1. USM input and the two exponential regimes
We adopt the principal-branch and boundary-value conventions of the USM paper. In particular,
$$\operatorname{arcsec}z:=\arccos(1/z),\qquad z\ne0,$$
with the principal branch of $\arccos$ and the boundary-value convention used there. The part of Theorem 3.2 needed here states that, for $0\lt y\le1$,
$$e^{i\operatorname{arcsec}y}=\tan\left(\frac12\arcsin y\right).\tag{1}$$
For $0\lt y\lt1$, the boundary value of $\operatorname{arcsec}y$ is purely imaginary; consequently, the left-hand side of (1), although written as a complex exponential, is real and positive.
Let $\theta\in(0,\pi)$. Since $\arcsin(\sin\theta)=\min\{\theta,\pi-\theta\}$, setting $y=\sin\theta$ in (1) gives the first coordinate.
Proposition 1.1 (Real-valued exponential coordinate). For $\theta\in(0,\pi)$,
$$\boxed{E(\theta):=e^{i\operatorname{arcsec}(\sin\theta)}=\tan\left(\frac{\min\{\theta,\pi-\theta\}}{2}\right)=\frac{\sin\theta}{1+|\cos\theta|}.}\tag{2}$$
Hence $E(\theta)\in(0,1]$.
To obtain a genuinely complex number from real Euclidean data, use the reciprocal argument $\csc\theta\ge1$. Define $\varepsilon_\theta=1$ for $0\lt\theta\le\pi/2$ and $\varepsilon_\theta=-1$ for $\pi/2\lt\theta\lt\pi$. Since
$$\operatorname{arcsec}(\csc\theta)=\arccos(\sin\theta)=\left|\frac{\pi}{2}-\theta\right|,$$
we obtain an oriented phase.
Proposition 1.2 (Oriented complex phase). For $\theta\in(0,\pi)$,
$$\boxed{\Phi(\theta):=e^{i\varepsilon_\theta\operatorname{arcsec}(\csc\theta)}=e^{i(\pi/2-\theta)}=\sin\theta+i\cos\theta.}\tag{3}$$
The phase has modulus $1$ and is nonreal unless $\theta=\pi/2$.
The two coordinates therefore behave in complementary ways: $E(\theta)$ uses a complex inverse secant but produces a real value, whereas $\Phi(\theta)$ uses a real inverse secant and produces a genuine complex phase.
2. Triangles
Let $ABC$ be a nondegenerate triangle with side lengths $a,b,c$ opposite $A,B,C$, respectively, semiperimeter
$$s=\frac{a+b+c}{2},$$
and area $\Delta\gt0$.
The standard metric relations
$$\sin A=\frac{2\Delta}{bc},\qquad \cos A=\frac{b^2+c^2-a^2}{2bc}\tag{4}$$
give a side-and-area form of the phase.
Proposition 2.1 (Triangle phase coordinate). Define $\Phi_A:=\Phi(A)$ and cyclically. Then
$$\boxed{\Phi_A=e^{i(\pi/2-A)}=\frac{4\Delta+i(b^2+c^2-a^2)}{2bc},}\tag{5}$$
with the analogous formulas for $\Phi_B$ and $\Phi_C$. Moreover,
$$\boxed{\Phi_A\Phi_B\Phi_C=i.}\tag{6}$$
Proof. Equation (5) follows by substituting (4) into (3). Since $A+B+C=\pi$,
$$\Phi_A\Phi_B\Phi_C=e^{i[(\pi/2-A)+(\pi/2-B)+(\pi/2-C)]}=e^{i\pi/2}=i.$$
2.1. A right-triangle specialization
Suppose $a^2+b^2=c^2$ and let $B$ be the acute angle opposite $b$. Then $\sin B=b/c$ and $\cos B=a/c$. Proposition 1.1 therefore gives
$$\boxed{e^{i\operatorname{arcsec}(b/c)}=\frac{b/c}{1+a/c}=\frac{b}{a+c}=\frac{c-a}{b}.}\tag{7}$$
The last equality follows from $b^2=c^2-a^2=(c-a)(c+a)$. Thus the real-valued USM exponential becomes a purely rational expression in the side lengths.
