The Student Room Group

Integration by substitution (help)

I understand that e^(x²) does not have an elementary derivative and the calculations to get that as a definite integral with limits infinity and negative infinity, the value obtained is √π. In my latest attempt at this, I replaced e^(x²) with e^(u²) from the original equation (since substitution was requested in the question) so that I wouldn't get confused and after getting an equation I don't think I can work with, I'm not sure if I'm approaching this question correctly at all.

Can someone help to point me in the right direction? Thanks

IMG_9566.jpg
(edited 3 weeks ago)
Reply 1
I understand that e^(x²) does not have an elementary derivative and the calculations to get that as a definite integral with limits infinity and negative infinity, the value obtained is √π. In my latest attempt at this, I replaced e^(x²) with e^(u²) from the original equation (since substitution was requested in the question) so that I wouldn't get confused and after getting an equation I don't think I can work with, I'm not sure if I'm approaching this question correctly at all.
Can someone help to point me in the right direction? Thanks
IMG_9566.jpg

u = (x-mu)/sqrt(2)sigma
should be straightforward
Reply 2
Original post by mqb2766
u = (x-mu)/sqrt(2)sigma
should be straightforward


So I tried this earlier and found dx in terms of du:
u = x/(√2σ) - μ/(√2σ)
du/dx = 1/(√2σ)
dx = (√2σ)du

And substituting this gave me: ∫(e^(u²))(√2σ)du, and I’m not sure where to go from there
Reply 3
Original post by ssccqq
So I tried this earlier and found dx in terms of du:
u = x/(√2σ) - μ/(√2σ)
du/dx = 1/(√2σ)
dx = (√2σ)du
And substituting this gave me: ∫(e^(u²))(√2σ)du, and I’m not sure where to go from there

Its e^(-u^2) and your definite integral limits should be similar and sqrt(2) and sigma are constants so its pretty much a write down answer
(edited 3 weeks ago)
Reply 4
Original post by mqb2766
Its e^(-u^2) and your definite integral limits should be similar and sqrt(2) and sigma are constants so its pretty much a write down answer


Sorry, you’re right, it’s -u²! I think I might be overcomplicating this, so I’m also really sorry for sounding stupid right now, but am I just supposed to compare with the original equation (∫(e^(-u²)du=√π) and add in the coefficient to make it fit?
Reply 5
Original post by ssccqq
Sorry, you’re right, it’s -u²! I think I might be overcomplicating this, so I’m also really sorry for sounding stupid right now, but am I just supposed to compare with the original equation (∫(e^(-u²)du=√π) and add in the coefficient to make it fit?

Sure youve a constant multiplier so take it outside the integral and im presuming youve worked out that the limits dont chage
(edited 3 weeks ago)
Reply 6
Original post by mqb2766
Sure youve a constant multiplier so take it outside the integral and im presuming youve worked out that the limits dont chage


Ah, thank you so much for your help! I’m so sorry for bugging you with this; I thought I had to find values for σ and μ (even though there’s only one equation, so pretty sure that’d be impossible) or something more complicated. Yeah, really straightforward, haha
Reply 7
Original post by ssccqq
Ah, thank you so much for your help! I’m so sorry for bugging you with this; I thought I had to find values for σ and μ (even though there’s only one equation, so pretty sure that’d be impossible) or something more complicated. Yeah, really straightforward, haha

Not sure of the context of the question, but Im sure you recognise that youre basically working out the normalising factor for a normal distribution. The mean mu doesnt affect things as it just shifts the location of the bump and as youre integrating between -inf and inf, thats irrelevant. However the scaling term 1/sqrt(2)sigma must affect the width and so the value of the integral.
Reply 8
Original post by mqb2766
Not sure of the context of the question, but Im sure you recognise that youre basically working out the normalising factor for a normal distribution. The mean mu doesnt affect things as it just shifts the location of the bump and as youre integrating between -inf and inf, thats irrelevant. However the scaling term 1/sqrt(2)sigma must affect the width and so the value of the integral.


I did find that picking variables σ and μ reminded me of normal distribution, but I couldn’t remember the specific equation. Interesting to know! This is just a standalone question for pre-uni revision, but I’ll definitely add normal distribution to the long list of what to read up on before uni

Quick Reply