DESolver
This is a python package for solving Initial Value Problems using various numerical integrators. Many integration routines are included ranging from fixed step to symplectic to adaptive integrators.
Documentation
Documentation is now available at desolver docs! This will be updated with new examples as they are written.
To Install:
Just type
pip install desolver
Use of PyTorch backend requires installation of PyTorch from here.
Minimal Working Example
This example shows the integration of a harmonic oscillator using DESolver.
import desolver as de
import desolver.backend as D
def rhs(t, state, k, m, **kwargs):
return D.array([[0.0, 1.0], [-k/m, 0.0]])@state
y_init = D.array([1., 0.])
a = de.OdeSystem(rhs, y0=y_init, dense_output=True, t=(0, 2*D.pi), dt=0.01, rtol=1e-9, atol=1e-9, constants=dict(k=1.0, m=1.0))
print(a)
a.integrate()
print(a)
print("If the integration was successful and correct, a[0].y and a[-1].y should be near identical.")
print("a[0].y = {}".format(a[0].y))
print("a[-1].y = {}".format(a[-1].y))
print("Maximum difference from initial state after one oscillation cycle: {}".format(D.max(D.abs(a[0].y-a[-1].y))))Implemented Integration Methods
Explicit Methods
Adaptive Methods
Runge-Kutta 14(12) (Feagin, 2009)
Runge-Kutta 10(8) (Feagin, 2009)
Runge-Kutta 8(7) (Dormand & Prince, 1980)
Runge-Kutta 4(5) with Cash-Karp Coefficients
Adaptive Heun-Euler Method
Fixed Step Methods
Symplectic BABs9o7H Method (Mads & Nielsen, 2015, BAB’s9o7H)
Symplectic ABAs5o6HA Method (Mads & Nielsen, 2015, ABAs5o6H)
Runge-Kutta 5 - The 5th order integrator from RK45 with Cash-Karp Coefficients.
Runge-Kutta 4 - The classic RK4 integrator
Midpoint Method
Heun’s Method
Euler’s Method
Euler-Trapezoidal Method
Implicit Methods
Adaptive Methods
Lobatto IIIC 4(2) (Kroulíková, 2017)
Radau IIA 5(2) (Kroulíková, 2017)
Fixed Step Methods
Backward Euler
Implicit Midpoint
Crank-Nicolson
Lobatto IIIA 2
Lobatto IIIB 2
Lobatto IIIC 2
Radau IA 3
Radau IIA 3
Lobatto IIIA 4
Lobatto IIIB 4
Gauss-Legendre 4
Radau IA 5
Radau IIA 6
References
Feagin, T. (2009). High-Order Explicit Runge-Kutta Methods. Retrieved from https://sce.uhcl.edu/rungekutta/
Dormand, J. R. and Prince, P. J. (1980) A family of embedded Runge-Kutta formulae. Journal of Computational and Applied Mathematics, 6(1), 19-26. https://doi.org/10.1016/0771-050X(80)90013-3
Mads, K. and Nielsen, E. (2015). Efficient fourth order symplectic integrators for near-harmonic separable Hamiltonian systems. Retrieved from https://arxiv.org/abs/1501.04345
Kroulíková, T. (2017). RUNGE-KUTTA METHODS (Master’s thesis, BRNO UNIVERSITY OF TECHNOLOGY, Brno, Czechia). Retrieved from https://www.vutbr.cz/www_base/zav_prace_soubor_verejne.php?file_id=174714
Changelog
Version 5.0.0
Fully ported backend to use autoray instead of manual specification of each backend, this enables a more portable setup
Fully removed pyaudi as a backend
Version 4.5.0
Updated to pyproject.toml setup and removed pyaudi support.
Added changelog to repository
Moved to pyproject.toml and hatch build system
Updated github workflow to use hatch
Reconfigured tests to have skip flags and have tests enabled by default
Version 4.2.0
Improved performance of implicit methods, added embedded implicit methods following Kroulíková (2017) for fully implicit adaptive integration.
Version 4.1.0
Initial release of implicit integration schemes that use a basic newton-raphson algorithm to solve for the intermediate states.
Version 3.0.0
PyAudi support has been finalised. It is now possible to do numerical integrations using gdual variables such as gdual_double, gdual_vdouble and gdual_real128 (only on select platforms, refer to pyaudi docs for more information). Install desolver with pyaudi support using pip install desolver[pyaudi]. Documentation has also been added and is available at desolver docs.
Version 2.5.0
Event detection has been added to the module. It is now possible to do numerical integration with terminal and non-terminal events.
Version 2.2.0
PyTorch backend is now implemented. It is now possible to numerically integrate a system of equations that use pytorch tensors and then compute gradients from these.
1 maintainer
Metadata
1 maintainer
Release files for desolver 5.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| desolver-5.1.0.tar.gz | 72.8 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| desolver-5.1.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 151.3 kB
Release files / desolver-5.1.0.tar.gz
| Download URL | desolver-5.1.0.tar.gz |
|---|---|
| Size | 72.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
14122739c4b2c47180d44cc3d184f6b60c580660a3dddd096eeaa83b31f7a816
|
|
BLAKE2b-256 checksum How to use checksums |
8cb3e238c1e9bd9a8a0156ce7753b708071b3e13248df8872a8120a1c466728e
|
| 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 Oct 16, 2025.
Transparency logRelease files / desolver-5.1.0-py3-none-any.whl
| Download URL | desolver-5.1.0-py3-none-any.whl |
|---|---|
| Size | 78.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
74aca16b815aa9335aa10c77ba434fcca33abdf025f0bc70a7e51e807f277688
|
|
BLAKE2b-256 checksum How to use checksums |
575e4f6cd82e821fa7dcbefaae89251257be6505969be21421ca4aa1ceb5cec0
|
| 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 Oct 16, 2025.
Transparency log
