## #1056 integrate(t^(-1.2),t);

closed
nobody
5
2007-03-03
2007-01-10
Anonymous
No

integrate(t^(-1.2),t);
gives

5.000000000000001
- -----------------
0.2
t

must be

5
- -----------------
0.2
t

## Discussion

• Raymond Toy - 2007-01-10
• status: open --> pending

• Raymond Toy - 2007-01-10

Logged In: YES
user_id=28849
Originator: NO

No. Compare with 1/(-1.2+1) -> 5.000000000000001

If you wanted the exact number -6/5, you should have said so instead of using the floating point number 1.2.

Setting status to pending.

• Nobody/Anonymous - 2007-01-10

Logged In: NO

But
integrate(t^(-1.2),t,t1,t2); ("positive" for all questions)
gives true
5 5
----- - -----
1/5 1/5
t1 t2

This is a bug or feature?

• Nobody/Anonymous - 2007-01-10
• status: pending --> open

• Raymond Toy - 2007-01-10

Logged In: YES
user_id=28849
Originator: NO

Good point. I would consider this a bug in the definite integration routines.

At the very least, it should warn about converting a float to a rational.

• Raymond Toy - 2007-01-16

Logged In: YES
user_id=28849
Originator: NO

I ran this again, and I actually get warnings about converting a float to a rational.

I think this is acceptable. I think it would be better if it didn't do this conversion, though.

• Nobody/Anonymous - 2007-01-31

Logged In: NO

no, the result is correct because
(%i27) 1/(-1.2+1);
(%o27) - 5.000000000000001
you are using floating point arithmetic and therefore you have rounding errors
actually 0.2 cannot be represented as float without rounding errors (if you are using a binary representation). use exact numbers (e.g t^(-6/5) ) if you want an exact result

• Robert Dodier - 2007-03-03

Logged In: YES
user_id=501686
Originator: NO

Marking this "Rejected". Agreed w/ comments by nobody dated 2007-01-31 02:17.

• Robert Dodier - 2007-03-03
• labels: --> Problem not in Maxima

• Robert Dodier - 2007-03-03
• status: open --> closed