Find the largest prime number p<1000 for which there exists a complex number z satisfying
the real and imaginary part of z are both integers;
∣z∣=p, and
there exists a triangle whose three side lengths are p, the real part of z3, and the imaginary part of z3.
解析
Solution
Assume that z=a+bi. Then,
z3=(a3−3ab2)+(3a2b−b3)i
Note that by the Triangle Inequality,
∣(a3−3ab2)−(3a2b−b3)∣<p⟹∣a3+b3−3ab2−3a2b∣<a2+b2
Thus, we know
∣a+b∣∣a2+b2−4ab∣<a2+b2
Without loss of generality, assume a>b (as otherwise, consider i3z=b+ai). If ∣a/b∣≥4, then
17b2≥a2+b2>∣a+b∣∣a2+b2−4ab∣≥∣b−4b∣∣16b2−16b2+b2∣=3b3
`Thus, this means b≤317 or b≤5. Also note that the roots of x2−4x+1 are 2±3, so thus if b≥6,
23b=(2(2−3)−4)b<a<4b
Note that
1000>p=a2+b2≥12b2+b2=13b2
so b2<81, and b<9. If b=8, then 163≤a≤32. Note that gcd(a,b)=1, and a≡b(mod2), so a=29 or 31. However, then 5∣a2+b2, absurd.
If b=7, by similar logic, we have that 143,soa=26.However,onceagain,5\mid a^2+b^2.Ifb=6,bythesamelogic,12\sqrt3, so a=23, where we run into the same problem. Thus b≤5 indeed.
If b=5, note that
(a+5)(a2+25−20a)<a2+25⟹a<20
We note that p=52+182=349 works. Thus, we just need to make sure that if b≤4, a≤18. But this is easy, as
p>(a+b)(a2+b2−4ab)≥(4+18)(42+182−4⋅4⋅18)>1000
absurd. Thus, the answer is 349.
Solution 2
Denote z=a+ib. Thus, a2+b2=p.
Thus,
z3=a(a2−3b2)+ib(−b2+3a2).
Because p, Re(z3), Im(z3) are three sides of a triangle, we have Re(z3)>0 and Im(z3)>0. Thus,
a(a2−3b2)b(−b2+3a2)>0,(1)>0.(2)
Because p, Re(z3), Im(z3) are three sides of a triangle, we have the following triangle inequalities:
Re(z3)+Im(z3)p+Re(z3)p+Im(z3)>p(3)>Im(z3)(4)>Re(z3)(5)
We notice that ∣z3∣=p3/2, and Re(z3), Im(z3), and ∣z3∣ form a right triangle. Thus, Re(z3)+Im(z3)>p3/2. Because p>1, p3/2>p. Therefore, (3) holds.
Conditions (4) and (5) can be written in the joint form as
Re(z3)−Im(z3)<p.(4)
We have
Re(z3)−Im(z3)=(a3−3ab2)−(−b3+3a2b)=(a+b)(a2−4ab+b2)
and p=a2+b2.
Thus, (5) can be written as
(a+b)(a2−4ab+b2)<a2+b2.(6)
Therefore, we need to jointly solve (1), (2), (6). From (1) and (2), we have either a,b>0, or a,b<0. In (6), by symmetry, without loss of generality, we assume a,b>0.
Thus, (1) and (2) are reduced to
a>3b.(7)
Let a=λb. Plugging this into (6), we get
((λ−2)2−3)<b1λ+1λ2+1.(8)
Because p=a2+b2 is a prime, a and b are relatively prime.
Therefore, we can use (7), (8), a2+b2<1000, and a and b are relatively prime to solve the problem.
To facilitate efficient search, we apply the following criteria:
To satisfy (7) and a2+b2<1000, we have 1≤b≤15. In the outer layer, we search for b in a decreasing order. In the inner layer, for each given b, we search for a. Given b, we search for a in the range 3b<a<1000−b2. We can prove that for b≥9, there is no feasible a. The proof is as follows.
For b≥9, to satisfy a2+b2<1000, we have a≤30. Thus, 3<λ≤930. Thus, the R.H.S. of (8) has the following upper bound
b1λ+1λ2+1<b1λ+1λ2+λ=bλ≤9930<2710.
Hence, to satisfy (8), a necessary condition is
((λ−2)2−3)<2710.
However, this cannot be satisfied for 3<λ≤930. Therefore, there is no feasible solution for b≥9. Therefore, we only need to consider b≤8.
We eliminate a that is not relatively prime to b.
We use the following criteria to quickly eliminate a that make a2+b2 a composite number.
For b≡1(mod2), we eliminate a satisfying a≡1(mod2).
For b≡±1(mod5) (resp. b≡±2(mod5)), we eliminate a satisfying a≡±2(mod5) (resp. a≡±1(mod5)).
\item For the remaining (b,a), check whether (8) and the condition that a2+b2 is prime are both satisfied.
The first feasible solution is b=5 and a=18. Thus, a2+b2=349.
\item For the remaining search, given b, we only search for a≥349−b2.
Following the above search criteria, we find the final answer as b=5 and a=18. Thus, the largest prime p is p=a2+b2=(349) .
If a<0,b>0, ∣a2−4ab+b2∣>∣a2+b2∣, making statement (∗) false. Combining with the former graph depicting possible ranges of a,b, by loss of generality, we assume a,b both >0 and exists in the first 30∘ of the circle.
To clearly visualize, we graph out ∣1+λ1+λ2∣ and ∣λ2−4λ+1∣ separately.
When λ is around 2+3, b reaches its maximum upper bound.
b2(1+λ2)<1000b2<66b≤8
Testing values of b in decreasing order, starting from 8, we test out each corresponding value of a(b⋅λ)by trying the two whole numbers closest to the real value of a.