I did not even remember how to do square roots from high school, but I vaguely recall it was similar to long division. (Thankfully I remember how to do long division.) I just went to youtube and refreshed my memory on how to do square roots.

Is there a "long root" process to do nth roots? N can be any integer of 2 or more. (N could be 1 too but that is trivial.) The radicand can be any positive number. The radicand does not hafta be a perfect square, perfect cube, or perfect n-power of anything.

Btw, I am aware of factorization. So $\sqrt{153} = \sqrt{9 * 17} = 3\sqrt{17}$. In this case, what I want to do is something like 2 into 17, but instead of long division, use a "long root" process for putting 2 into 17. The process would go on forever, much like 25/7 goes on forever because the remainder never "settles". I would just stop when I get, say, 3 decimal places or however many I think is accurate enough.


  • $\sqrt{68}$

  • $\sqrt[3]{401}$

  • $\sqrt[7]{50}$

  • $\sqrt[21]{675}$

  • $\sqrt[n]{x}$

  • $\begingroup$ Perhaps you mean something like here? $\endgroup$ – S.C.B. Feb 25 '17 at 12:23
  • $\begingroup$ @S.C.B. Yes but that appears to be a method only for square roots. I want n roots. So, say, $\sqrt[5]{167}$. $\endgroup$ – DrZ214 Feb 25 '17 at 12:28

In a nutshell, the algorithm for square root works because of the identity: $$ y =(a+x)^2 = a^2 + 2ax + x^2 = a^2 + (2a+x) x $$ In the algorithm you iteratively determine increasing (adding one decimal at the time) values for $a$ by looking for the largest value of $x'$ at the appropriate decimal place so that $$y - a^2 \geq (2a+x') x'$$ You then essentially replace $a$ by $a+x'$ and continue for the next decimal place.

I suppose you could do this for e.g. cube roots by looking at $$ y = (a+x)^3 = a^3 + (3 a^2+3x+x^2) x $$ but it looks rather nasty to carry out in practice.

  • $\begingroup$ Does x' mean the derivative of x with respect to x? Or does x' mean x to some power? $\endgroup$ – DrZ214 Feb 25 '17 at 13:40
  • $\begingroup$ No I just wrote $x'$ to distinguish it from the exact value of $x$ for which you have equality. The description I give is like e.g. homeschoolmath.net/teaching/square-root-algorithm.php The method is very similar to a Newton iteration (in which you solve $2ax = y-a^2$, so without the term $x^2$). The Newton method is probably preferable for many reasons and works easily for any $n$'th root. $\endgroup$ – H. H. Rugh Feb 25 '17 at 13:43

I'm having trouble understanding exactly what you are asking about but I'll focus on the following "The process would go on forever, much like 25/7 goes on forever because the remainder never "settles". I would just stop when I get, say, 3 decimal places or however many I think is accurate enough."

You might want to grab a book on introductory real analysis. R. P. Burn's Numbers and Functions has very good chapters on sequences and completeness that will satisfy your needs.

For example, for any $a$, $0\leq a-\frac{\lfloor a10^n \rfloor}{10^n}<\frac{1}{10^n}$ ($\lfloor x \rfloor$ here is called floor $x$, which is the greatest integer less than $x$). From this we can deduce that for any number $a$ there is a sequence of rational numbers which tends to it. This is because the sequence $(1/10^n)$ tends to zero and so by the squeeze rule for null sequences, the sequence $(\frac{\lfloor a10^n \rfloor}{10^n}-a)$ tends to zero, so by the definition of limit we see how the result is proved. Remember that $\frac{\lfloor a10^n \rfloor}{10^n}$ is always rational.

The chapter on completeness will go over how every nth root of a number exists and is unique.

Hope that points you in the right direction.

  • $\begingroup$ The process of long division: pick a number, write it on the top, multiply that number by the divisor, write the product under the dividend, subtract, bring down the next number. I left out some special things like make sure the product is as close to the goal without going above it, sometimes you hafta write a 0, and other stuff. The process works with any 2 numbers, including divisors with 2 or more digits. But now I want a process for roots. I want to know the process for the square root, cube root, 4th root, 5th root, and in general, n root, even multi-digit roots like 21, 150, etc. $\endgroup$ – DrZ214 Feb 25 '17 at 13:19
  • $\begingroup$ The thing you wrote seems to prove that a real root exists for any n and radicand that are real numbers. Just knowing it exists does not help me find a way to extract the root, at least, not that I can see. $\endgroup$ – DrZ214 Feb 25 '17 at 13:21
  • $\begingroup$ @DrZ214, what I wrote does not prove real root exists for any number. It is a sketch for a proof that shows every real number is the limit of a sequence of rational numbers with terminating decimals. I apologise if I have misunderstood your question. Maybe the definition of $nth$ root will help you? for example $3.3=3^2$ so $\sqrt {3^2}=3$, $7.7.7=7^3$ so $\sqrt[3]{7^3}=7$. $\endgroup$ – shredalert Feb 25 '17 at 13:28
  • $\begingroup$ I want any roots of any number, not just perfect powers. $\sqrt{7} = 2.645...$ according to calculator, but what process do you apply to get that? $\sqrt[3]{11} = 2.223...$, but what process do you apply? $\endgroup$ – DrZ214 Feb 25 '17 at 13:39
  • $\begingroup$ You might find this helpful math.stackexchange.com/questions/1102865/… $\endgroup$ – shredalert Feb 25 '17 at 14:07

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.