Log in

View Full Version : Software That Does...What?


Brad Adrian
05-03-2003, 09:00 AM
<div class='os_post_top_link'><a href='http://www.handango.com/PlatformProductDetail.jsp?siteId=311&platformId=2&productType=2&catalog=0&sectionId=0&productId=68399' target='_blank'>http://www.handango.com/PlatformPro...productId=68399</a><br /><br /></div>Every now and then I come across a Pocket PC application whose description is so self-explanatory that I feel compelled to try it out. I'll let this description of Lisa for PocketPC 2002 by Poliplus Software speak for itself:<br /><br /><img src="http://www.pocketpcthoughts.com/images/adrian/lisa.gif" /><br /><br />"Lisa is a "classical Feedback Control Systems" application designed to handle typical calculations in the 's' and 't' domains...Lisa supports Bode Analysis and Design, Root Locus Analysis and Design and Time analysis."<br /><br />Features include:<br /><br /> Line approximation for phase and magnitude plots that allows users to adjust an existing transfer function or completely design one from scratch.<br /><br /> Line approximation of root locus allowing the user to drag, add or delete poles and zeros and instantly see a line approximation sketch featuring : poles, zeros, asymptotes, angles of arrival and departure, real axis points and breakaway points.<br /><br /> Provides symbolic functions for Laplace transform and its inverse. This allows users to modify time equations and see its effect in the frequency domain or vice-versa.<br /><br />I guess my only remaining question is: HUH?

pro_worm
05-03-2003, 09:36 AM
Looks like some advanced calculus to me - s and t are variables speed and time. while line approx. are used to predict where a curve (a graph) will go at some point; for example, if you know what it looks like from 1 to 4 seconds, line approx. will estimate what it looks like at 5 seconds using a form of oiler's method. What the bullet points are referring to then is the ability to create graphs through existing functions, or (as the second bullet indicates) by adding or deleting "dots." Zeros are where the function intercepts the x-axis, you all know what asymtotes are, and angle of departure and arrival indicate the angle at which the curve hits its ending and beggining.

The last bullet point is referring to the ability to change the variable of time to see its effects on the x-axis "domain" of the graph.

Basically, its a specialized graphing calculator.

:wink: And I should know, because I have an exam on this on Thursday :evil:

Joff
05-03-2003, 10:53 AM
Hi Conrad,

Just a small correction, I think the s-domain refers to the Laplace transform. And the t-domain probably refers to the time domain.

I wouldn't like you to fail your exam :wink:

rave
05-03-2003, 11:39 AM
Feedback control systems is a subject matter usually taken up in electrical engineering, but has applications in other fields as well.

Suppose we have a system that controls something, say, a cruise control system. For this particular system, we'd be interested in a number of things, like:

* how fast the system adjusts your car to be running at the correct speed, or

* how far off your actual speed is from the value that you have set, or

* whether the car actually reaches the target speed, or does it just speed up, then slow down, then speed up, then slow down, but never hitting the mark exactly.

You can actually predict these and other system properties by making a mathematical model and running it through some evaluations, or "tests". These "tests" are the root locus, Nyquist plots and Bode plots, among others. What is usually done is to plot these graphs and judge system performance from the shape of the plot. These things are exactly what this piece of software helps you to do. :)

bjornkeizers
05-03-2003, 01:33 PM
Looks like some advanced calculus to me - s and t are variables speed and time. while line approx. are used to predict where a curve (a graph) will go at some point; for example, if you know what it looks like from 1 to 4 seconds, line approx. will estimate what it looks like at 5 seconds using a form of oiler's method. What the bullet points are referring to then is the ability to create graphs through existing functions, or (as the second bullet indicates) by adding or deleting "dots." Zeros are where the function intercepts the x-axis, you all know what asymtotes are, and angle of departure and arrival indicate the angle at which the curve hits its ending and beggining.

Or, as he said:

HUH?

See, now this I understand..

