I demonstrated our math library at the last HPCC. It allows to plot any kind of graphs (including inequalities)
Unlike most plotters, it can plot implicit functions (in the form of g(x,y)=f(x,y)) and will always draw a "true" representation of the function. As a bonus, it can also identify where the function i undefined or discontinued and display it as such. It also allows tracing
> I demonstrated our math library at the last HPCC. > It allows to plot any kind of graphs (including inequalities)
> Unlike most plotters, it can plot implicit functions (in the form of > g(x,y)=f(x,y)) and will always draw a "true" representation of the > function. As a bonus, it can also identify where the function i > undefined or discontinued and display it as such. It also allows tracing
Wow! This is the killer app for hp50g! Well, the name, EDGE, was meant to be the name of a killer app, wasn't it? :-)
For a few days, I have been writing a SysRPL program to do range analysis. I need it to evaluate the precision of experimental data. For example, as explained in EDGE's whitepaper, say you measure 1±0.1 meters and 2±0.1 meters, i.e. you know the two lengths are in the ranges [0.9 1.1] and [1.9 2.1] (in meters). Then, the sum of them is 3±0.2 meters. Simple when it's just an addition, but much more difficult if there is a lot of algebraic manipulation e.g. with trigonometric functions. I never thought to use range analysis to speed up graphing.
Once again, EDGE seems to give very impressive results. I wish I had a hp50g to give it a try. Congratulations!
On Nov 3, 6:20 am, JYA <nos...@nospam.blah> wrote:
> The code for the HP50 isn't publicly available yet.
> Jean-Yves
Looks great. Can't wait to try it out.
In your paper, you mention that equations like rotated ellipses and lemniscates are not part of senior-high curriculum. Actually, we do cover these very shapes.
Some years back (okay, back in the MS-DOS days actually), I wrote a graphing program for school use that plotted implicit functions with recursive subdivision as you described. Even on my 286-12, it could plot 640x480 in just a few seconds. I got the idea of using recursive subdivisions from the Fractint program which used something similar (along with other speed-ups) to generate fractals in amazingly short times.
When I got my 49g+, I was thrilled that it could do truth plots, but when I saw the pixel-by-pixel method it used, I realized that it was practically unusable.
JYA wrote: > On 2007-11-03 07:44:48 +1100, sc_use...@hotmail.com said:
>> Wow. Sounds incredible. Where can I get it for the 50g? (or is it >> still in beta testing?)
> The code for the HP50 isn't publicly available yet.
> Jean-Yves
Wow. I'm completely impressed. Will the demo page be staying up for the foreseeable future? I'd like to pass the link along to some friends, but it'd be frustrating for them if it disappeared. Cool stuff!
The fact that you designed this to run on the 50g is really exciting. This is truly the first calculator app I'd pay for if it works as well as you say it does. (and, judging by the demo page, it looks like it does)
So, count me as one of the first in line to download/buy whenever it comes out!
On 2007-11-04 02:17:43 +1100, "Greg M." <grego...@insightbb.com> said:
> Wow. I'm completely impressed. Will the demo page be staying up for the > foreseeable future? I'd like to pass the link along to some friends, but > it'd be frustrating for them if it disappeared. Cool stuff!
yes, we have no plan to remove the online version in the short term future... we will probabl add features such as tracing (which the library can do already)
> The fact that you designed this to run on the 50g is really exciting. This > is truly the first calculator app I'd pay for if it works as well as you > say it does. (and, judging by the demo page, it looks like it does)
it wasn't designed to run on the hp50 per say, but on any embedded platform with restricted CPU power and memory. Traditional algorithm made in unpractical until now.
We hope to bring such plotting capabilities to any calculators as such it will require a lots of talks with the concerned parties. The ability to identify undefined area and discontinuities is I believe a major feature and would make a huge difference to students. it would likely require a curriculum change however as edge can do in seconds what used to take a lot of time. teachers may not like it as identifying where the function is undefined is often the major theme of an exercise.
Jean-Yves
-- They who would give up an essential liberty for temporary security, deserve neither liberty or security (Benjamin Franklin)
On 2007-11-04 00:39:27 +1100, Wes <wjltemp...@yahoo.com> said:
> In your paper, you mention that equations like rotated ellipses and > lemniscates are not part of senior-high curriculum. Actually, we do > cover these very shapes.
those are actually the only kind of implicit functions studied, but they are barely scratching the surface of it. I believe it is due to the lack of tools to study those (especially on calculators)
> Some years back (okay, back in the MS-DOS days actually), I wrote a > graphing program for school use that plotted implicit functions with > recursive subdivision as you described. Even on my 286-12, it could > plot 640x480 in just a few seconds. I got the idea of using recursive > subdivisions from the Fractint program which used something similar > (along with other speed-ups) to generate fractals in amazingly short > times.
as per our white paper, recursive subdivision doesn't work for all type of implicit functions, as such its use is limited. it also works better for inequalities. it would also fail to determine in a set manner if an area is undefined or not and would leave the difficult area unexplored. EDGE will always find a solution in a short time, this is guaranteed.
> When I got my 49g+, I was thrilled that it could do truth plots, but > when I saw the pixel-by-pixel method it used, I realized that it was > practically unusable.
truth plot can only do inequalitiies, it's hopeless. HP calculators are the worse when it comes to advance plotting
TI and Casio are using the zero contour technique, which has a complexity of O(N^2)... but at least they can do it... it's just damn slow!
