Hi everyone! Can anyone discuss with me partial fraction decomp? I know how to do it but I don't understand why it's okay to set up the problems in certain ways...Here is what I mean>>> for example s+1/s^2(s+2)^3 is set up like this>>>A/s +B/s^2 + C/(s+2) + D/ (s+2)^2 + E/(s+2)^3 the reason I'm confused is because it seems like it should be set-up like this instead>>> A/s +B/s + C/(s+2) + D/ (s+2) + E/(s+2) Can anyone explain this to me? Thanks soooo much! :o)
Hi again amistre! aren't you tired of me yet? :o)
you already have a root of s defined A/s + B/s = (A+B)/s which is just redundant
almost, bladders getting full :)
when you say root, I am thinking of where the function crosses the x-axis...what does a root of "s" actually mean?
notice this: A/s + B/s^2 when we add fractions, which is what decomp is doing, we are trying to convert it into all the possible setups A(s)/s(s) + B/s^2 = (As+B)/s^2
the roots are the factors of the setup
okay yes...roots are factors! totally get that
gimme just a sec to work that example on paper to prove to myself...one sec
\[\frac{5}{2*3^2*5^3}\] notice the prime factorization of the denominator \[\frac{5}{2*3^2*5^3}=\frac{A}{2}+\frac{B_0}{3}+\frac{B_1}{3^2}+\frac{C_0}{5}+\frac{C_1}{5^2}+\frac{C_2}{5^3}\] these are all the possible formations that we can compost it into
lol, compost
so you are saying if you get a common denominator on the right side, and add, it will equal the left side then?
the decomposition is accounting for ALL possible combinations of the summation. its up to us to determine which ones are actually valid.
spose we ignore some of them \[\frac{5}{2*3^2*5^3}=\frac{A}{2}+\frac{B}{3^2}+\frac{C}{5^3}\] we havent accounted for all possible summations, have we?
well according to what you said, no...but my intuition wants to say, "I need to prove that to myself"
we dont get a truely decompsed set up if we do not account for ALL possible combinations.
okay...i get now that we have to "completely decompose" and write out all combination on the right...but can you answer this yes or no please...if we get a common denominator on the right side and add all those terms, will it equal the left side?
yes, that is after all how you add up fractions
okay...just making sure! :o)
because I tried to add up your example >A(s)/s(s) + B/s^2 = (As+B)/s^2 and I wasn't getting anywhere
you still have to solve for A and B
if both sides are to be equal, then they have to be equal for all values of s
so humor me for a sec...take your example>>>A(s)/s(s) + B/s^2 = (As+B)/s^2 and prove it for me
I can't get it
hi zepdrix...I see you spying on us!
wrote it all out but I can't prove itI
wait! got it!!!! :o) YAY!
okay amistre!, not that I ever doubted you, but I proved it to myself now! You have magical powers and I believe you 100%!
so bottom line, write out every facotr on the right to completely "compost" haha
(s+1)/s^2(s+2)^3 = A/s +B/s^2 + C/(s+2) + D/ (s+2)^2 + E/(s+2)^3 let s = 0, and cover up the bad parts (0+1)/0^2(0+2)^3 = A/0 +B/0^2 + C/(0+2) + D/ (0+2)^2 + E/(0+2)^3 1/8 = C/2 + D/4 + E/8 1 = 4C + 2D + E ------------------------------- let s = -2, and cover up the bad parts (-2+1)/(-2)^2(-2+2)^3 = A/-2 +B/-2^2 + C/(-2+2) + D/ (-2+2)^2 + E/(-2+2)^3 -1/4 = -A/2 + B/4 -1 = -2A + B ------------------------------- now we have some conditions defined that must hold let B = 2A - 1 let E = 1 - 4C - 2D and plug them in the setup, let s=-1 and see what the results are eventually you get to a point where you can start defining all the tops
yay!!
question...I completely get what you did to solve for B and E...but what made you choose s=-1? I would have never done that
its something small, and it makes the top left go to 0, remember, in order for left and right to be equal, they must be true for ALL values of s, so when we narrow it down out of the simpler stuff we start branching out (s+1)/s^2(s+2)^3 = A/s +(2A-1)/s^2 + C/(s+2) + D/ (s+2)^2 + (1 - 4C - 2D )/(s+2)^3 0= -A +(2A-1) + C + D + (1 - 4C - 2D ) 0= A -1 + C + D + 1 - 4C - 2D 0= A - 3C - D A = 3C + D must be true
notice that A in terms of C and D reduce us to 2 unknowns
makes the top left of "what" go to zero?