You can actually predict these and other system properties by making a mathematical model and running it through some evaluations, or "tests". These "tests" are the root locus, Nyquist plots and Bode plots, among others. What is usually done is to plot these graphs and judge system performance from the shape of the plot. These things are exactly what this piece of software helps you to do.

Uh huh. Im sure this is the best piece of software in its category :lol:

Probably the only one in that category too 8O

JimDantin
05-03-2003, 01:34 PM
It's written in Java -- no wonder I don't understand it!

Ed Hansberry
05-03-2003, 01:39 PM
Hi Conrad,

Just a small correction, I think the s-domain refers to the Laplace transform. And the t-domain probably refers to the time domain.
You beat me to the punch Joff. That is exactly what I was going to say. Really!

gliscameria
05-03-2003, 03:13 PM
I really don't think any of that matter, considering it will probably run entirely too slow for anyone to use. (looks to be based on Formulae 1, which was SLOOOW, slow like doom2 on a ps/2)

David Johnston
05-03-2003, 04:00 PM
Feedback control systems is a subject matter usually taken up in electrical engineering, but has applications in other fields as well.

Suppose we have a system that controls something, say, a cruise control system. For this particular system, we'd be interested in a number of things, like:

* how fast the system adjusts your car to be running at the correct speed, or

* how far off your actual speed is from the value that you have set, or

* whether the car actually reaches the target speed, or does it just speed up, then slow down, then speed up, then slow down, but never hitting the mark exactly.

You can actually predict these and other system properties by making a mathematical model and running it through some evaluations, or "tests". These "tests" are the root locus, Nyquist plots and Bode plots, among others. What is usually done is to plot these graphs and judge system performance from the shape of the plot. These things are exactly what this piece of software helps you to do. :)

Well, duh!

BugDude10
05-03-2003, 04:07 PM
Well, FINALLY! This is probably the KILLER APP I've been waiting for! This, and something for me to track my biorhythms.

malcolmsharp
05-03-2003, 04:16 PM
OK, guys... it's not like I needed proof that some high level GEEKS hung out here or anything!

Math geeks... sigh... give me PDA and comic book geeks ANY day of the week, but not MATH geeks!

Jason Dunn
05-03-2003, 04:44 PM
OK, guys... it's not like I needed proof that some high level GEEKS hung out here or anything!

Math geeks... sigh... give me PDA and comic book geeks ANY day of the week, but not MATH geeks!

Don't worry, I'm the furthest thing from being a math geek - I nearly failed math in high school and never took any math past that. :mrgreen:

Bearjaw
05-03-2003, 05:03 PM
Bunch of Geeks! LOL

trachy
05-03-2003, 05:15 PM
Until they incorporate flux capacitor support into this thing, it's worthless to me. ;-)

srs
05-03-2003, 05:18 PM
somebody really needs to port maxima (http://maxima.sourceforge.net/) to the ppc.[/url]

szamot
05-03-2003, 05:35 PM
WOW this is a great topic, I can read all the words and understand them because they are in English but I still have not clue what you guys are on about. Make that duh on my end. I am out of here, math gives me the shakes and conjures up memories best colored by Prozac.

JonMisurda
05-03-2003, 06:01 PM
somebody really needs to port maxima (http://maxima.sourceforge.net/) to the ppc.[/url]

It already has been: http://www.rainer-keuchel.de/wince/maxima-ce.html

Jon

opus
05-03-2003, 06:25 PM
One way to grok control theory is by thinking about your house thermostat.

If you have a device that can instantly inject arbitrary amounts of heat or cold into your house, you have an easy feedback system. Too cold? Just add the right amount of heat. Too hot? Add some cold. Writing a program to control the house's temperature is trivial.

Now think of the actual system. All we can do is either turn the furnace on full blast, or turn it off. Furthermore there's a five minute lag between turning on the furnace and feeling the heat it generated. Also, once the furnace goes either on or off, it has to stay that way for a couple of minutes.

