uniolympiads English

IMC 2009, Ημέρα 1, Θέμα 5

Let nn be a positive integer. An nn-simplex in Rn\mathbb{R}^n is given by n+1n + 1 points P0,P1,…,PnP_0, P_1, \dots, P_n, called its vertices, which do not all belong to the same hyperplane. For every nn-simplex SS we denote by v(S)v(S) the volume of SS, and we write C(S)C(S) for the center of the unique sphere containing all the vertices of SS.

Suppose that PP is a point inside an nn-simplex SS. Let SiS_i be the nn-simplex obtained from SS by replacing its ii-th vertex by PP. Prove that

v(S0)C(S0)+v(S1)C(S1)+⋯+v(Sn)C(Sn)=v(S)C(S).v(S_0)C(S_0) + v(S_1)C(S_1) + \dots + v(S_n)C(S_n) = v(S)C(S).
Πρωτότυπο από το επίσημο PDFLet n be a positive integer. An n-simplex in R n is given by n + 1 points P 0 , P 1 , . . . , P n , called its vertices, which do not all belong to the same hyperplane. For every n-simplex S we denote by v(S) the volume of S, and we write C(S) for the center of the unique sphere containing all the vertices of S. Suppose that P is a point inside an n-simplex S. Let S i be the n-simplex obtained from S by replacing its i-th vertex by P . Prove that v(S 0 )C(S 0 ) + v(S 1 )C(S 1 ) + · · · + v(S n )C(S n ) = v(S)C(S).

Η μεταγραφή συγκρίθηκε σύμβολο προς σύμβολο με το πρωτότυπο, όχι ακόμα από άνθρωπο.

Χρόνος σε αυτό το θέμα 0:00

Δούλεψε τουλάχιστον 45 λεπτά πριν ανοίξεις τη λύση.

Επίσημη λύση

Solution 1. We will prove this by induction on nn, starting with n=1n = 1. In that case we are given an interval [a,b][a,b] with a point p∈(a,b)p \in (a,b), and we have to verify

(b−p)b+p2+(p−a)p+a2=(b−a)b+a2,(b-p)\frac{b+p}{2} + (p-a)\frac{p+a}{2} = (b-a)\frac{b+a}{2},
which is true.

Now let assume the result is true for n−1n-1 and prove it for nn. We have to show that the point

X=∑j=0nv(Sj)v(S)O(Sj)X = \sum_{j=0}^n \frac{v(S_j)}{v(S)}O(S_j)
has the same distance to all the points P0,P1,…,PnP_0, P_1, \ldots, P_n. Let i∈{0,1,2,…,n}i \in \{0,1,2,\ldots,n\} and define the sets Mi={P0,P1,…,Pi−1,Pi+1,…,Pn}M_i = \{P_0, P_1, \ldots, P_{i-1}, P_{i+1}, \ldots, P_n\}. The set of all points having the same distance to all points in MiM_i is a line hih_i orthogonal to the hyperplane EiE_i determined by the points in MiM_i. We are going to show that XX lies on every hih_i. To do so, fix some index ii and notice that
X=v(Si)v(S)O(Si)+v(S)−v(Si)v(S)⋅∑j≠iv(Sj)v(S)−v(Si)O(Sj)⏟YX = \frac{v(S_i)}{v(S)}O(S_i) + \frac{v(S)-v(S_i)}{v(S)}\cdot\underbrace{\sum_{j\ne i}\frac{v(S_j)}{v(S)-v(S_i)}O(S_j)}_{Y}
and O(Si)O(S_i) lies on hih_i, so that it is enough to show that YY lies on hih_i.