-- They who would give up an essential liberty for temporary security, deserve neither liberty or security (Benjamin Franklin)
On Nov 3, 11:28 am, Khanh-Dang <khanh-d...@w.fr.invalid> wrote:
> say you measure 1±0.1 meters > and 2±0.1 meters, i.e. you know the two lengths are in the ranges [0.9 > 1.1] and [1.9 2.1] (in meters). Then, the sum of them is 3±0.2 meters.
I'm not sure if this applies to your experimental data, but usually when doing calculations with data, the uncertainties don't just add. For instance, in your example (1 +- 0.1) + (2 +- 0.1) = (3 +- 0.14). In general, (A+-a)+(B+-b) = (A+B +- sqrt(a^2+b^2)).
On my 50g, I have routines for doing common operations (+-*/^) with uncertainties, so that doing calculations with uncertainties is essentially no more difficult than doing regular calculations (thanks to RPN)
{1. 0.1} {2. 0.1} ?ADD
instead of
1. 2. +
If the uncertainty in your data has a uniform distribution instead of the more typical normal distribution, then your result of (3 +- 0.2) is correct.
> When I got my 49g+, I was thrilled that it could do truth plots, > but when I saw the pixel-by-pixel method it used, > I realized that it was practically unusable.
On Sat, 03 Nov 2007 10:59:37 -0700, Wes <wjltemp...@yahoo.com> wrote: >On Nov 3, 11:28 am, Khanh-Dang <khanh-d...@w.fr.invalid> wrote:
>> say you measure 1±0.1 meters >> and 2±0.1 meters, i.e. you know the two lengths are in the ranges [0.9 >> 1.1] and [1.9 2.1] (in meters). Then, the sum of them is 3±0.2 meters.
>I'm not sure if this applies to your experimental data, but usually >when doing calculations with data, the uncertainties don't just add. >For instance, in your example (1 +- 0.1) + (2 +- 0.1) = (3 +- 0.14). >In general, (A+-a)+(B+-b) = (A+B +- sqrt(a^2+b^2)).
>On my 50g, I have routines for doing common operations (+-*/^) with >uncertainties, so that doing calculations with uncertainties is >essentially no more difficult than doing regular calculations (thanks >to RPN)
> {1. 0.1} {2. 0.1} ?ADD
>instead of
> 1. 2. +
>If the uncertainty in your data has a uniform distribution instead of >the more typical normal distribution, then your result of (3 +- 0.2) >is correct.
>-wes
If the absolute errors are much smaller than the measurements, in addition or subtraction they add. It is only when the absolute errors are of comparable magnitude to the measurements, that the square of the errors need to be used. In multiplication and division, it is the relative errors are added. Harold A. Climer Dept. Of Physics, Geology, and Astronomy University of Tennessee at Chattanooga 309 Grote Hall 423-425-4546 Harold-Cli...@utc.edu
> I demonstrated our math library at the last HPCC. > It allows to plot any kind of graphs (including inequalities)
> Unlike most plotters, it can plot implicit functions (in the form of > g(x,y)=f(x,y)) and will always draw a "true" representation of the > function. As a bonus, it can also identify where the function i > undefined or discontinued and display it as such. It also allows tracing
It includes some trig function , the ability to report bugs (if any) and features some UI improvements like rounding the cursor coordinates and the ability to change how the axes are drawn so it doesn't hide the generated graph
> It includes some trig function , the ability to report bugs (if any) and > features some UI improvements like rounding the cursor coordinates and > the ability to change how the axes are drawn so it doesn't hide the > generated graph
> > It includes some trig function , the ability to report bugs (if any) and > > features some UI improvements like rounding the cursor coordinates and > > the ability to change how the axes are drawn so it doesn't hide the > > generated graph
In our ongoing quest to perfection, the Hydrix team have continued working on our EmbeDded Graphic Engine.
It now works "a la" Google Map. Much more dynamic you can drag the graph around ; double click on a point to center etc..
You can also click on the various examples to directly generate the graph. And finally, it is possible to get a direct link on the graph you've generated, so you can post here some cool graph.
Have a play and let us know how you go ..
Cheers Jean-Yves
-- They who would give up an essential liberty for temporary security, deserve neither liberty or security (Benjamin Franklin)
On Fri, 04 Apr 2008 12:57:13 -0500, JYA wrote: > Dear all.
> In our ongoing quest to perfection, the Hydrix team have continued > working on our EmbeDded Graphic Engine. > Have a play and let us know how you go ..
> On 2007-11-04 02:17:43 +1100, "Greg M." <grego...@insightbb.com> said:
>> Wow. I'm completely impressed. Will the demo page be staying up for the >> foreseeable future? I'd like to pass the link along to some friends, but >> it'd be frustrating for them if it disappeared. Cool stuff!
> yes, we have no plan to remove the online version in the short term > future... > we will probabl add features such as tracing (which the library can do > already)
>> The fact that you designed this to run on the 50g is really exciting. >> This >> is truly the first calculator app I'd pay for if it works as well as you >> say it does. (and, judging by the demo page, it looks like it does)
> it wasn't designed to run on the hp50 per say, but on any embedded > platform with restricted CPU power and memory. > Traditional algorithm made in unpractical until now.
> We hope to bring such plotting capabilities to any calculators as such it > will require a lots of talks with the concerned parties. > The ability to identify undefined area and discontinuities is I believe a > major feature and would make a huge difference to students. it would > likely require a curriculum change however as edge can do in seconds what > used to take a lot of time. teachers may not like it as identifying where > the function is undefined is often the major theme of an exercise.
> Jean-Yves
> -- > They who would give up an essential liberty for temporary security, > deserve neither liberty or security (Benjamin Franklin)
|