2.2. Heron’s formula from the phase product
Set
$$X_A=b^2+c^2-a^2,\qquad X_B=c^2+a^2-b^2,\qquad X_C=a^2+b^2-c^2.\tag{8}$$
Substitution of the metric forms into (6) yields
$$(4\Delta+iX_A)(4\Delta+iX_B)(4\Delta+iX_C)=8ia^2b^2c^2.\tag{9}$$
The right-hand side has zero real part. Hence
$$(4\Delta)^3-4\Delta(X_AX_B+X_BX_C+X_CX_A)=0.$$
Since $\Delta\gt0$,
$$16\Delta^2=X_AX_B+X_BX_C+X_CX_A.\tag{10}$$
A direct expansion gives
$$X_AX_B+X_BX_C+X_CX_A=2a^2b^2+2b^2c^2+2c^2a^2-a^4-b^4-c^4,$$
and
$$2a^2b^2+2b^2c^2+2c^2a^2-a^4-b^4-c^4=(a+b+c)(-a+b+c)(a-b+c)(a+b-c).$$
Since the four factors are $2s$, $2(s-a)$, $2(s-b)$, and $2(s-c)$, respectively, (10) gives:
Theorem 2.2 (Heron).
$$\boxed{\Delta^2=s(s-a)(s-b)(s-c).}\tag{11}$$
The role of the exponential in this proof is explicit: the additive closure $A+B+C=\pi$ becomes the multiplicative closure (6), while each phase packages the sine and cosine metric data of one vertex into a single complex number.
2.3. Comparison with Edwards’s complex proof
A different complex-number proof of Heron’s formula was published by Miles Dillon Edwards in 2007; an online reproduction of Edwards’s proof is also available. It is worth distinguishing that argument from the phase proof above, because the two share the same broad mechanism—turning an angle-sum relation into multiplication in $\mathbb C$—but use different geometric coordinates.
In Edwards’s construction, let $I$ be the incenter, let $r$ be the inradius, and write the tangent-length decomposition
$$a=y+z,\qquad b=x+z,\qquad c=x+y,$$
so that
$$s=x+y+z,\qquad x=s-a,\qquad y=s-b,\qquad z=s-c.$$
The three right triangles formed by the inradius give angles $\alpha,\beta,\gamma$ at $I$ satisfying
$$\alpha+\beta+\gamma=\pi.$$
The complex numbers $r+ix$, $r+iy$, and $r+iz$ have arguments $\alpha$, $\beta$, and $\gamma$, respectively. Their product is therefore negative real, and its imaginary part vanishes:
$$0=\operatorname{Im}\left[(r+ix)(r+iy)(r+iz)\right]=r^2(x+y+z)-xyz.\tag{12}$$
Thus
$$r^2=\frac{xyz}{s}=\frac{(s-a)(s-b)(s-c)}{s},$$
and $\Delta=rs$ gives Heron’s formula.
The contrast with the present proof is precise. Edwards uses non-unit complex numbers attached to the incenter right triangles and extracts an imaginary part because their product lies on the real axis. Here the coordinates are unit phases attached directly to the vertex angles, and after metric substitution one extracts a real part because the phase product equals $i$. The present construction does not use the incircle; more importantly for the purpose of this article, the same phase language extends to the general quadrilateral identity of Section 3 and its modulus form in Section 4.
3. General convex quadrilaterals
Let $ABCD$ be a convex quadrilateral with
$$AB=a,\qquad BC=b,\qquad CD=c,\qquad DA=d,$$
semiperimeter
$$s=\frac{a+b+c+d}{2},$$
area $K\gt0$, and opposite angles
$$\alpha=\angle BAD,\qquad \gamma=\angle BCD.$$
Define
$$P:=(s-a)(s-d),\qquad Q:=(s-b)(s-c).\tag{13}$$
3.1. The general half-angle identities
The following pair was established by the present author in “Two Identities and their Consequences”, Theorem 5, as a generalization of the cyclic half-angle identities in Theorem 1 of that article. We include a short proof for completeness.
Lemma 3.1. For every convex quadrilateral as above,
$$ad\sin^2\frac{\alpha}{2}+bc\cos^2\frac{\gamma}{2}=P,\tag{14}$$
and
$$bc\sin^2\frac{\gamma}{2}+ad\cos^2\frac{\alpha}{2}=Q.\tag{15}$$
Proof. Let $p=BD$. The law of cosines in triangles $ABD$ and $BCD$ gives
$$p^2=a^2+d^2-2ad\cos\alpha=b^2+c^2-2bc\cos\gamma,$$
so
$$ad\cos\alpha-bc\cos\gamma=\frac{a^2+d^2-b^2-c^2}{2}.\tag{16}$$
Using the half-angle formulas,
$$ad\sin^2\frac{\alpha}{2}+bc\cos^2\frac{\gamma}{2}=\frac{ad+bc+bc\cos\gamma-ad\cos\alpha}{2},$$
hence
$$ad\sin^2\frac{\alpha}{2}+bc\cos^2\frac{\gamma}{2}=\frac{ad+bc}{2}+\frac{b^2+c^2-a^2-d^2}{4}=(s-a)(s-d)=P.$$
The second identity is obtained similarly, or by subtracting the first from $ad+bc=P+Q$.