A map f:R>0→Rnf : \mathbb{R}_{>0} \to \mathbb{R}^n will be called affine if there are points A,B∈RnA, B \in \mathbb{R}^n such that f(λ)=λA+(1−λ)Bf(\lambda) = \lambda A + (1-\lambda)B. Consider the ray gg starting in PiP_i and passing through PP. For λ>0\lambda > 0 let Pλ=(1−λ)P+λPiP_\lambda = (1-\lambda)P + \lambda P_i, so that PλP_\lambda is an affine function describing the points of gg. For every such λ\lambda let SjλS_j^\lambda be the nn-simplex obtained from SS by replacing the jj-th vertex by PλP_\lambda. The point O(Sjλ)O(S_j^\lambda) is the intersection of the fixed line hjh_j with the hyperplane orthogonal to gg and passing through the midpoint of the segment PiPλ‾\overline{P_iP_\lambda} which is given by an affine function. This implies that also O(Sjλ)O(S_j^\lambda) is an affine function. We write φj=v(Sj)v(S)−s(Si)\varphi_j = \frac{v(S_j)}{v(S)-s(S_i)}, and then

Yλ=∑j≠iφjO(Sjλ)Y_\lambda = \sum_{j\ne i}\varphi_jO(S_j^\lambda)
is an affine function. We want to show that Yλ∈hiY_\lambda \in h_i for all λ\lambda (then specializing to λ=1\lambda = 1 gives the desired result). It is enough to do this for two different values of λ\lambda.

Let gg intersect the sphere containing the vertices of SS in a point ZZ; then Z=PλZ = P_\lambda for a suitable λ>0\lambda > 0, and we have O(Sjλ)=O(S)O(S_j^\lambda) = O(S) for all jj, so that Yλ=O(S)∈hiY_\lambda = O(S) \in h_i. Now let gg intersect the hyperplane EiE_i in a point QQ; then Q=PλQ = P_\lambda for some λ>0\lambda > 0, and QQ is different from ZZ. Define TT to be the (n−1)(n-1)-simplex with vertex set MiM_i, and let TjT_j be the (n−1)(n-1)-simplex obtained from TT by replacing the vertex PjP_j by QQ. If we write v′v' for the volume of (n−1)(n-1)-simplices in the hyperplane EiE_i, then

v′(Tj)v′(T)=v(Sjλ)v(S)=v(Sjλ)∑k≠iv(Skλ)=λv(Sj)∑k≠iλv(Sk)=v(Sj)v(S)−v(Si)=φj.\begin{aligned} \frac{v'(T_j)}{v'(T)} = \frac{v(S_j^\lambda)}{v(S)} &= \frac{v(S_j^\lambda)}{\sum_{k\ne i}v(S_k^\lambda)} \\ &= \frac{\lambda v(S_j)}{\sum_{k\ne i}\lambda v(S_k)} = \frac{v(S_j)}{v(S)-v(S_i)} = \varphi_j. \end{aligned}
If pp denotes the orthogonal projection onto EiE_i then p(O(Sjλ))=O(Tj)p(O(S_j^\lambda)) = O(T_j), so that p(Yλ)=∑j≠iφjO(Tj)p(Y_\lambda) = \sum_{j\ne i}\varphi_jO(T_j) equals O(T)O(T) by induction hypothesis, which implies Yλ∈p−1(O(T))=hiY_\lambda \in p^{-1}(O(T)) = h_i, and we are done.

Solution 2. For n=1n = 1, the statement is checked easily.

Assume n≥2n \ge 2. Denote O(Sj)−O(S)O(S_j) - O(S) by qjq_j and Pj−PP_j - P by pjp_j. For all distinct jj and kk in the range 0,…,n0, \ldots, n the point O(Sj)O(S_j) lies on a hyperplane orthogonal to pkp_k and PjP_j lies on a hyperplane orthogonal to qkq_k. So we have

{⟨pi,qj−qk⟩=0⟨qi,pj−pk⟩=0\begin{cases} \langle p_i, q_j - q_k\rangle = 0 \\ \langle q_i, p_j - p_k\rangle = 0 \end{cases}
for all j≠i≠kj \ne i \ne k. This means that the value ⟨pi,qj⟩\langle p_i, q_j\rangle is independent of jj as long as j≠ij \ne i, denote this value by λi\lambda_i. Similarly, ⟨qi,pj⟩=μi\langle q_i, p_j\rangle = \mu_i for some μi\mu_i. Since n≥2n \ge 2, these equalities imply that all the λi\lambda_i and μi\mu_i values are equal, in particular, ⟨pi,qj⟩=⟨pj,qi⟩\langle p_i, q_j\rangle = \langle p_j, q_i\rangle for any ii and jj.

