0
$\begingroup$

This problem is related to a time constant (herein called $x$) possessed by an article heated in a constant-temperature oven. Below I have simplified the presentation of the problem so as to concentrate on the math aspects. I do not want to get off onto discussing times and temperatures.

Is there a direct symbolic solution for $X$ in the following equation? i.e, $X=$ _________
(Even a converging series would be nice.)

$J\left(\exp\left(\frac ax\right) -1\right) = K\left(\exp\left(\frac bx\right) -1\right)$

Assume (for laboratory use):

$J, K, a, b$ are real;

$a > b > 0$;

$|K| > |J| $

In any given experiment, $K$ and $J$ have the same polarity (either both are positive, or both are negative)

One set of numerical values:

$J=18; K=52; a=3.8; b=2.25; x= 1.824551$

I converted the equation to an infinite series, but all of the $x$'s were in the denominator:

$$... + (Kb^n-Ja^n)/(x^nn!) + \ldots = 0 \;\;\;(n=1, 2, 3, \ldots)$$

I am not a mathematician. I came across this problem while working in a laboratory in 1983, and have worked on it off and on since then.

$\endgroup$
0
$\begingroup$

In the most general case, an equation such as $$f(x)=J \left(e^{a/x}-1\right)-K \left(e^{b/x}-1\right)$$ does not have any analytic solution (the exception would be if $a$ was a small multiple of $a$ $2,3,4$ because in this case, a simple change of variable transforms the equation to a polynomial).

If the solution $x$ was large, we could replace $x$ by $\frac{1}{y}$ and expand (just as you did) as a Taylor series built at $y=0$. This would lead to an infinite expansion like $$f(y)=y (a J-b K)+\frac{1}{2} y^2 \left(a^2 J-b^2 K\right)+\frac{1}{6} y^3 \left(a^3 J-b^3 K\right)+O\left(y^4\right)$$ Limited to second order, the approximate solution would then be $$y=\frac{1}{x}=-\frac{2 (a J-b K)}{a^2 J-b^2 K}$$ Using your numbers, we should then find that $x=-\frac{37}{1080}\simeq -0.0342593$ which is just awful compared to what you give. Expansion in series could be only useful for estimation purposes under a lot of conditions.

All of that being said, only numerical methods should be used to solve your equation. A simple root finding method is Newton which, starting from a "reasonable" guess $x_0$ will update it according to $$x_{n+1}=x_n-\frac {f(x_n)} {f'(x_n)}$$ For the case you gave, let us use $x_0=2$. The iterates will then be $1.75961$, $1.81760$, $1.82447$, $1.82455$ which is the solution for six significant figures.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.