Engineering Physics Labs (errors)

Peleus

Ok, just a quick question here about the physics labs. Does anyone get how to fill in the analysis of accuracy and uncertainty section of the report? For the last two i just made up some random stuff which i knew was wrong. There is some formula we have to use to work out percentage error etc. but i don't know what it is cos it was never explained to us. What did yous put in this section?

ps:/ any JF engineers get that lol email from me? apparently we can use the engineering mailing lists.

Jonathan

Peleus wrote: »

ps:/ any JF engineers get that lol email from me? apparently we can use the engineering mailing lists.

Dont use the mailing lists.. People will dislike you immensely for sending crap around.

Actually.. funny story. One of the guys last year sent the autocad project that we were meant to send to a different email address to the whole class

uprooted shane

happined a few times over the year actually!

PurpleFistMixer

Dunno if your labs are the same as mine in TP, but if you got the lab book there's loads of stuff about error calculation in it (page 17 or something if I recall)... you mean stuff like finding the error in something which is a product and all that?
As far as filling it out goes, lord knows. *looks angrily at her empty lab report in the corner* Think you just write stuff like "the error in the x quantity was found to be y by the formula", then include some workings or something... Dunno really.

TheAmateur

Peleus wrote: »

Ok, just a quick question here about the physics labs. Does anyone get how to fill in the analysis of accuracy and uncertainty section of the report? For the last two i just made up some random stuff which i knew was wrong. There is some formula we have to use to work out percentage error etc. but i don't know what it is cos it was never explained to us. What did yous put in this section?

I've no idea either, someone tried to explain it to me (apparently you use standard deviation on all your experimental values or something) so I just put plus or minus 1mm or ohm or whatever I was measuring... jjust to fill the space really.

Peleus wrote: »

ps:/ any JF engineers get that lol email from me? apparently we can use the engineering mailing lists.

Yeah i just got your email back there, what other kind of replies did you get?

Podge_irl

There are proper ways to do it and I sometimes put a fair amount of time into it. Its quite easy to miscalculate errors when you are running experimentally recorded values through a load of functions. How hard they penalise you for getting it wrong is highly dependent on the grader. Other times I simply made up the section (the experimental error in made-up results is zero anyway....).

devinejay

I can see how mailing lists can be abused but today, I lol'd, and I'm not ashamed to say it!

Edwardius

You have a manual. Read it and stop being lazy. There's a section on accuracy and uncertainties. I (like most others) was giving 2 marks out of ten for this section last year, which makes it fairly important. You're fooling yourself if you make up the errors.

Boston

I remember the error calculation bit was a simple concept once I thought about it. Something about relating to the distance between samples in your sample set. The formula in the book was a bit head wrecking though!

About the mailing list, you can be removed from them if someone complains.

Peleus

Dead Ed wrote: »

You have a manual. Read it and stop being lazy. There's a section on accuracy and uncertainties. I (like most others) was giving 2 marks out of ten for this section last year, which makes it fairly important. You're fooling yourself if you make up the errors.

I tried reading the manual but the formula they give just doesn't make sense to me. i'll try reading it again...

ApeXaviour

Like Dead_ed and boston I do some demonstrating for labs. There are different types of uncertainty. Instrumental, statistical counting etc. As Ed said, it's in the book and it's not that difficult a concept, if you're in university doing engineering we can assume you've got at least some penchant for maths, and this is pretty basic stuff so either there's a mental block going on or it's laziness.


Say you have x +/- Δx, and y +/- Δy

If you want to add the values, you add the errors. If you want to subtract the values, you still add the errors (think about it...).

If you want multiply or divide then you must work out the percentage or proportion of each error as compared to the reading. An error of 100 isn't that much when your value is 1,000,000. So you divide the error by its the reading Δx / x, and add it to your the proportion of your other reading Δy / y and you'll get Δz / z, where z = x * y, or in the case of division z = x / y. Δz is your error in z, in case that wasn't apparent...

For more complicated expressions. Say you needed to work out the the error for z, where z = m^2 (m squared). You'll need to use the a formula

if a = f(x), then Δa = Δx * df(x)/dx.

So in that case, Δz = Δm * 2 * m

Make sense?

Boston

so x = 3000 y =2000 error in both = +/- 5

Δy / y = 0.0025
Δx / x = 0.0017

z = x*y = 600,0000
Δz / z = (Δy / y ) + (Δx / x) = .0042
Δz = 600,0000 * .0042 = 25200

Correct?

