# Logarithmic differentiation

Show 40 post(s) from this thread on one page
Page 1 of 2 12 Last
• January 2nd 2010, 09:32 PM
db5vry
Logarithmic differentiation
Given that $f(x) = x^{-\sqrt x}$ for $x > 0$,

(a) obtain an expression for f′ (x), [4]
(b) find the x-coordinate of the stationary point on the graph of f and determine whether this point is a maximum or a minimum. [4]

Now for the first part, I obtained a derivative of

$(x^{-\sqrt x}) \frac{-1}{\sqrt x} - \frac{lnx}{2\sqrt x}$ through logarithmic differentiation. I think this is correct, but the problem I have is knowing what to do in the second part of the question. How would you go about finding the x-coordinate and how can you tell what kind of point it is?

Thanks again for your time and patience :)
• January 2nd 2010, 09:34 PM
Drexel28
Quote:

Originally Posted by db5vry
Given that $f(x) = x^{-\sqrt x}$ for $x > 0$,

(a) obtain an expression for f′ (x), [4]
(b) find the x-coordinate of the stationary point on the graph of f and determine whether this point is a maximum or a minimum. [4]

Now for the first part, I obtained a derivative of

$(x^{-\sqrt x}) \frac{-1}{\sqrt x} - \frac{lnx}{2\sqrt x}$ through logarithmic differentiation. I think this is correct, but the problem I have is knowing what to do in the second part of the question. How would you go about finding the x-coordinate and how can you tell what kind of point it is?

Thanks again for your time and patience :)

Solve for zero. Use the second derivative test.
• January 2nd 2010, 09:42 PM
db5vry
Quote:

Originally Posted by Drexel28
Solve for zero. Use the second derivative test.

So would I take $\frac{d^2y}{dx^2}$ of the derivative I obtained?

It seems a very complicated function and I would find that enormously difficult to even attempt! Would you describe it as a lengthier process than the first, because both parts of the question are worth 4 marks each and the second part sounds like a higher volume of work (Shake)
• January 2nd 2010, 09:45 PM
Drexel28
Quote:

Originally Posted by db5vry
So would I take $\frac{d^2y}{dx^2}$ of the derivative I obtained?

It seems a very complicated function and I would find that enormously difficult to even attempt! Would you describe it as a lengthier process than the first, because both parts of the question are worth 4 marks each and the second part sounds like a higher volume of work (Shake)

Plug in local values. Be careful of this method though. Or, alternatively "cheat" and just graph it. Once again though, I would advise using the second derivative test or the slightly more innocuous first derivative test.
• January 2nd 2010, 09:54 PM
db5vry
Quote:

Originally Posted by Drexel28
Plug in local values. Be careful of this method though. Or, alternatively "cheat" and just graph it. Once again though, I would advise using the second derivative test or the slightly more innocuous first derivative test.

I like your idea of plugging in a value :) I couldn't cheat as this would be for an exam which doesn't allow graphic calculators - on my course syllabus it says nothing about taking second derivatives (It is an A-Level course for further maths) so I assume that plugging in values must be what they are asking for (Nod)

I plugged in x=1 as I recognised that ln 1 = 0 so thought that would automatically make the rest of it zero - but then I checked it and it did not mean this was the case, as it is only being subtracted from something else.

I'm struggling here but I hope I can learn how to answer it using your recommended method of second derivatives - I can only do this with basic/parametric equations but f'(x) in this question looks beyond my ability to differentiate a second time.
• January 2nd 2010, 10:24 PM
Drexel28
Quote:

Originally Posted by db5vry
I like your idea of plugging in a value :) I couldn't cheat as this would be for an exam which doesn't allow graphic calculators - on my course syllabus it says nothing about taking second derivatives (It is an A-Level course for further maths) so I assume that plugging in values must be what they are asking for (Nod)

I plugged in x=1 as I recognised that ln 1 = 0 so thought that would automatically make the rest of it zero - but then I checked it and it did not mean this was the case, as it is only being subtracted from something else.

I'm struggling here but I hope I can learn how to answer it using your recommended method of second derivatives - I can only do this with basic/parametric equations but f'(x) in this question looks beyond my ability to differentiate a second time.

Why is the second derivative so hard. Let $f(x)=x^{-\sqrt{x}}$. I made a slight clerical error. That being said, why don't you try the second derivative? You can validate it here wolframalpha.com. It will even tell you how to compute the derivative.
• January 2nd 2010, 10:39 PM
db5vry
Quote:

Originally Posted by Drexel28
Why is the second derivative so hard. Let $f(x)=x^{-\sqrt{x}}$. I made a slight clerical error. That being said, why don't you try the second derivative? You can validate it here wolframalpha.com. It will even tell you how to compute the derivative.

