls back to the dense path.
+
+### Fixed
+- **Correctness (critical, multivariate Hawkes likelihood)**:
+  `MultivariateHawkes._log_likelihood_dim` summed `log λ_m(t)` over
+  every event in the system, then summed across `m` — an `M`-fold
+  overcounting of the log-intensity term. The textbook likelihood is
+  `Σ_n log λ_{k_n}(t_n) − Σ_m Λ_m(T)`, i.e. each event contributes
+  `log` of its own *source* dim's intensity exactly once. Verified
+  against a hand-computed reference on a 2-d sim (current −63.0 vs
+  correct −44.4 on an identical data set). Fix causes multivariate MLE
+  parameter RMSE on the `mv_exp_5d` benchmark to drop from 0.167 → ~0.1,
+  now within 2× of tick on the same data. This bug was silently
+  affecting every multivariate Hawkes fit since the library's
+  inception.
+- **Performance (univariate JAX MLE)**: each fit was paying ~370 ms on
+  fresh JIT compilation because `_make_jit_neg_loglik_*` returned a new
+  `@jax.jit`-decorated closure per call (closing over `events_jax` /
+  `T_jax`). JAX caches by function identity, so the trace was
+  recompiled every fit. The neg-log-likelihood functions are now lifted
+  to module scope and the `value+grad` / Hessian closures are cached by
+  `(kernel kind, n_components, r)`. Result: `uni_exp_small` fit dropped
+  from **376 ms → 5 ms** steady-state (~75× faster), bringing univariate
+  intensify within ~5× of `tick.HawkesExpKern` on the same data while
+  still fitting the decay rate jointly.
+- **Performance (multivariate JAX MLE)**: `_fit_multivariate_numpy` now
+  has a JAX-JIT fast path for the all-`ExponentialKernel` case
+  (indirectly required by the flat-vector parameter layout already).
+  Uses a dense N×N causal lag matrix with gathered per-pair
+  `(α_mk, β_mk)` from `(source_i, source_j)` indexing, and caches the
+  compiled `value+grad` per `M`. Multivariate `mv_exp_5d` fit dropped
+  **77 s → 1.1 s** (~70× faster). Mixed-kernel multivariate matrices
+  still take the numpy fallback path (unchanged).
+- Pre-0.2.0 callers of `_make_jit_neg_loglik_exp` /
+  `_make_jit_neg_loglik_sum_exp` / `_make_jit_neg_loglik_power_law` /
+  `_make_jit_neg_loglik_approx_pl` and the old
+  `_jax_hessian_std_errors(neg_ll_fn, x_opt, names)` signature still
+  work — they're shimmed to the cached implementation.
+
+## [0.2.0] - 2026-04-20
+
+Public launch release. Consolidates correctness, API polish, and the
+full documentation, CI, and scientific-credibility scaffolding needed
+for lab adoption. Includes verified head-to-head benchmarks vs
+[tick](https://github.com/X-DataInitiative/tick).
+
+### Fixed
+- **Correctness (critical)**: `_recursive_compensators` in the time-rescaling
+  diagnostic had incorrect math for `ExponentialKernel` /
+  `SumExponentialKernel` — the computed cumulative intensities were off by a
+  factor of β, causing KS p-values to be catastrophically small under
+  well-specified models. Replaced with the correct recursion (equivalent
+  to the verified general pairwise-integral path, but O(N) instead of
+  O(N²)). Unknown recursive kernels now fall back to the general path
+  rather than applying the buggy formula. Before this fix the fast path of
+  `time_rescaling_test` and `qq_plot` produced wrong p-values for every
+  exponential-family Hawkes fit.
+
+### Added
+- `benchmarks/` with README, `reference_dataset.py`, `run_intensify.py`,
+  `run_tick.py`, and verified head-to-head results in
+  `docs/benchmarks.md`. Datasets use portable `.npy`+`.json` pairs so the
+  same files work across NumPy 1.x and 2.x environments. pyhawkes was
+  evaluated and dropped (transitive deps reference
+  `scipy.misc.logsumexp`, removed in SciPy 1.0).
+- `tests/test_textbook_cases.py`: parameter-recovery on seeded simulations,
+  log-likelihood scaling, well-specified-vs-misspecified KS sanity check.
+- Community infrastructure: `CITATION.cff`, `CONTRIBUTING.md`,
+  `CODE_OF_CONDUCT.md`, `SECURITY.md`, GitHub issue + PR templates.
+- Documentation: expanded `docs/getting_started.md` with regularization
