|
> > ACK! Don't they teach fixed point any more? I encourage anyone to post an
> > application where the dynamic range of signed 32 bit values is outside the
> > range of numbers which would be useful to an RCX application!
> >
> > I'm not flaming, I'm just trying to figure out why eveyone is soooo fixated
> > on floating point.
>
> Well, we are trying to teach AI, not embedded systems :) Towards that
> end, we've sort of tried to simplify things as much as possible for the
> kids who are taking the class, since most of them will probably never do
> an embedded system again. That means doing things (like floats) that are
> commonly available on "real" machines. I think at least one person will
> tackle the task of converting the work to fixed point (among other
> things) to reduce processor time and overhead as a final project. I'll
> keep you up to date.
> -Luis
OK, time to get out the soapbox. If you are trying to teach AI, and are
using floating point math, then you probably have really fast computers
with floating point processors that are sometimes quicker than equivalent
fixed point calcs......
I do hope that the students are getting some numerical methods banged into
their heads, otherwise how can they accurately asses the errors in their
AI algorithms?
Which brings me back to fixed point. Once you show someone the simplicity
and elegance of simple, small, fixed point numbers, they may be able to
do the complex AI algorithms on embedded machines.
If they don't learn fixed point, they may see a simple 8-bit fixed point
processor and say it can't do AI because it has no FP numbers....
Just my 2cents, any other views out there on engineering education?
Cheers,
Ralph Hempel - P.Eng
--------------------------------------------------------
Check out pbFORTH for LEGO Mindstorms at:
<http://www.hempeldesigngroup.com/lego/pbFORTH>
--------------------------------------------------------
Reply to: rhempel at bmts dot com
--------------------------------------------------------
|
|
Message has 3 Replies: | | RE: AI and even more exiciting stuff
|
| (...) Depends exactly what you're teaching in the class, and how much time you have. Spending a session teaching fixed point math may well mean that you have to miss out a useful AI tool. If the students already understand how to call, say (...) (25 years ago, 18-Oct-99, to lugnet.robotics)
| | | Re: AI and even more exiciting stuff
|
| (...) Here in Karlsruhe, engineering education is 95% math in the first year, and at least 60% in the second. Never mind your major - you're doing maths. Linear algebra, real and complex analysis of multiple variables, numerics, stochastics, (...) (25 years ago, 18-Oct-99, to lugnet.robotics)
| | | Re: AI and even more exiciting stuff / off-topic
|
| (...) Ah, I almost forgot. I saw an applet somewhere on the University network that translates some higher language, possibly C or a subset of Java, into output suitable for universal turing machines. Not quite the performance king, but interesting. (...) (25 years ago, 18-Oct-99, to lugnet.robotics)
|
Message is in Reply To:
16 Messages in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|