let s=1, and s=2 to get 2 equations in 2 unknowns
you really need to pay attention to what you post lol (s+1)/s^2(s+2)^3 is our left side ^^^^ the top of it -1+1 = 0
A = 3C + D B = 2A - 1 = 6C + 2D - 1 E = 1 - 4C - 2D and we have only C and D left to solve ...
the simpler stuff is used up, so we simply have to define 2 equations for a particular s value that hasnt been used and work the results
i was paying attention to what I posted...I just didn't understand, so I asked...not that you clarified, I get it...thanks
:)
I hate these systems of equations...can't I just distribute and make a matrix?
maybe, dunno how that would go tho :) (s+1)/s^2(s+2)^3 = (3C+D)/s +(6C + 2D - 1)/s^2 + C/(s+2) + D/ (s+2)^2 + (1 - 4C - 2D )/(s+2)^3 let s=1 2/27 = (3C+D) +(6C + 2D - 1) + C/3 + D/9 + (1 - 4C - 2D )/27 2 = 3(27)C+ 27D + 6(27)C + 2(27)D - 27 + 9C + 3D + 1 - 4C - 2D 28 = [3(27)+6(27) +5]C+ [27 + 2(27) + 1]D 28 = 248 C+ 82 D ----------------------------------- let s=2 3/16(6)^3 = (3C+D)/2 +(6C + 2D - 1)/4 + C/6 + D/ 6^2 + (1 - 4C - 2D )/6^3 3/16 = 108(3C+D) +54(6C + 2D - 1) + 36C + 6D + (1 - 4C - 2D ) 3/16 = 108(3)C +108D +54(6)C + 54(2)D - 54 + 36C + 6D + 1 - 4C - 2D oy vey this is long
amistre, i really do admire your work ethic but I honestly would rather gargle peanut butter! wouldn't a matrix be much easier?
(53(16) +3)/16 = [108(3) +54(6)+32] C + [108+54(2)+4] D im not sure i have the mental capacity to figure out a matrix solution lol
okay...well I will do the matirx and then compare your answer...I will need like 5 minutes...brb
28 = 248 C+ 82 D, (53(16) +3)/16 = (108(3) +54(6)+32) C + (108+54(2)+4) D http://www.wolframalpha.com/input/?i=28+%3D+248+C%2B+82+D%2C+%2853%2816%29+%2B3%29%2F16+%3D+%28108%283%29+%2B54%286%29%2B32%29+C+%2B+%28108%2B54%282%29%2B4%29+D assuming i made absolutely no errors ..... C = -14389/9600, D = 11699/2400
yikes...gimme a sec
http://www.wolframalpha.com/input/?i=decompose+%28s%2B1%29%2F%28s%5E2%28s%2B2%29%5E3+%29 im sure i made some faux pauxes
apparently D = 0 so yeah, oh well
yeah, i made some errors along the way, but the method is sound regardless of my ineptitudity
this is really good practice for my test! :o) getting there!
1 0 1 0 0 0 6 1 4 1 0 0 12 6 4 2 1 0 8 12 0 0 0 1 0 8 0 0 0 1 equals>>>
A=-.0625 = -1/16 B=.125 = 1/8 C=.0625 = 1/16 D=0 = 0 E=-.25 = -1/4
I can't get wolfram to answer the question...but I think it's correct
(s+1)/s^2(s+2)^3 = A/s +B/s^2 + C/(s+2) + D/ (s+2)^2 + E/(s+2)^3 (s+1)/(s+2)^3 = As +B + Cs^2/(s+2) + Ds^2/ (s+2)^2 + Es^2/(s+2)^3 (s+1) = As(s+2)^3 +B(s+2)^3 + Cs^2(s+2)^2 + Ds^2(s+2) + Es^2 let s=-2,-1,0,1,2 -1 = A0 +B0 + C0 + D0 -4E ; E = 1/4 -1/4 = -A +B + C + D 1 = 8B ; B = 1/8 -3/8 = -A + C + D 3 = 8A -8C -8D ------------------------- 2 = A(3)^3 +(3)^3/8 + C(3)^2 + D(3) + 1/4 -6.5 = 108A + 36C + 12D --------------------------- -14 = 128A + 64C + 16D rref{{8,-8,-8,3},{108,36,12,-6.5},{128,64,16,-14}} bah!! i still messed it up long the way ...