With a messy feedback system it's easy to create a broken control program that would send the temperature spinning out to the extremes. It turns out that you can characterize feedback systems like this with mathematical models (in your handy PPC), which can tell you how to make a control program that will give a stable system.

Any time you have a sensor, a control, and a lag between adjusting the control and seeing the result with the sensor, i.e. almost any computer-controlled mechanical device, you can analyze it with control theory.

MikeUnwired
05-03-2003, 06:37 PM
Now a reason to stick with PPC's over Palm! Lisa is the type app that could really drag me away from Bejeweled.

Can this thing tell you how many cups of sugar it takes to get to the moon?

B
05-03-2003, 08:21 PM
Until they incorporate flux capacitor support into this thing, it's worthless to me. ;-)

If you are really interested, check out http://www.DeepDMC.com (http://www.DeepDMC.com/) - I'm sure you could adapt it to be controlled from your PocketPC. Just add your own locomotive (http://www.pocketpcthoughts.com/forums/viewtopic.php?p=108344) and who knows when you'll end up. :D

kuyars
05-03-2003, 08:57 PM
Actually, this is such a coincidence. I'm a junior studying Computer Engineering and I just had a quiz on plotting these graphs last friday!

Just a correction, s refers to the frequency domain, while t refers to the time domain. Just as how you can measure how long something takes to do something (time domain), you can similarly measure how many cycles in can perform in a certain time (frequency domain).

What this program basically does is, suppose you have a circuit and you apply a voltage to it, called Vin. Now, you measure the voltage coming out of the circuit, let's call that Vout.

Now the ratio of Vout / Vin is called the transfer function. It's basically a ratio of what the circuit outputs versus what it inputs. These are normally used in amplifiers, filters, etc. Think about it: an amplifier takes a signal coming in, increases it, and then something larger comes out. The transfer function is graphed in an industry standard form, called a Bode Plot. Where the x-axis is the frequency, and the y-axis is the magnitude of the transfer function.

So a Bode plot is just a way for electrical engineers to quickly see the behavior of a circuit!

But I suppose the correct explanation should have been: do you all care? :wink:

deich
05-03-2003, 09:52 PM
I would have killed for this when I was in school studying Electrical Engineering. Bode plots are great for control systems. They are also great to understand how stable amplifiers are. You wouldn't want to amplifier to oscillate would you? (in othe words, screech) Bode plots also give you insight on how you might get a little more performance out of your amplifier (more base anybody.? Or clearer highs?)

Mathematicians would say these are linear differential equiations. But being engineers, we use Laplace Transforms to turn these differential equations into simple algebra. And, if you're an old timer, you can sketch a pretty accurate bode plot with nothing more than paper, pencil, and a 4-function calculator (or even a slide rule). Sure it's a tough job, but somebody has to do it.

Actually, I haven't done any of this in years. The IC guys took most of the hardware design a long time ago, leaving the rest of us to hack software. But even hacking software can be fun.

Nellwaskilled
05-03-2003, 10:00 PM
somebody really needs to port maxima (http://maxima.sourceforge.net/) to the ppc.[/url]

ehrrr...
it was done almost 2 years ago, including port of Gnuplot .

http://www.rainer-keuchel.de/wince/maxima-ce.html

Brad Adrian
05-03-2003, 10:36 PM
Can this thing tell you how many cups of sugar it takes to get to the moon?
Yes.

Dom
05-03-2003, 11:37 PM
What an app ! Can somebody please now port "Crocodile Clips" from the Desktop to Pocket PC ? If you never saw it try a search on Google. An amazing physics program that let you build things "virtually" then see the results. There's an inventor inside all of us :0))
Dom

Nellwaskilled
05-03-2003, 11:57 PM
What an app ! Can somebody please now port "Crocodile Clips" from the Desktop to Pocket PC ? If you never saw it try a search on Google. An amazing physics program that let you build things "virtually" then see the results. There's an inventor inside all of us :0))
Dom

Have you try this below?
without the pretty graphic, but acomplish the simulation just the same.

http://www.swiss.ai.mit.edu/~jaffer/SIMSYNCH

Use this version of SCM, it seems to have the cleanest instal
http://www.symbolictools.de/public/pocketconsole/applications/scm/index.htm

Deslock
05-04-2003, 12:09 AM
I bet that would've come in handy back when I took Controls (Mechanical Engineering class that was a big pain in the ass).