We claim that for such pip_i and qiq_i, the volumes

Vj=∣det⁡(p0,…,pj−1,pj+1,…,pn)∣V_j = |\det(p_0, \ldots, p_{j-1}, p_{j+1}, \ldots, p_n)|
and
Wj=∣det⁡(q0,…,qj−1,qj+1,…,qn)∣W_j = |\det(q_0, \ldots, q_{j-1}, q_{j+1}, \ldots, q_n)|
are proportional. Indeed, first assume that p0,…,pn−1p_0, \ldots, p_{n-1} and q0,…,qn−1q_0, \ldots, q_{n-1} are bases of Rn\mathbb{R}^n, then we have
Vj=1∣det⁡(q0,…,qn−1)∣∣det⁡((⟨pk,ql⟩))k≠jl<n∣==1∣det⁡(q0,…,qn−1)∣∣det⁡((⟨pk,ql⟩))l≠jk<n∣=∣det⁡(p0,…,pn−1)det⁡(q0,…,qn−1)∣Wj.\begin{aligned} V_j &= \frac{1}{|\det(q_0, \ldots, q_{n-1})|}\left|\det\left((\langle p_k, q_l\rangle)\right)_{\substack{k\ne j\\ l<n}}\right| = \\ &= \frac{1}{|\det(q_0, \ldots, q_{n-1})|}\left|\det((\langle p_k, q_l\rangle))_{\substack{l\ne j\\ k<n}}\right| = \left|\frac{\det(p_0, \ldots, p_{n-1})}{\det(q_0, \ldots, q_{n-1})}\right|W_j. \end{aligned}
If our assumption did not hold after any reindexing of the vectors pip_i and qiq_i, then both pip_i and qiq_i span a subspace of dimension at most n−1n-1 and all the volumes are 0.

Finally, it is clear that ∑qjWj/det⁡(q0,…,qn)=0\sum q_jW_j/\det(q_0, \ldots, q_n) = 0: the weight of pjp_j is the height of 0 over the hyperplane spanned by the rest of the vectors qkq_k relative to the height of pjp_j over the same hyperplane, so the sum is parallel to all the faces of the simplex spanned by q0,…,qnq_0, \ldots, q_n. By the argument above, we can change the weights to the proportional set of weights Vj/det⁡(p0,…,pn)V_j/\det(p_0, \ldots, p_n) and the sum will still be 0. That is,

0=∑qjVjdet⁡(p0,…,pn)=∑(O(Sj)−O(S))v(Sj)v(S)==1v(S)(∑O(Sj)v(Sj)−O(S)∑v(Sj))=1v(S)(∑O(Sj)v(Sj)−O(S)v(S)),\begin{aligned} 0 &= \sum q_j\frac{V_j}{\det(p_0, \ldots, p_n)} = \sum(O(S_j) - O(S))\frac{v(S_j)}{v(S)} = \\ &= \frac{1}{v(S)}\left(\sum O(S_j)v(S_j) - O(S)\sum v(S_j)\right) = \frac{1}{v(S)}\left(\sum O(S_j)v(S_j) - O(S)v(S)\right), \end{aligned}
q.e.d.

Μεταγραφή από το επίσημο κείμενο λύσεων, σελίδες 3–4. Η μεταγραφή δεν έχει ελεγχθεί δεύτερη φορά· αν έχεις αμφιβολία, κοίτα το πρωτότυπο.

Θέμα αρχείου: η εκφώνηση είναι το επίσημο κείμενο (μεταγραμμένο, με το πρωτότυπο παραπάνω). Η ταξινόμηση ανά αντικείμενο και η δυσκολία εκτιμήθηκαν αυτόματα· υποδείξεις και λύση στα ελληνικά προστίθενται σταδιακά.

Πώς πήγε;