I actually used that site to validate the first derviative that I took :) and I had a look to see what the second one would be, but that is far beyond my capability which is why I find it difficult (Thinking) I did try, and took logs of both sides and I couldn't find a way to get any further. Sorry but I'm not sure the question is looking for the answer through those methods - maybe the plugging in the values that was suggested? How would you go about doing that? I have never done this before and I would like to know how :)
• January 2nd 2010, 10:43 PM
Drexel28
Quote:

Originally Posted by db5vry
I actually used that site to validate the first derviative that I took :) and I had a look to see what the second one would be, but that is far beyond my capability which is why I find it difficult (Thinking) I did try, and took logs of both sides and I couldn't find a way to get any further. Sorry but I'm not sure the question is looking for the answer through those methods - maybe the plugging in the values that was suggested? How would you go about doing that? I have never done this before and I would like to know how :)

This method is EXTREMELY sketchy. The concept is this. Having found a critical point $c$ of $f(x)$ we can assume (by the way the question is worded) that $f(c)$ is either a local maximum or local minimum. To determine which it is take $f(c)$ and compare it to $f\left(c+\varepsilon\right)$ for some arbitrarily small $\varepsilon$. The sketchiness is that if you use $\varepsilon=\tfrac{1}{2}$ was it small enough.
• January 2nd 2010, 10:55 PM
db5vry
Quote:

Originally Posted by Drexel28
This method is EXTREMELY sketchy. The concept is this. Having found a critical point $c$ of $f(x)$ we can assume (by the way the question is worded) that $f(c)$ is either a local maximum or local minimum. To determine which it is take $f(c)$ and compare it to $f\left(c+\varepsilon\right)$ for some arbitrarily small $\varepsilon$. The sketchiness is that if you use $\varepsilon=\tfrac{1}{2}$ was it small enough.

I see what you mean. I've tried this value and also some smaller values and none of them come close to equating to zero. The question seems to be looking for a very particular answer and it sounds definitely very hit and miss.

The question is on this past paper, just for you to get a bit of background:

http://www.wjec.co.uk/uploads/papers/w08-977-01.pdf it's the ninth and last question on the paper. I just have this feeling that I'm missing something completely obvious but I don't know what it is :(
• January 2nd 2010, 10:58 PM
Drexel28
Quote:

Originally Posted by db5vry
I see what you mean. I've tried this value and also some smaller values and none of them come close to equating to zero. The question seems to be looking for a very particular answer and it sounds definitely very hit and miss.

The question is on this past paper, just for you to get a bit of background:

http://www.wjec.co.uk/uploads/papers/w08-977-01.pdf it's the ninth and last question on the paper. I just have this feeling that I'm missing something completely obvious but I don't know what it is :(

You seem to have used the method I described to find a zero of $f'(x)$. Instead, it is used to determine, when found, whether that zero admits a maximum or minimum value.
• January 2nd 2010, 11:05 PM
db5vry
Quote:

Originally Posted by Drexel28
You seem to have used the method I described to find a zero of $f'(x)$. Instead, it is used to determine, when found, whether that zero admits a maximum or minimum value.

I now interpret what you mean as inserting a small value as x and seeing whether it is positive or negative - I put in 0.2 and got a negative value, which would mean that it is a maximum point?

Sorry if I have misunderstood again, it must be frustrating but I massively appreciate your help :)
• January 2nd 2010, 11:07 PM
Drexel28
Quote:

Originally Posted by db5vry
I now interpret what you mean as inserting a small value as x and seeing whether it is positive or negative - I put in 0.2 and got a negative value, which would mean that it is a maximum point?

Sorry if I have misunderstood again, it must be frustrating but I massively appreciate your help :)

$f(c)$ is by assumption either a local maximum or local minimum. So if $f(c) then we may conclude that it is a minimum, etc.
• January 2nd 2010, 11:15 PM
db5vry
Quote:

Originally Posted by Drexel28
$f(c)$ is by assumption either a local maximum or local minimum. So if $f(c) then we may conclude that it is a minimum, etc.

Well if I called f(c) f(0.2), I obtained a value of -0.89 to 2dp.
If I called f(E) f(0.02), I obtained 11.75.

The smaller value has seen the sign change. I didn't think that would have happened but I'm sure it's incorrect :(
• January 2nd 2010, 11:22 PM
Drexel28
Quote:

Originally Posted by db5vry
Well if I called f(c) f(0.2), I obtained a value of -0.89 to 2dp.
If I called f(E) f(0.02), I obtained 11.75.

The smaller value has seen the sign change. I didn't think that would have happened but I'm sure it's incorrect :(

Is .2 the zero of $f'(x)$?
• January 2nd 2010, 11:26 PM
db5vry
Quote:

Originally Posted by Drexel28
Is .2 the zero of $f'(x)$?

It isn't, I am not sure how to do that at all. I feel increasingly out of my depth with this question (Shake)
Show 40 post(s) from this thread on one page
Page 1 of 2 12 Last