In addition to opus' thermostat, another common controls/feedback example is cruise control in your car. The car's "controller" (computer) needs to monitor the car's speed to determine whether or not to add fuel... basically a feedback loop is used to do this. BTW, one of the reasons you can't use cruise control at low speeds is that there is a somewhat narrow vehicle speed range that the feedback loop is accurate for (at least that was the case back when I studied the subject 6 or 7 years ago).

Anyway, this program deals with the higher math that engineers/programmers use to create systems like that. It's actually really interesting stuff... well, when you're not stressing over being tested on it anyway :wink:

lurch
05-04-2003, 12:53 AM
Man I could've used this back in school too... I remember doing Laplace transforms and Diff. Eq's that took up 4-8 pages... oh yeah! :)

This would've been very handy.

deich
05-04-2003, 12:57 AM
I have to admit I haven't done it, but it can't be very hard to make a cruise control that would work well at lower speeds. Today, they're based on embedded computers (like everything else), so there's no trouble adapting the control parameters to the car's speed. The challenge would be making it work well in lower gears, where a small change in gas has a big change in velocity. (As anybody knows who has a stick -- it's just harder to drive in 1st gear).

I think there's another reason. Cruise control is just a bad idea at low speeds. I had a buddy who put an ad-on unit on his car. He could cruise at 55, then stop for a stop light. When the light turned green, he pressed "resume" and let the cruise control peal out. It was really cool. And really stupid.

ziggy
05-04-2003, 02:27 AM
somebody really needs to port maxima (http://maxima.sourceforge.net/) to the ppc.[/url]

Does Maxima no longer have pictures of babes in scant clothing?

ziggy
05-04-2003, 02:28 AM
Wait, I just figured the whole program out! The answer is 42.

ctmagnus
05-04-2003, 02:41 AM
&lt;sigh> So many engineers, so little oxygen. :wink:

xoiph
05-04-2003, 07:15 PM
Just a correction, s refers to the frequency domain, while t refers to the time domain. Just as how you can measure how long something takes to do something (time domain), you can similarly measure how many cycles in can perform in a certain time (frequency domain).

Actually, you're all wrong. S refers to the number of Skittles you can eat in one minute, while T refers to the speed of light while mumbo jumbo blah blah blah... :mrgreen:

Kirkaiya
05-04-2003, 07:27 PM
One way to grok control theory is by thinking about your house thermostat.

If you have a device that can instantly inject arbitrary amounts of heat or cold into your house, you have an easy feedback system. Too cold? Just add the right amount of heat. Too hot? Add some cold. Writing a program to control the house's temperature is trivial.

Now think of the actual system. All we can do is either turn the furnace on full blast, or turn it off. Furthermore there's a five minute lag between turning on the furnace and feeling the heat it generated. Also, once the furnace goes either on or off, it has to stay that way for a couple of minutes.

With a messy feedback system it's easy to create a broken control program that would send the temperature spinning out to the extremes. It turns out that you can characterize feedback systems like this with mathematical models (in your handy PPC), which can tell you how to make a control program that will give a stable system.

Any time you have a sensor, a control, and a lag between adjusting the control and seeing the result with the sensor, i.e. almost any computer-controlled mechanical device, you can analyze it with control theory.

This was a geat explanation for laypeople.

Reading thru the product description on the homepage here (ppcthoughts) almost gave me the shivers, particularly the Nyquist plots... lol.. As part of my engineering degree (mechanical) we had to take a class in control systems.

Anyway - this program seems aimed at using non-symbolic calcuations to display the results of changing variables in a feedback control system.

