The Student Room Group

"Impossible" logarithm question?

We were given the following simultaneous equations to solve (equations [1] and [2] respectively):

log(x+y)=0[br]2logx=log(y1)\log({x+y})=0[br]2\log{x}=\log({y-1})

(only dealing with real numbers)

This is my thought process:

I started with equation [1]. We can rearrange it to

B0=x+yB^0=x+y

where BB is the base of the logarithm. But any base to the power of 0 equals 1.

Therefore,

x+y=1y1=xx+y=1\\y-1=-x

Substituting this into equation [2] gives

2logx=log(x)2\log{x}=\log({-x})

However, assuming xx is real, either one of xx or x-x is negative, or x=x=0x=-x=0. In both cases, at least one of the logarithms is undefined.

--------

Is there something I'm missing here?
Thanks!
(edited 9 years ago)
Reply 1
I agree with your reasoning. There are no solutions.
Debatable- some would argue x = -1, y = 2 is a solution, but that requires a little fiddling with the second equation for it to be defined.

The question you have to ask is are 2log(x) and log(x^2) equivalent for -ve x.
Reply 3
Original post by Desk-Lamp
Debatable- some would argue x = -1, y = 2 is a solution, but that requires a little fiddling with the second equation for it to be defined.

The question you have to ask is are 2log(x) and log(x^2) equivalent for -ve x.


log(x) is undefined for x<=0, so x=-1 cannot possibly be a solution.
Original post by james22
log(x) is undefined for x<=0, so x=-1 cannot possibly be a solution.


log(x) has no real value for x<= 0, but that doesn't mean it's undefined. In this case the solutions have to be real numbers but that doesn't prevent you working through the complex plane to get a result.
Reply 5
Original post by Desk-Lamp
log(x) has no real value for x<= 0, but that doesn't mean it's undefined. In this case the solutions have to be real numbers but that doesn't prevent you working through the complex plane to get a result.


If you are going to get into complex numbers, then you need to define what definition of log you are using. The typical definition does not define it for real x less than or equal to 0, and it is in fact never defined at 0 whatever you do.
the standard definition of the logarithm applies exactly the same for numbers < 0, there are simply no real solutions to log(-x).
It's fairly simple to show this without ever needing to consider something undefined- Start with the Euler Identity

1=eπi -1 = e^{\pi i}

x=xeπi -x = xe^{\pi i}

log(x)=log(xeπi) log(-x) = log(xe^{\pi i})

2log(x)=2log(xeπi) 2log(-x) = 2log(xe^{\pi i})

2log(x)=log(x2e2πi) 2log(-x) = log(x^2 e^{2 \pi i})

e2πi=1e^{2 \pi i} = 1, therefore 2log(x)=log(x2)2log(-x) = log(x^2)
Reply 7
Original post by Desk-Lamp
the standard definition of the logarithm applies exactly the same for numbers < 0, there are simply no real solutions to log(-x).
It's fairly simple to show this without ever needing to consider something undefined- Start with the Euler Identity

1=eπi -1 = e^{\pi i}

x=xeπi -x = xe^{\pi i}

log(x)=log(xeπi) log(-x) = log(xe^{\pi i})

2log(x)=2log(xeπi) 2log(-x) = 2log(xe^{\pi i})

2log(x)=log(x2e2πi) 2log(-x) = log(x^2 e^{2 \pi i})

e2πi=1e^{2 \pi i} = 1, therefore 2log(x)=log(x2)2log(-x) = log(x^2)


You have to define what you mean by log before you use it. We can do this easily in the real case, in the complex case there are infinite valid definitions and the usual one rules out negative reals.
Original post by james22
You have to define what you mean by log before you use it. We can do this easily in the real case, in the complex case there are infinite valid definitions and the usual one rules out negative reals.


I'm working with the same definition everbody uses- log(x) is the number you have to raise the base to to get x. As far as I can see, this poses no restriction to negative x.

Indeed, it seems a strange reason to cut out valid solutions midway through evaluating an expression simply because you have decided you don't want to.
Reply 9
Original post by Desk-Lamp
I'm working with the same definition everbody uses- log(x) is the number you have to raise the base to to get x. As far as I can see, this poses no restriction to negative x.

Indeed, it seems a strange reason to cut out valid solutions midway through evaluating an expression simply because you have decided you don't want to.


That is not the definition of logs for the complex plane. It only works on the positive really line because exponents are 1-1. In the complex plane such a definition makes no sense because there are infinite different numbers that would work. For example what is log(1) (base e)? Is it 0, or 2*pi*i, or 4*pi*i?
Original post by james22
That is not the definition of logs for the complex plane. It only works on the positive really line because exponents are 1-1. In the complex plane such a definition makes no sense because there are infinite different numbers that would work. For example what is log(1) (base e)? Is it 0, or 2*pi*i, or 4*pi*i?


These are all valid solutions to the equation x = ln(1). We're not trying to define a one-one function here, this is just solving an equation.

Just because the square root function only gives out positive values, it doesn't mean -1 isn't a solution to the equation x^2 = 1
Original post by Desk-Lamp
These are all valid solutions to the equation x = ln(1). We're not trying to define a one-one function here, this is just solving an equation.

Just because the square root function only gives out positive values, it doesn't mean -1 isn't a solution to the equation x^2 = 1


I was showing why the definition of log you used did not work (it would lead to a multi-valued function which is not allowed here). You still haven't said what you mean by log, and until you define it you cannot use it.
Original post by james22
I was showing why the definition of log you used did not work (it would lead to a multi-valued function which is not allowed here). You still haven't said what you mean by log, and until you define it you cannot use it.


Point not taken there. Just exactly why is are "multi-valued functons not allowed here"?

This seems an obvious parallel with the square root function- just because we neglect -ve values when we want a one-one function, it doesn't mean we have to impose restrictions when solving equations.
Original post by Desk-Lamp
Point not taken there. Just exactly why is are "multi-valued functons not allowed here"?

This seems an obvious parallel with the square root function- just because we neglect -ve values when we want a one-one function, it doesn't mean we have to impose restrictions when solving equations.


We aren't imposing restrictions, we just need a definition of the log function to work with.

If you want to start using multifunctions then go ahead, but the question would say very specifically if we were supposed to be using log as a multifunction since that is not a typical use of it. If you allow a multivalued function then you need to accept that log(x) is not a number, but really a set (various ways of handling multifunctions), so it becomes a very non-standard equation. Log almost always means the usual definition, and when used on teh complex plane it is defined specifically (unless you have been told to use a certain one unless stated otherwise).
I think this has become somewhat beyond the point. All the question is asking is to find values of x and y for which these relationships are true. If you plug in these values, you can evaluate both suides to the same real number, hence those x and y values are solutions.

Anything beyond this is not necessary for the purposes of this question.
Original post by Desk-Lamp
I think this has become somewhat beyond the point. All the question is asking is to find values of x and y for which these relationships are true. If you plug in these values, you can evaluate both suides to the same real number, hence those x and y values are solutions.

Anything beyond this is not necessary for the purposes of this question.


What you have just said is false though, you cannot use logs on non-positive real numbers without defining what log you are using. if you are using the standard logarithm then it is actually impossible to find the logs of negative reals. You cannot possibly evaluate both sides to be the same since you cannot evaluate logs of negative reals.

Quick Reply

Latest