Two elementary consequences of (13) are
$$P+Q=ad+bc,\qquad P-Q=\frac{b^2+c^2-a^2-d^2}{2}.\tag{17}$$
Combining the second relation with (16) gives the useful equivalent form
$$Q-P=ad\cos\alpha-bc\cos\gamma.\tag{18}$$
This is the bridge from the generalized half-angle identities to the imaginary part of the complex relation below.
3.2. The general complex-exponential identity
Define the oriented phases
$$\Phi_\alpha:=\Phi(\alpha)=e^{i(\pi/2-\alpha)},\qquad \Phi_\gamma:=\Phi(\gamma)=e^{i(\pi/2-\gamma)}.$$
Thus
$$\Phi_\alpha=\sin\alpha+i\cos\alpha,\qquad \overline{\Phi_\gamma}=\sin\gamma-i\cos\gamma.$$
The decomposition of the quadrilateral along $BD$ supplies the real part:
$$2K=ad\sin\alpha+bc\sin\gamma.\tag{19}$$
The generalized half-angle identities, through (18), supply the imaginary part. Combining the two gives the central identity.
Theorem 3.2. For every convex quadrilateral $ABCD$,
$$\boxed{ad\,\Phi_\alpha+bc\,\overline{\Phi_\gamma}=2K+i(Q-P)=2K+\frac{i}{2}(a^2+d^2-b^2-c^2).}\tag{20}$$
Equivalently,
$$\boxed{ad\,e^{i(\pi/2-\alpha)}+bc\,e^{-i(\pi/2-\gamma)}=2K+\frac{i}{2}(a^2+d^2-b^2-c^2).}\tag{21}$$
Using the principal inverse secant, the phase factors appearing in (20) can also be written as
$$\Phi_\alpha=e^{i\varepsilon_\alpha\operatorname{arcsec}(\csc\alpha)},\qquad \overline{\Phi_\gamma}=e^{-i\varepsilon_\gamma\operatorname{arcsec}(\csc\gamma)},$$
with $\varepsilon$ defined as in Proposition 1.2.
Proof. By definition,
$$ad\Phi_\alpha+bc\overline{\Phi_\gamma}=ad(\sin\alpha+i\cos\alpha)+bc(\sin\gamma-i\cos\gamma),$$
so
$$ad\Phi_\alpha+bc\overline{\Phi_\gamma}=(ad\sin\alpha+bc\sin\gamma)+i(ad\cos\alpha-bc\cos\gamma).$$
Now apply (19) and (18).
Thus (20) packages two real metric identities into one complex relation:
$$\operatorname{Re}:\qquad 2K=ad\sin\alpha+bc\sin\gamma,$$
and
$$\operatorname{Im}:\qquad Q-P=ad\cos\alpha-bc\cos\gamma=\frac{a^2+d^2-b^2-c^2}{2}.$$
4. Bretschneider from the modulus
In the earlier MATINF article, Bretschneider’s formula was obtained by multiplying the two generalized half-angle identities and rearranging the result (Theorem 6). The calculation below packages the same underlying metric data into the modulus of the single complex relation (20).
Taking the squared modulus of (20) gives
$$4K^2+\frac14(a^2+d^2-b^2-c^2)^2=|ad\Phi_\alpha+bc\overline{\Phi_\gamma}|^2,$$
hence
$$4K^2+\frac14(a^2+d^2-b^2-c^2)^2=a^2d^2+b^2c^2+2abcd\operatorname{Re}(\Phi_\alpha\Phi_\gamma),$$
and therefore
$$4K^2+\frac14(a^2+d^2-b^2-c^2)^2=a^2d^2+b^2c^2-2abcd\cos(\alpha+\gamma).\tag{22}$$
The last equality uses
$$\Phi_\alpha\Phi_\gamma=e^{i[\pi-(\alpha+\gamma)]}.$$
From (17),
$$4PQ=(ad+bc)^2-\frac14(a^2+d^2-b^2-c^2)^2.\tag{23}$$
Substituting (23) into (22), and using
$$1+\cos(\alpha+\gamma)=2\cos^2\frac{\alpha+\gamma}{2},$$
yields the classical general area formula.
Theorem 4.1 (Bretschneider). For every convex quadrilateral,
$$\boxed{K^2=(s-a)(s-b)(s-c)(s-d)-abcd\cos^2\left(\frac{\alpha+\gamma}{2}\right).}\tag{24}$$
Thus Bretschneider’s formula is the modulus statement associated with the complex-exponential identity (20). In this sense, the modulus proof is a complex-exponential reformulation of the half-angle route used in the earlier MATINF paper.