http://www.wolframalpha.com/input/?i=decompose+%28s%2B1%29%2F%28s%5E2%28s%2B2%29%5E3+%29 this is the wolfs reply
@amistre64 Got it! YAY! My matrix worked! Take a peak...it says it's true! http://www.wolframalpha.com/input/?i= [%28s%2B1%29%2F%28s^2%28s%2B2%29^3%29]%3D[%28-1%2F16%29%2Fs+%2B+%281%2F8%29%2Fs^2+%2B+%281%2F16%29%2F%28s%2B2%29+%2B+%280%29%2F%28s%2B2%29^2+%2B+%28-1%2F4%29%2F%28s%2B2%29^3]
I can't get the dumb link to post correctly
your not on chrome ... are you
firefox
just paste this into wolfram [(s+1)/(s^2(s+2)^3)]=[(-1/16)/s + (1/8)/s^2 + (1/16)/(s+2) + (0)/(s+2)^2 + (-1/4)/(s+2)^3]
btw, your link didn't show anything
firefox is going bad i spose ....
so that's good news...I did it correctly, and I never took linear algebra yet either! :o)
bladders at its limit, so im heading towards bed. gnite :)
a couple questions... 1) this problem was like A/s + B/ s^2 etc I have seen problem like Ax+B/(something) etc... when do you know to combine letters instead of keeping the separate?
but yeah, to answer the question, we need to account for ALL the possible combinations of the linear roots.
well, the one simplifies to the other A/s + B/s^2 adds up to (As+B)/s^2 so its really a matter of how you want to go about it
tis is why i said the top needs to be one degree less then the bottom
so when I have seen problems set up like Ax+B/(something) etc... in the past, they could have just separated it further? They just took a shortcut or something?
they shortcutted it yes
okay neat! I forgot my second question and it was easy but important darn it! :O(
i remember!
now when we have a prime factor such as (s^3+3) we can split it into imaginary roots, and the same thing happens in the sum, we get (As^2 + Bs + C)/(s^3+3)
pretty sure thats not a prime factor but hey, its late lol
when solving these crazy problem with matrices and your calculator... the calculator gives you answers in decimals, but what if you get an answer like .66666667...the calc won't give you a fraction for that...any suggestions?
s^2 + 3 has imaginary roots, thats a better example
repeating decimals can be "summed" to a fraction S = .6666666.... 10S = 6.6666666...... 10S - S = 9S = 6 S = 6/9 = 2/3
so this way you can always use your calc to solve your systems of equations, even if the calculator won't give you a fraction right?
S = .5134823823823... 1000 S = 5134.823823823... 100000S = 513482.3823823... (100000 - 1000) S = 513482 - 5134 S can then be solved as a fraction
ugh. forgot i did a triple repeat ... fill in the detials eh
so you start with S = .5134823823823... do you have to just make up 2 new values like 1000 S = 5134.823823823... 100000S = 513482.3823823... ? I mean, can you sum the steps in words?
the idea is to get it so that you can subtract the decimals parts and be left with whole numbers
okay, I will look into how to do it
S = .5134823823823... 1000S brings the nonrepeating stuff out front, moves the deci 4 places 1000S = 5134. 823823823... we have triples repeating in this case so another 3 zeros shoould pull out enough stuff 1000000S = 5134823. 823823823... now the subtraction eliminates the decimal expansions 1000000S = 5134823. 823823823... - 1000S = - 5134. 823823823... --------------------------------- a S = b.0000000000 S = b/a
.777777... = 7/9 .23232323... = 23/99 .512512512512.. = 512/999
so the amount of zeros you choose is equal to how many repeating places there are?
depends on the number, if they are repeating from the start, then yeah otherwise we need to adjust it so that we are subtracting out the decimal expansion
.n repeat ..... 10^n gets the first n parts out of the way, and then for the repetaing part we multiply out another 10^k for some k repeating terms
so if you have .777777... you set it equal to like X so x=.777777... then 100x = 77.7777... and 1000x=777.777... (1000-100)x=(777.777...-77.777...) x=700/900 x=7/9
.666666666 10^0 pulls out the first 0 non repeating terms 10^0 * 10^1 pulls out the 1 repeating term (10^1 - 10^0)S = 6.6666..... - 0.6666..... 9S = 6
thats fine too
.066666... x=.066666... (1000-100)x = 66.666...-6.666... x=600/900 x=2/3
just as long as you have more zeros than like non repeating spaces right?
not quite 1000(.06666) = 66.6 100(.06666) = 6.6 900 x = 60 x = 60/900 = 6/90 - 2/30
=1/15
oh no! I got it wrong?
Join our real-time social learning platform and learn together with your friends!