ApeXaviour

Correct. Except to have errors in excess of one significant figure is a bit meaningless. If the main value is inaccurate to 25200, then how can the error be accurate to 3 figures? That 5200 should be rounded off.

So the answer would be better described as:
6,000,000 +/- 30,000

Or more commonly in scientific notation:
6*10^6 +/- 0.03*10^6


The same removal of significant figures is done in the readings. If you'll allow me Boston, to waste some time and alter your example slightly

x = 3136 y =2017 error in both = +/- 5

Δy / y = 0.0025
Δx / x = 0.0016

z = x*y = 6,325,312
Δz / z = (Δy / y ) + (Δx / x) = 0.0041
Δz = 6,325,312 * 0.0041 = 25,934

Again you round off the error to 30,000, but because this is the limit of the accuracy, the value itself should be rounded off too. Your answer in this case becomes:

6,330,000 +/- 30,000
or
6.33*10^6 +/- 0.03*10^6

Brian

Since you guys seem to be having so much fun, would you mind doing mine for me?

Boston

Very true apex. Excellent example.


Baza: Me and crash already did your CS assignment.

Brian

That wasn't the assignment, it was just me messing around. The hard part of the assignment is
1. Working with someone else
2. Doing stuff we haven't covered [ie aligning text to left and right in the same line].

Jonathan

The problem with pair programming is that some people either get computer programming or they dont. If you look at the grades for 1e3 last year, there were a huge number of people that got in the 90s, and a huge number of people either failing, or barely passing. Very little people in between.

If you get two competent programmers in a pair, you will spot the errors, but also think of other ways of doing the problem.

Brian

The guy that I'm paired with seems fairly competent, but I really would prefer to work on my own for something like programming..

Boston

Having to write code which someone else needs to understand is good for you. Stops you being sloppy.

Brian

Perhaps, but it gets frustrating when neither of us know what we're doing and both our ideas on how to do it are different.

Jonathan

Is the main problem the algorithms or the syntax?

Boston

Baza210 wrote: »

Perhaps, but it gets frustrating when neither of us know what we're doing and both our ideas on how to do it are different.

True. On small projects its hard to divide up who does what. Theres a book called thinking in C++. Its free online. Its the best C++ learning resource out there.

TheAmateur

Agh I just realised I never picked up a lab report sheet at the last lab, can I just go in on monday and get one then?

Brian

http://mindview.theunixman.com/ThinkingInC-Beta3.zip

mathew

Baza210 wrote: »

That wasn't the assignment, it was just me messing around. The hard part of the assignment is
1. Working with someone else
2. Doing stuff we haven't covered [ie aligning text to left and right in the same line].

cout << fixed << setprecision(2) << endl;
    cout << setw(10) << left << name << endl;
    cout << setw(15) << left << "Hours \t" << setw(10) << hours << endl;
    cout << setw(10) << "Hourly rate \t" << setw(10) << setprecision(2) << wage << endl;
    cout << endl;
    cout << setw(10) << "Gross Wage \t" << setw(10) << grosswage << endl;
    cout << setw(10) << "Tax @ 20% \t" << setw(10) << setprecision(2) << tax << endl;
    cout << setw(10) << "PRSI @ 2.5% \t" << setw(10) << setprecision(2) << prsi << endl;
    cout << setw(10) << "Union Dues \t" << setw(10) << uniondues << endl;
    cout << "-------------------------------" << endl;
    cout << setw(10) << "Net Pay \t" << setw(10) << netpay << endl << endl;

\t is tab, which tabs to the end of the set width.. very handy operator.. similar to \n which you have probably been told about

Brian

Would it be possible to have something like this

cout << "Enter your desired printout width: ";
cin >> width;
cout << setw(width) << left << name << endl;

?

Peleus

Baza210 wrote: »

Would it be possible to have something like this

cout << "Enter your desired printout width: ";
cin >> width;
cout << setw(width) << left << name << endl;

?

but why would the operator care what width the printout was?

mathew wrote: »

cout << fixed << setprecision(2) << endl;

    cout << setw(10) << left << name << endl;

    cout << setw(15) << left << "Hours \t" << setw(10) << hours << endl;

    cout << setw(10) << "Hourly rate \t" << setw(10) << setprecision(2) << wage << endl;

    cout << endl;

    cout << setw(10) << "Gross Wage \t" << setw(10) << grosswage << endl;

    cout << setw(10) << "Tax @ 20&#37; \t" << setw(10) << setprecision(2) << tax << endl;

    cout << setw(10) << "PRSI @ 2.5% \t" << setw(10) << setprecision(2) << prsi << endl;

    cout << setw(10) << "Union Dues \t" << setw(10) << uniondues << endl;

    cout << "-------------------------------" << endl;

    cout << setw(10) << "Net Pay \t" << setw(10) << netpay << endl << endl;

Thank you for this. I was incredibly hungover this morning when she was explaining the setw(n) thing. and basically everything else you have there.