5. Particular cases
5.1. Cyclic quadrilateral: Brahmagupta
If $ABCD$ is cyclic, then
$$\alpha+\gamma=\pi.$$
Therefore
$$\cos^2\left(\frac{\alpha+\gamma}{2}\right)=0,$$
and (24) immediately becomes:
Corollary 5.1 (Brahmagupta).
$$\boxed{K^2=(s-a)(s-b)(s-c)(s-d).}\tag{25}$$
Cyclicity makes the two phase factors appearing in (20) coincide. Since $\gamma=\pi-\alpha$,
$$\overline{\Phi_\gamma}=\Phi_\alpha,$$
so (20) becomes
$$(ad+bc)\Phi_\alpha=2K+\frac{i}{2}(a^2+d^2-b^2-c^2),$$
or
$$\boxed{e^{i(\pi/2-\alpha)}=\frac{4K+i(a^2+d^2-b^2-c^2)}{2(ad+bc)}.}\tag{26}$$
This is the cyclic quadrilateral analogue of the triangle phase coordinate.
The general half-angle identities also collapse to the familiar cyclic pair. Indeed,
$$\cos^2\frac{\gamma}{2}=\sin^2\frac{\alpha}{2},\qquad \sin^2\frac{\gamma}{2}=\cos^2\frac{\alpha}{2},$$
so (14)–(15) give
$$\boxed{\sin^2\frac{\alpha}{2}=\frac{(s-a)(s-d)}{ad+bc},\qquad \cos^2\frac{\alpha}{2}=\frac{(s-b)(s-c)}{ad+bc}.}\tag{27}$$
These are precisely the cyclic identities used as the starting point in the earlier MATINF article, Theorem 1.
5.2. Bicentric quadrilateral
If the cyclic quadrilateral is also tangential, Pitot’s relation gives
$$a+c=b+d=s.$$
Hence
$$s-a=c,\qquad s-b=d,\qquad s-c=a,\qquad s-d=b.$$
Brahmagupta therefore reduces to:
Corollary 5.2 (Bicentric area formula).
$$\boxed{K=\sqrt{abcd}.}\tag{28}$$
The same area formula was treated in the earlier MATINF paper, Theorem 4; here it appears as an immediate specialization of the Brahmagupta branch of the complex-exponential framework.
5.3. Heron as a degenerate cyclic limit
Let one side of a cyclic quadrilateral tend to zero, say $d\to0^+$, while the remaining sides tend to those of a nondegenerate triangle. Then
$$s\longrightarrow\frac{a+b+c}{2},$$
and Brahmagupta’s formula tends to
$$K^2=s(s-a)(s-b)(s-c),$$
which is Heron’s formula. This limiting route is complementary to the independent triangle phase proof in Section 2. It is a degenerate limit rather than a specialization within the class of convex quadrilaterals, since the limiting side length is zero.
5.4. What cyclicity means in phase language
For a general quadrilateral, the complex identity (20) contains the two opposite-angle phase factors $\Phi_\alpha$ and $\overline{\Phi_\gamma}$. Cyclicity is exactly the condition
$$\alpha+\gamma=\pi\quad\Longleftrightarrow\quad\overline{\Phi_\gamma}=\Phi_\alpha.$$
Thus a cyclic quadrilateral is precisely the case in which the two phase factors appearing in the general identity collapse to one. In this sense, Brahmagupta’s formula is the one-phase specialization of the two-phase Bretschneider identity.
6. Summary and outlook
1. For $\theta\in(0,\pi)$, the USM secant–sine identity provides the real-valued exponential coordinate $E(\theta)$.
2. Passing to the reciprocal argument and orienting the principal inverse secant gives the unit phase
$$\Phi(\theta)=e^{i(\pi/2-\theta)}=\sin\theta+i\cos\theta.$$
3. For triangles, the closure $A+B+C=\pi$ becomes $\Phi_A\Phi_B\Phi_C=i$; the real part of the resulting metric product yields Heron’s formula. This phase proof is structurally related to, but geometrically distinct from, Edwards’s earlier incenter-based complex proof.
4. For general convex quadrilaterals, the generalized opposite-angle half-angle identities from the earlier MATINF paper give $Q-P=ad\cos\alpha-bc\cos\gamma$. Together with the area decomposition, this yields the two-phase relation
$$ad\Phi_\alpha+bc\overline{\Phi_\gamma}=2K+i(Q-P)=2K+\frac{i}{2}(a^2+d^2-b^2-c^2).$$
Its modulus yields Bretschneider’s formula.
5. Under cyclicity, the two phase factors $\Phi_\alpha$ and $\overline{\Phi_\gamma}$ coincide, Bretschneider reduces to Brahmagupta, and the general half-angle formulas reduce to the classical cyclic half-angle identities. The bicentric formula follows as a specialization, while Heron is recovered again as a degenerate cyclic limit.
No hay comentarios:
Publicar un comentario