numba-integrators 0.2.1

Numerical integrators using Numba

Numba Integrators

Numba Integrators is collection numerical integrators based on the ones in SciPy. Aim is to make them faster and much more compatible with Numba.

Table of Contents

• Quick start guide
  • The first steps
    • Installing
    • Importing
    • Example
    • Example of the advanced function

Quick start guide

Here's how you can start

The first steps

Installing

Install Numba Integrators with pip

pip install numba-integrators

Importing

Import name is not the same as install name, numba-integrators.

import numba_integrators

Example

import numba as nb
import numba_integrators as ni
import numpy as np

@nb.njit(nb.float64[:](nb.float64, nb.float64[:]))
def f(t, y):
    '''Differential equation for sine wave'''
    return np.array((y[1], -y[0]))

y0 = np.array((0., 1.))

solver = ni.RK45(f, 0.0, y0,
                 t_bound = 1, atol = 1e-8, rtol = 1e-8)

t = []
y = []

while ni.step(solver):
    t.append(solver.t)
    y.append(solver.y)

print(t)
print(y)

Example of the advanced function

import numba as nb
import numba_integrators as ni
import numpy as np

@nb.njit
def f(t, y, parameters):
    '''Differential equation for sine wave'''
    auxiliary = parameters[0] * y[1]
    dy = np.array((auxiliary, -y[0])) + parameters[1]
    return dy, auxiliary

t0 = 0.
y0 = np.array((0., 1.))
parameters = (2., np.array((-1., 1.)))

# Numba type signatures
parameters_signature = nb.types.Tuple((nb.float64, nb.float64[:]))
auxiliary_signature = nb.float64
solver_type = ni.RK45

Solver = ni.Advanced(parameters_signature, auxiliary_signature, solver_type)
solver = Solver(f, t0, y0, parameters,
                t_bound = 1, atol = 1e-8, rtol = 1e-8)

t = []
y = []
auxiliary = []

while ni.step(solver):
    t.append(solver.t)
    y.append(solver.y)
    auxiliary.append(solver.auxiliary)

print(t)
print(y)
print(auxiliary)

Changelog

0.2.1 2023-08-10

• Advanced mode solver to handle functions with parameters and auxiliary output

0.1.2 2023-08-06

• Fixes

0.1.1 2023-08-05

• Initial working version

0.0.3 2023-05-14

• Inital working state