Equation self-substitution

122 Views Asked by At

Let $f(y)=g(x)$ be an equation.

I remember my grade school teacher warning students against self-substitution in solving equations, which is the procedure described as follows:

Assuming $f(.)$ to be invertible, we solve for $y = f^{-1}(g(x))$. Then we substitute this expression for $y$ in the original equation:
$f(y)=g(x)$
$f(f^{-1}(g(x)))=g(x)$
$g(x)=g(x)$
Which is a tautology, since the substitution has introduced no new constraint and is thus a practically useless procedure.

Now, assume the equation can be expressed as $F(y,y)=f(y)=g(x)$. For instance, $F(a,b)$ could be $a^3+b$ and $F(y,y)=y^3+y$.

Questions:
1. Is it valid to apply the substitution $y = f^{-1}(g(x))$ to only one of the two $y$’s? Specifically, is it valid to say $F(y, f^{-1}(g(x)))=g(x)$?
2. If I wish to avoid the substitution, can $F(y, f^{-1}(g(x)))=g(x)$ be achieved through other means - through manipulations of the original equation $F(y,y)=g(x)$, i.e. by performing elementary operations on both sides?

1

There are 1 best solutions below

2
On

I'm not really sure what your second question means; let me address the first.

The substitution your teacher warned you about is still valid - it won't lead to incorrect statements - it's just useless. So, you can do it, but there's no reason to.

As to the substitution you describe. First, let me describe an example. Suppose we're given the equation "$y+y=3x$." Then we have $F(a, b)=a+b$, $f(y)=y+y$ (so $f^{-1}(y)={y\over 2}$) and $g(x)=3x$. Then $y=f^{-1}(g(x))$, so this can be rewritten as $${3x\over 2}+y=3x.$$ I take this to be an example of what you're describing; if I'm wrong, please correct me. (Note that we can perform this style of substitution in many ways, e.g. to get ${3x\over 4}+{3y\over 2}=3x.$)

  • Is it valid? Sure, why wouldn't it be? Substituting an equal for an equal is always valid, that's what "equal" means.

  • Is it useful? Well, not really. In order to do this we need to already know $f^{-1}$, so the only possible use it could have is if $g$ were hard to invert, but $F$ and $g$ combined was easy to invert in the sense that the function $$G[f^{-1}(g(x)), g(x)]$$ were easy to compute - where $G$ is the "semi-inverse" of $F$: $G(a, b)=c$ if $F(a, c)=b$. Think of $G$ as taking two inputs, $a$ and $b$, and saying: "If $b$ were the output of an expression of the form $F(a, -)$, what would go in the blank?" The key point is that, because $$F(f^{-1}(g(x)), y)=g(x),$$ we have $$y=G[f^{-1}(g(x)), g(x)].$$ But I can't think of a case where this happens, so I'm going to say that this is probably not useful (except maybe in very contrived circumstances). I'd be happy to be corrected, though!