I am trying to solve an ODE with solve_ivp, but I am getting strange errors. Documentation: https://docs.scipy.org/doc/scipy/reference/generated/scipy.integrate.solve_ivp.html

from scipy.integrate import solve_ivp as solve
import numpy as np
import matplotlib.pyplot as plt

def fun(t, y, beta, eps):
    u = 1/eps * (-1/3 * y[0]**3 + y[0] + y[1])
    v = beta - y[0]
    y[0]= u
    y[1]= v

    return y

eps = 0.05
beta = 1.1
y0 = np.array([2, 2])
t = np.linspace(0, 1000, num=10000)
Y = solve(fun=lambda t, y: fun(t, y, beta, eps), t_span=(t[0],t[-1]), y0=y0 , method='RK45', t_eval=[1,2,3,4]).y

ValueError: need at least one array to concatenate

Note that I already had to deviate from the documentation, as args argument does not work either. Instead, I had to redefine my function which gets 2 additional arguments using the lambda method. I found this here: https://stackoverflow.com/questions/48245765/pass-args-for-solve-ivp-new-scipy-ode-api

Am I doing something wrong or is this method extremely buggy?


1 Answer 1


Your issue is that you are modifying y in your function. If you return, [u,v] instead of resetting the values in y, it seems to run just fine. I believe this is because the solver will store and use intermediate values elsewhere, but your function is having the side effect of altering these stored values.

  • $\begingroup$ Thank you very much Tyberius!!! I completely forgot that Python always passes by reference. Didn't even think about this. $\endgroup$
    – reloh100
    Commented Apr 2, 2020 at 1:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.