In chapter V of volume 2 of Dickson's "History of the theory of numbers" (p.191-195), which collects results on "rational" triangles (triangles with integer side lengths), apear several results on the so called "Curtius's problem" (also called "the problem of three shooters"). This problem is concerned with establishing a rule for finding rational triangles which also have circumscribed circles with integer radius.

On p.195, appears the following remark:

C.F Gauss, whose attention had been called to Curtius's problem by Schumacher, stated that the sides of every triangle such that each side and the radius $r$ of the circumscribed circle are integers are of the form: $$4abfg(a^2+b^2), \pm4ab(f+g)(a^2f-b^2g), 4ab(a^2f^2+b^2g^2)$$ where $a,b,f,g$ are positive integers, while $r = (a^2+b^2)(a^2f^2+b^2g^2)$... Many writers derived Gauss's formula.

I checked in Gauss's werke, and those formulas are given in a letter to Schumacher from 1847, which appears in volume 12 under the title "solution to arithmetical task". What i want to know is how to put this isolated result in modern mathematical framework? i didn't find much material in the internet on problems related to rational triangles. In particular i'd like to know how is Gauss's solution to this problem related to number theoretical ideas (to which branch of number theory is it related?)?

  • 2
    $\begingroup$ The problem is from the area which is called Diophantine equations. When you write the formula en.wikipedia.org/wiki/Circumscribed_circle relating the sides to the circumradius, you obtain an equation which has to be solved in integers. $\endgroup$ – Alexandre Eremenko Sep 17 '19 at 12:23
  • 2
    $\begingroup$ @ Alexandre Eremenko - Yes, the title of volume 2 of Dickson's book is "Diophantine analysis" (it's one of those question in which it's not clear what i'm asking...). But i'm trying to dig deeper - do Gauss's formulas contain new idea about Diophantine equations? or maybe it's fairly easy to derive them (maybe i'm missing something?)? and if it's a difficult problem then what modern number-theoretic idea is it related to? $\endgroup$ – user2554 Sep 17 '19 at 12:35

"Curtius's problem" is Schumacher's personal name repeated only by Dickson. By scaling, one can reduce the problem to that about triangles with rational sides and the circumradius, which is equivalent to the more commonly considered problem about triangles whose sides and the area are rationals, since $r=abc/4A$. These, or the version where the sides and the area are integers, are now called Heron triangles. The solution, for the integer version, goes back to Brahmagupta, and one can see that Gauss's formulas are related to Brahmagupta's (who stated them in words, of course): $$a=n(m^2+k^2);\ \ b=m(n^2+k^2);\ \ c=(m+n)(mn-k^2).$$ These formulas, just like those for Pythagorean triples, relate Heron triangles to Diophantine equations. The long known connection between Diophantine equations and elliptic curves was made particularly prominent in the Wiles's proof of the Last Fermat theorem, and elliptic curves provide the most common perspective on Heron triangles in recent research, see e.g. Heron Triangles and their Elliptic Curves by Halbeisen-Hungerbuhler. They are also quite popular in recreational mathematics, and e.g. Yiu's survey relates them to Gaussian integers. But I did not see a connection explicitly made to Curtius's version of the problem, or Gauss's contribution to it.

More remotely, with the additional condition that the triangle be right angled, this leads to the still unsolved problem of congruent numbers. The problem goes back to al-Khazin (c. 940 AD), and a number is congruent when it is the area of such a triangle. A simple criterion for a number to be congruent would follow from a very famous conjecture, also about elliptic curves, the Birch and Swinnerton-Dyer conjecture. It is one of the millennium problems.

  • $\begingroup$ thanks for this answer! ( and i'm sorry for the delay in the response). It helps a lot to know that this aspect of "recreational number theory" is connected with the deep topic of elliptic curves. Just one question: can you elaborate in more details how are Gauss's formulas related to Brahmagupta's? Because i couldn't see yet the algebraic connection. $\endgroup$ – user2554 Sep 21 '19 at 13:06
  • $\begingroup$ By the way, i found a new reference to Gauss's result:mathpages.com/home/kmath474.htm, under the title "highly heronian triangles". $\endgroup$ – user2554 Sep 21 '19 at 21:20
  • $\begingroup$ @user2554 You should be able to get Brahmagupta's formulas from Gauss's by reparametrizing and setting one of the parameters to 1. There is also a scaling by 4 to make sure that the radius is an integer (it may only be rational even if the area is an integer). I did not do the calculations. Nice find, I wonder who the author is, they might have done it somewhere. Also, Wikipedia says, contrary to MathPages, that Brahmagupta's formula gives all Heron triangles up to scale, but there is an issue with the scaling factor. I wonder if Gauss's form resolves it. $\endgroup$ – Conifold Sep 21 '19 at 23:55

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.