+  shorthand, backend-switching, and input-validation notes; new
+  `docs/user_guide/diagnostics.md` covering the (corrected) time-rescaling
+  theorem; new `docs/user_guide/simulation.md`; kernel-coverage table in
+  `docs/user_guide/inference.md`.
+- README rewrite: PyPI / Python / license / CI badges, "Why intensify?"
+  comparison vs `tick` and `pyhawkes`, citation block, doc links.
+- CI: multi-OS matrix (ubuntu / macOS / windows), mypy job, Sphinx
+  docs-build job with `-W` treat-warnings-as-errors, README-quickstart
+  doctest job (`MPLBACKEND=Agg`), PyPI publish workflow on tags using
+  OIDC trusted publishing.
+- Pre-commit: `codespell`, `check-yaml`, `check-toml`,
+  `check-added-large-files`, `end-of-file-fixer`, `trailing-whitespace`,
+  `mixed-line-ending`.
+
+## [0.1.1] - 2026-04-19
+
+API-consistency release. No correctness changes; surface-area polish.
+
+### Added
+- `MarkedHawkes` and `NonlinearHawkes` MLE now support every kernel type
+  the univariate helpers understand: `ExponentialKernel` (signed and
+  unsigned), `SumExponentialKernel`, `PowerLawKernel`, `ApproxPowerLawKernel`,
+  `NonparametricKernel`. Previously both processes raised
+  `NotImplementedError` for anything other than unsigned `ExponentialKernel`.
+- `regularization="l1"` / `"elasticnet"` string shorthand accepted by
+  `MLEInference.fit`, resolving to default `L1()` / `ElasticNet()`.
+  Instances still pass through unchanged.
+- Backend proxy: `get_backend()` returns a lazy proxy that delegates every
+  attribute to the currently active backend. Module-level
+  `bt = get_backend()` captures now pick up runtime `set_backend()` switches.
+- `tests/test_mle_kernel_expansion.py`: parametric regression suite for the
+  expanded MLE kernel surface + regularization shorthand + backend proxy.
+
+### Changed
+- `hawkes_mle_bounds` and `hawkes_mle_apply_vector` now preserve
+  `ExponentialKernel.allow_signed` so signed kernels round-trip through MLE
+  correctly when reused by `NonlinearHawkes`.
+
+### Fixed
+- Pre-existing test/spec mismatches aligned with documented behavior:
+  `endogeneity_index` saturates at 1.0 at critical branching (per docstring);
+  `FitResult.connectivity_matrix()` raises `TypeError` (not `ValueError`)
+  when called on a non-multivariate process — test expectations updated.
+
+## [0.1.0] - 2026-04-19
+
+First versioned release. Focus: correctness on real data, stable packaging.
+
+### Added
+- `__version__` attribute on the top-level package.
+- `py.typed` marker for PEP 561 type-checker support.
+- `Kernel.scale(factor)` method on all built-in kernels so
+  `project_params` can enforce stationarity uniformly across kernel types
+  (`ExponentialKernel`, `SumExponentialKernel`, `PowerLawKernel`,
+  `ApproxPowerLawKernel`, `NonparametricKernel`).
+- `RuntimeWarning` emitted from every MLE fit path when the SciPy optimizer
+  does not converge — previously only the multivariate path warned.
+- `RuntimeWarning` when the fitted multivariate Hawkes spectral radius
+  remains ≥ 1 after projection (non-stationary fit).
+- Minimum-version pins on all runtime dependencies (`numpy>=1.24`,
+  `jax>=0.4.20`, `scipy>=1.11`, `optax>=0.1.7`, `matplotlib>=3.7`).
+- Project URLs: Documentation, Issues, Changelog.
+- Classifiers: `Operating System :: OS Independent`,
+  `Topic :: Scientific/Engineering :: Mathematics`, `Typing :: Typed`.
+
+### Changed
+- `UnivariateHawkes.project_params` and `MultivariateHawkes.project_params`
+  now use `Kernel.scale()`, so non-`alpha` kernels (e.g. `SumExponentialKernel`,
+  `NonparametricKernel`) are actually projected instead of silently skipped.
+- Version bumped from `0.1.0-alpha` to `0.1.0`.
+
+### Fixed
+- Replaced runtime-validation `assert` statements in
+  `intensify/core/kernels/nonparametric.py` and
+  `intensify/core/processes/marked_hawkes.py` with explicit `ValueError` /
+  `RuntimeError` raises. Asserts are stripped under `python -O` and previously
+  hid real errors.

