Finding contstant in remainder term for linearization of $exp(x)$

108 Views Asked by At

I've managed to find the remainder term for the linearization of $\exp(x)$ about $x=0$ in Lagrange form: $$ R_1(x)=\exp(θ_Lx)\frac 12x^2 \text{ ,where } θ_L∈[0,1]. $$

My question is how would I find the value of $θ_L$ when $x$ is tending to $0$? I've tried to rearrange the above equation given that $R_1(x):=f(x)-P_k(x)$ to make $θ_L$ the subject and then find the limit of the equation when $x$ tends to $0$ but that didn't work. Any help would be appreciated.

1

There are 1 best solutions below

2
On

Not sure if I understand the question, so I ask the OP to read the following and clarify the question accordingly. (Update: see the end of the post for the likely answer).

Lagrange remainder has the following (exact) form:

$$R_n=\int_{x_0}^x f^{(n+1)}(t)\frac{(x-t)^n}{n!} dt$$

In this case $f(t)=e^t$ and $x_0=0$, so:

$$R_1=\int_{0}^x e^t(x-t) dt=e^x-1-x$$

Which is of course, trivially true, and doesn't help much.

However, we can use the Mean-Value theorem which states that if a function is differentiable and continuous on $[a,b]$, there's at least one number $c$ such that:

$$f'(c)=\frac{f(b)-f(a)}{b-a}$$

From this theorem the following conclusion follows for some function $g(x)$ which is integrable and doesn't change sign on $[a,b]$:

$$\int_a^b f(t)g(t)dt=f(c) \int_a^b g(t)dt$$

Here number $c \in [a,b]$ can be different than in the previous example (depending on $g(x)$). Getting back to $R_n$, we have:

$$R_n=f^{(n+1)}(c) \int_{x_0}^x \frac{(x-t)^n}{n!} dt=f^{(n+1)}(c) \frac{(x-x_0)^{n+1}}{(n+1)!}$$

Thus, we have an equation for our case:

$$R_1=e^x-1-x=e^c \frac{x^2}{2}$$

$$e^c=2 \frac{e^x-1-x}{x^2}$$

Obviously, $c=c(x)$.

Now the OP seems to want $\lim_{x \to 0} c(x)$.

$$\lim_{x \to 0} e^c=2 \lim_{x \to 0} \frac{e^x-1-x}{x^2}=1$$

$$\lim_{x \to 0} c=0$$

Just as it should be because for $x \to 0$, because the quadratic term close to $0$ should be $x^2/2$.


If for example, we wanted to find $c$ for $x=1$, then we would have:

$$e^c=2(e-2)$$

$$c=\log 2+\log(e-2)$$

And the remainder (again, trivially true):

$$R_1=2(e-2)\frac{1}{2}=e-2 $$


Edit

However, if the OP searches $c$ in another form:

$$c(x)=\theta(x) x$$

Then of course there's a way to find the limit for $\theta$ from the above:

$$\theta(x) x= \log \left(2 \frac{e^x-1-x}{x^2} \right)$$

$$\theta_L= \lim_{x \to 0} \frac{1}{x} \log \left(2 \frac{e^x-1-x}{x^2} \right) = \frac{1}{3}$$

The above follows from:

$$\lim_{x \to 0} \frac{1}{x} \log \left(2 \frac{e^x-1-x}{x^2} \right) =\lim_{x \to 0} \frac{1}{x} \log \left(2 \frac{x^2/2+x^3/6}{x^2} \right)=\lim_{x \to 0} \frac{1}{x} \log \left(1+\frac{x}{3} \right)=\frac{1}{3}$$