Performing LaPlace transforms (ughh.. talk about bad memories) without resorting to numerical methods (ie., solving the equations using numbers, which are actually an estimation, rather than by manipulating the symbols of the equation) and then plotting the results is a pretty neat trick !

Luckily for me, I hated that class enough that I have repressed most memory of it. Oh - and for those who were wondering at the labels, "Imaginary" and "Real" on the quadrants in the animated GIF, those are referring to whether one of the components of the point (which as x & y coords, or s and t, whatever you wish) has the Square-root of -1 as a factor (some solutions to simple math problems come in the form of SQR(-1) * k, where k is any constant).

I kinda miss working with the vector math - it's all fuzzy to me now... my memory is like a damn sieve, and the higher-level math has leaked out the most.

deich
05-04-2003, 08:01 PM
Nice comment about imaginary numbers. Here's another way to think about it. Visualize the number line. Positive numbers to the right. Negative numbers to the left. Now imagine a positive vector of some length. You can convert that vector from positive to negative by rotating it 180 degrees. So multiplying by -1 is the same as 180 degree rotation.

Imaginary numbers ("i", or "j" if you are an EE) are rotated 90 degrees. i * i = -1. Or, rotating a positive vector 90 degrees makes it imaginary. Rotating it another 90 degrees makes it negative.

If you're an EE, imaginary numbers are very useful because they can be used to understand the AC phase shift between voltage and current (or power factor).

I'm amazed by how many "scientific calculators" don't support imaginary numbers. It's one reason I like Calc98.

lurch
05-04-2003, 10:02 PM
Hmm. I think this thread was used to show off how smrt everybody is.

I am so smart! SMRT!

rave
05-05-2003, 02:55 AM
Well, now that we know that they do this stuff for the PPC, can I now assume it's not too much to ask for a MATLAB/Octave clone for the PPC? :)

Nellwaskilled
05-05-2003, 03:58 AM
you just have to start investigating which developer is willing to do it i guess.

Brad Adrian
05-05-2003, 04:28 AM
&lt;sigh> So many engineers, so little oxygen. :wink:
LOL!!! :D :D :D

Landis
05-05-2003, 02:58 PM
will estimate what it looks like at 5 seconds using a form of oiler's method.
That's one way to remember how to pronounce Euler :)

I really admire how this stuff seems to come so easy to all you math geeks. I can see the usefullness, even the beauty of mathmatics but it often makes me feel like a moron :(

Guess I'll just have to be content as a sideline fan :wink:

s_p_kay
05-05-2003, 08:37 PM
Doesn't anybody study electrical engineering anymore? :(
This app is great for students taking Controls, Adv. Electronics, etc.
where you are required to do these plots regularly. :D

Kirkaiya
05-05-2003, 11:44 PM
You know what I think somebody ought to do with all the storage capacity of SD and CF cards, and a 400 MHz processor?

Somebody should create an improved version of "Eliza", or a similar conversational AI, so that when we're all alone, and bored, we'd have somebody to talk to!!

Remember Eliza? You could say things like, "Do you like baseball?", and it would be all, "do you think my liking baseball is important? Lets talk about your mother"

:lol:

What's the state of AI anyway? When are we going to see Wintermute (from the Neuromancer-series, by William Gibson)? i just read an interesting book by Vernor Vinge last week, called, "Fire upon the Deep", in which superintelligent AI's (or entities that resulted from collective minds of normal intelligences) are called, "Powers".

hmm... I wonder when my PocketPC officially becomes smarter than me? lol

lurch
05-07-2003, 10:49 PM
hmm... I wonder when my PocketPC officially becomes smarter than me? lol
I think it already did... 8)


sorry, couldn't resist. :) (you know I'm kidding)

deich
05-08-2003, 03:05 AM
Doesn't anybody study electrical engineering anymore?

Given the state of jobs in EE, I wouldn't recommend it. I definately don't want my son studying EE.