Brian

Peleus wrote: »

but why would the operator care what width the printout was?

Perhaps the operator wouldn't care about what width it was, but if my above coding works then it could be applied for variable decimal places. ie

cout << "To how many decimal places would you like the value calculated to be?";
cin >> places;
cout << "Result: " << setprecision(places) << a*b;

mathew

yep that can be done Baza. Pretty much anywhere a value needs to be entered, it can be done so with a variable..

Jonathan

You can have great fun with variable variables :P

Boston

Baza, the only way to learn to program is to do it and see what happens. Get a copy of Dev-C++ if you're stuff in windows.

Brian

I hope to not be stuck in XP for too much longer. I'll be switching to Intrepid or OS X fairly soonish.

Jonathan

Boston wrote: »

Get a copy of Dev-C++ if you're stuff in windows.

If you just wanna be compiling your programs to see if they work, you might be better off with cygwin + make.

TheAmateur

Boston wrote: »

Baza, the only way to learn to program is to do it and see what happens. Get a copy of Dev-C++ if you're stuff in windows.

Just downloaded this, it seems pretty cool but when I tried to access the list of functions in the help file it said the list wasn't there... and that's probably the most useful part of the help file! Also, when I hit run on a test file cmd just flashed up and then disappeared (I'm on windows xp), anyone know why that'd be?

Boston

jmccrohan wrote: »

If you just wanna be compiling your programs to see if they work, you might be better off with cygwin + make.

Thats a better option, but Dev-C++ is the quickest and simplest windows base IDE I've come accross for C++.

TheAmateur wrote: »

Just downloaded this, it seems pretty cool but when I tried to access the list of functions in the help file it said the list wasn't there... and that's probably the most useful part of the help file! Also, when I hit run on a test file cmd just flashed up and then disappeared (I'm on windows xp), anyone know why that'd be?

What list of function? I haven't used Dev-C++ in years.

Brian

Cygwin wouldn't download for me because it installs itself from the internet and couldn't figure out the Trinity proxy

Boston

Theres an option for inputting the proxy. Did you remember to change the port?

Brian

I know, I input the proxy [proxyB.tcd.ie] and chose port 8080, but there was no entry box for authentication, and no prompt after clicking connect/download.

Jonathan

Dev c++ complied programs will flash up cause windows does not leave the program on screen once terminated. Your options are run the program from the command line, or you can add system.pause(0) (i cant remember the exact line off the top of my head.) this will cause the program to stay on screen when it halts.

Boston

or have a cin command at the bottom of your program so it waits for input before it exits.

Jonathan

Baza210 wrote: »

I know, I input the proxy [proxyB.tcd.ie] and chose port 8080, but there was no entry box for authentication, and no prompt after clicking connect/download.

There should be, i used TCDwifi to download repositories on friday..

mathew

Boston wrote: »

or have a cin command at the bottom of your program so it waits for input before it exits.

or get a mac..... :P

Brian

@mathew- please see r/b/m ; )

Jonathan

Baza210 wrote: »

@mathew- please see r/b/m ; )

Whats r/b/m?

Brian

It's this thread desu!

Jonathan

Baza210 wrote: »

It's this thread desu!

You and your acronyms..

Brian

Go and hang out with everyone else in the engineering thread. There's loads of people there.

Jonathan

I propose an official engineering thread.

Brian

Officiality always seemed pointless to me. This isn't ONM or some crap where people are competing to have Teh OFFICIAL SMASH BROS BRAWL THREAD!

PurpleFistMixer

What exactly does programming have to do with errors in Engineering Physics Labs? Are engineers required to write programs to calculate their errors?

mathew

PurpleFistMixer wrote: »

What exactly does programming have to do with errors in Engineering Physics Labs? Are engineers required to write programs to calculate their errors?

No just some serious off topic posting going on...