intensify-0.2.0/CITATION.cff

@@ -0,0 +1,38 @@
+cff-version: 1.2.0
+message: "If you use intensify in your research, please cite it as below."
+title: "intensify: A modern Python point process library with deep Hawkes specialization"
+abstract: >-
+  intensify is an open-source Python library for simulating, fitting, and
+  diagnosing point process models — with emphasis on Hawkes processes for
+  quantitative finance and computational neuroscience. It provides
+  univariate, multivariate, marked, and nonlinear Hawkes variants; a
+  pluggable kernel family (exponential, sum-of-exponentials, power-law,
+  approximate power-law, nonparametric); MLE inference with JAX autodiff
+  and a NumPy fallback; and goodness-of-fit tooling based on the
+  time-rescaling theorem.
+type: software
+authors:
+  - family-names: Hill
+    given-names: Matthew
+license: MIT
+repository-code: "https://github.com/hillmatt7/intensify"
+url: "https://hillmatt7.github.io/intensify"
+version: 0.2.0
+date-released: "2026-04-19"
+keywords:
+  - point-process
+  - hawkes-process
+  - self-exciting
+  - maximum-likelihood
+  - jax
+  - quantitative-finance
+  - computational-neuroscience
+  - time-rescaling
+preferred-citation:
+  type: software
+  authors:
+    - family-names: Hill
+      given-names: Matthew
+  title: "intensify: A modern Python point process library with deep Hawkes specialization"
+  year: 2026
+  url: "https://github.com/hillmatt7/intensify"

intensify-0.2.0/CODE_OF_CONDUCT.md

@@ -0,0 +1,43 @@
+# Contributor Covenant Code of Conduct
+
+## Our pledge
+
+We as members, contributors, and leaders pledge to make participation in
+the `intensify` community a harassment-free experience for everyone,
+regardless of age, body size, visible or invisible disability, ethnicity,
+sex characteristics, gender identity and expression, level of experience,
+education, socio-economic status, nationality, personal appearance, race,
+caste, color, religion, or sexual identity and orientation.
+
+We pledge to act and interact in ways that contribute to an open, welcoming,
+diverse, inclusive, and healthy community.
+
+## Our standards
+
+Examples of behavior that contributes to a positive environment:
+
+- Demonstrating empathy and kindness toward other people.
+- Being respectful of differing opinions, viewpoints, and experiences.
+- Giving and gracefully accepting constructive feedback.
+- Taking responsibility and apologizing to those affected by our mistakes.
+- Focusing on what is best not just for us as individuals but for the
+  overall community.
+
+Examples of unacceptable behavior:
+
+- Sexualized language or imagery; sexual attention or advances of any kind.
+- Trolling, insulting or derogatory comments, personal or political attacks.
+- Public or private harassment.
+- Publishing others' private information without explicit permission.
+- Other conduct that could reasonably be considered inappropriate in a
+  professional setting.
+
+## Enforcement
+
+Instances of abusive, harassing, or otherwise unacceptable behavior may be
+reported to the maintainers via GitHub's private security advisory channel
+or by email. All complaints will be reviewed and investigated promptly and
+fairly.
+
+This Code of Conduct is adapted from the
+[Contributor Covenant](https://www.contributor-covenant.org/), version 2.1.

intensify-0.2.0/CONTRIBUTING.md

@@ -0,0 +1,68 @@
+# Contributing to intensify
+
+Thank you for considering a contribution. This project aims to be a reliable,
+well-tested building block for point-process research in labs and industry.
+
+## Ground rules
+
+- **Correctness first.** Every numerical change should ship with a test that
+  asserts the quantity it affects, not just that the code runs.
+- **No silent fallbacks.** If a code path can produce a wrong answer under
+  certain conditions (non-stationary fit, degenerate data, optimizer stall),
+  surface it as a `warning` or `raise`, never as a silent sentinel.
+- **Backend-agnostic.** Anything under `intensify/core/` must work with
+  both the JAX and NumPy backends. Use `from ...backends import get_backend`
+  and call backend methods through the returned proxy.
+
+## Development setup
+
+```bash
+git clone https://github.com/hillmatt7/intensify
+cd intensify
+
+# Create a virtualenv and install with dev extras
+python -m venv .venv
+source .venv/bin/activate
+pip install -e ".[dev,docs]"
+
+pre-commit install
+```
+
+## Running checks
+
+```bash
+# Lint + format
+hatch run lint:check
+
+# Tests (with coverage)
+pytest tests/
+
+# Docs
+hatch run docs:build
+```
+
+## Pull request checklist
+
+- [ ] Tests cover the new or changed behavior.
+- [ ] `pytest tests/` passes locally (no new warnings unless intentional).
+- [ ] `ruff check intensify tests` is clean.
+- [ ] Public API changes are documented in `CHANGELOG.md` under `## [Unreleased]`.
+- [ ] If you changed math, a citation (paper or textbook chapter) appears in
+      the docstring or in-line comment.
+
+## Reporting bugs
+
+Open an issue at
+<https://github.com/hillmatt7/intensify/issues/new/choose>.
+Please include a minimal reproducer, the observed vs expected behavior,
+your Python version, `intensify.__version__`, and the active backend
+(`intensify.get_backend_name()`).
+
+## Security
+
+See [`SECURITY.md`](SECURITY.md) for reporting security issues privately.
+
+## Community
+
+This project follows the [Contributor Covenant](CODE_OF_CONDUCT.md). By
+participating you agree to uphold it.