7

Some of my math teachers said that $i=\sqrt{-1}$ is the wrong definition for $i$ and that the correct definition is $i^2=-1$. If the second definition if true, then does it mean $\sqrt{-1}= ±i$? Which of the two definitions is true?

Masacroso
  • 30,417

4 Answers4

7

From a formal algebraic perspective, $\Bbb{C}$ is the field $\Bbb{R}[x]/(x^2+1)$ which is the field you get by adjoining to $\Bbb{R}$ the roots of $x^2 + 1$ and all the linear combinations of the form $a+ib$, where $a, b \in \Bbb{R}$. You really need to add just one of them (we call it $i$), because the other is just $0 + i(-1)$. Notice that it doesn't matter which we pick as $i$: if I pick one and you picked the other, we would never notice we made different choices.

Infact, the Galois group (automorphisms preserving the original field, thus fixing $\Bbb{R}$ in this case) of $\Bbb{C}$ has exactly two automorphisms: the identity, sending $i$ to itself, and the conjugation, sending $i$ to its opposite. By definition, automorphisms preserve structure, thus confirming the choice of $i$ doesn't really matter.

seldon
  • 1,382
4

The square root symbol does not make sense when applied to a negative number. Granted, people often write it that way for convenience. $i$ is defined by the equation $i^2=-1$. Yes, if there is a field containing a root of $x^2+1$ then if $i$ is such a root so is $-i$. When we define a field such as $\mathbb C$ we are implicitly choosing a root of $x^2+1$.

lulu
  • 70,402
  • 2
    The $\sqrt.$ here is different from the normal $\sqrt. :\Bbb R^{\ge0}\rightarrow\Bbb R^{\ge0}$. We're defining a 'new version' with the same symbol for convenience as it extends the old one. – Shuri2060 Jul 22 '17 at 14:05
  • @Shuri2060 You can't define $i$ and $\sqrt x$ in the one line. That is, you can't define $i$ in terms of $\sqrt x$ and simultaneously define $\sqrt x$ in terms of $i$. – lulu Jul 22 '17 at 17:45
  • It depends, but I think $\sqrt.$ is usually first extended so that its domain is $\Bbb R$ with the requirements that it remains a bijection. Then $i$ is a 'shorthand' for $\sqrt{-1}$ much like $2$ is a shorthand for $S(1)$. – Shuri2060 Jul 22 '17 at 17:48
  • @Shuri2060 Well, I'd define that extension by saying that $x<0\implies \sqrt x=\sqrt {|x| } , i$. With $i$ defined as one of the roots of $x^2+1$ over $\mathbb R$. How would you define it? Is there some nice axiomatic way to do it? Anyway, I have to imagine that the OP means to start with $i$ and then extend the definition of $\sqrt x$, no? – lulu Jul 22 '17 at 17:53
  • I see - in which case I think your way of defining it would probably be best. But perhaps there is some nice axiomatic way of doing it - I'll think about it. – Shuri2060 Jul 22 '17 at 18:12
4

$\sqrt{\cdot}$ is seen in most cases as a function in $[0,\infty)$, so it have a unique value associated to each non-negative real. In this tradition we generally define $i:=\sqrt{-1}$, that can be thought informally as a kind of extension to the classical square root function.

Indeed we can extend the definition of the square root to any complex number, setting it as the principal value of $\sqrt z:=e^{1/2\ln z}$ with the same result.

However is true that $(-i)^2=-1$.

Masacroso
  • 30,417
3

$\sqrt{-1}$ is denoted as $i$ and $i$ is a solution of the algebraic equation $x^2+1=0$. That is why $i^2=-1$.

One could define $i=-\sqrt{-1}$ and using this notation he could built the Complex analysis. That is up to him.

MAN-MADE
  • 5,381