[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Set value to a custom significance. (no display)
From: |
Joris Spekreijse |
Subject: |
RE: Set value to a custom significance. (no display) |
Date: |
Tue, 2 Mar 2010 08:16:17 +0100 |
Good morning,
Thanks for the answers.
We are trying two things for our project to convert a MatLab formula into C for
an MPU.
First thing is to see what happens when we use less bits and different fixed
point settings for value in our formula. I am going to solve this with some old
fashioned bit shifting.
Secondly we have strong (do not know the English term) significance. When a
measurement device has a significance of 3 it is wrong to have an answer with a
significance of 4 and use this result in further calculations. In that case the
rounding trick should do his work.
Yours,
Joris
-----Original Message-----
From: John W. Eaton [mailto:address@hidden
Sent: Monday, March 01, 2010 8:09 PM
To: Judd Storrs
Cc: Joris Spekreijse; address@hidden
Subject: Re: Set value to a custom significance. (no display)
On 1-Mar-2010, Judd Storrs wrote:
| First of all, the machine native floating point is base 2, not base 10
| so this is going to be an approximation. However, something like this
| may work by removing the remainder at each step:
|
| octave:1> format long
| octave:2> a = 0.499/2
| a = 0.249500000000000
| octave:3> a = round(100*a)/100
| a = 0.250000000000000
| octave:3> b = a * 2
| b = 0.500000000000000
| octave:4> b = round(100*b)/100
| b = 0.500000000000000
|
| You could use round, ceil or floor depending on how you want rounding to work.
I recently added a chop function:
http://hg.savannah.gnu.org/hgweb/octave/rev/27777bd27e71
But I have to ask: Why do you think you want to do this? What are
you trying to accomplish?
jwe