# Indefinite integral is incorrect

`indefinite_integral(sqrt(1+cos(x)**2), x).full_simplify()`

gives `1/6*sin(x)^3`

, which is incorrect.

Indefinite integral is incorrect

`indefinite_integral(sqrt(1+cos(x)**2), x).full_simplify()`

gives `1/6*sin(x)^3`

, which is incorrect.

3

This is a bug. Furthermore, it is *not* a `maxima`

bug, as it is often the case.
Here, Sage truly screws things up itself:

Maxima doesn't give a false answer:

`sage: maxima.integrate(sqrt(1+cos(x)^2),x).sage() integrate(sqrt(cos(x)^2 + 1), x)`

When one tries to "ease" the problem,

maxima doesn't recognize the "obvious", but does not give a false answer:

`sage: maxima.integrate(sqrt(1-m*sin(x)^2),x).sage() integrate(sqrt(-m*sin(x)^2 + 1), x)`

Sage does:

`sage: integrate(sqrt(1-m*sin(x)^2),x) 1/4*m*x - 1/8*m*sin(2*x)`

BTW: what is expected:

```
sage: elliptic_e(x,1/2).diff(x)
sqrt(-1/2*sin(x)^2 + 1)
```

One can easily check that `sympy`

, `giac`

and `fricas`

all fail to integrate,
but that none of them gives misleading "answers".

This one does not seem to be related to existing indefinite integral bugs, and is an original, genuine, Sage-specific one. Reported as Trac #26563.

Asked: **
2018-10-26 07:08:37 -0600
**

Seen: **119 times**

Last updated: **Oct 27 '18**

Error integrate with Fricas as algorithm fresnel_sin called

Segmentation fault when multiplying by variable

Stack overflow in boolean test

why I can not get a numerical answer by show(integral(x^x,x,1,2))

Free nilpotent Lie algebra of step 3 with 11 generators

Copyright Sage, 2010. Some rights reserved under creative commons license. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.

Where did you get the

`indefinite_integral`

function ?in 8.5.b0:

@tmonteli, I use sage 8.1

Using integral(sqrt(1+cos(x)**2), x).full_simplify() as suggested still results in an answer which is incorrect, doesn't it? This is a nonelementary integral. See here or here