University of Tokyo science mathematics 2018: problems and solutions

Original publication: Updated:

These solutions are this site's independent explanations, not official University of Tokyo answers. The original author judged the first four problems comparatively accessible, with more geometric work in the last two. Difficulty ratings and score suggestions reflect that historical assessment. See the past-paper review.

Problem 1: a trigonometric function's variation and limits

For 0<x<π0<x<\pi, study the increase and decrease of

f(x)=xsin⁡x+cos⁡x,f(x)=\frac x{\sin x}+\cos x,

and find its limits as x→0+x\to0+ and x→π−x\to\pi-.

Solution

Differentiate and factor:

f′(x)=cos⁡x(sin⁡xcos⁡x−x)sin⁡2x.f'(x)=\frac{\cos x(\sin x\cos x-x)}{\sin^2x}.

Put G(x)=sin⁡xcos⁡x−xG(x)=\sin x\cos x-x. Then G(0)=0G(0)=0 and G′(x)=cos⁡2x−1<0G'(x)=\cos2x-1<0 on (0,π)(0,\pi), so G(x)<0G(x)<0. Thus f′f' is negative before π/2\pi/2, zero there, and positive afterward. The minimum is f(π/2)=π/2f(\pi/2)=\pi/2.

The sine limit gives

lim⁡x→0+f(x)=2.\boxed{\lim_{x\to0+}f(x)=2.}

For the other endpoint, put t=π−x→0+t=\pi-x\to0+:

f(π−t)=π−tsin⁡t−cos⁡t=πsin⁡t−tsin⁡t−cos⁡t⟶+∞.f(\pi-t)=\frac{\pi-t}{\sin t}-\cos t =\frac\pi{\sin t}-\frac t{\sin t}-\cos t\longrightarrow+\infty.

Hence the second limit is +∞+\infty. The divergent numerator term is π\pi, which must not be replaced by tt. The useful calculus habit is to differentiate a sign-controlling factor when its sign is not immediately clear; see standard limits.

Problem 2: binomial coefficients divided by factorials

Define

an=1n!(2n+1n)(n≥1).a_n=\frac1{n!}\binom{2n+1}{n}\quad(n\ge1).
  1. For n≥2n\ge2, write an/an−1=qn/pna_n/a_{n-1}=q_n/p_n in lowest terms, with positive integer denominator pnp_n, and find pn,qnp_n,q_n.
  2. Find all n≥1n\ge1 for which ana_n is an integer.

Part 1

Factorial cancellation gives

anan−1=2(2n+1)n(n+1).\frac{a_n}{a_{n-1}}=\frac{2(2n+1)}{n(n+1)}.

Euclid's algorithm gives gcd⁡(2n+1,n)=gcd⁡(2n+1,n+1)=1\gcd(2n+1,n)=\gcd(2n+1,n+1)=1. Since n(n+1)n(n+1) is even and 2n+12n+1 is odd, the reduced form is

pn=n(n+1)2,qn=2n+1.\boxed{p_n=\frac{n(n+1)}2,\qquad q_n=2n+1.}

Part 2

We have a1=3a_1=3 and a2=5a_2=5. For n≥3n\ge3,

an=a1∏j=2nqjpj.a_n=a_1\prod_{j=2}^{n}\frac{q_j}{p_j}.

Every factor in the product numerator is odd, while the denominator includes p3=6p_3=6. No factor of two in that denominator can cancel with the odd numerator. Thus the result cannot be an integer. Therefore

n=1,2.\boxed{n=1,2.}

A nonintegral last ratio alone would not prove a nonintegral product; the uncancelled factor of two is the decisive argument.

Problem 3: the area swept by two moving points

Let C be the part of y=x2y=x^2 with −1≤x≤1-1\le x\le1. Let O=(0,0)O=(0,0), A=(1,0)A=(1,0), and k>0k>0. As P moves on C and Q moves on segment OA, define R by

OR→=1kOP→+kOQ→.\overrightarrow{OR}=\frac1k\overrightarrow{OP}+k\overrightarrow{OQ}.

Find the area S(k)S(k) of R's region, and its limits as k→0+k\to0+ and k→∞k\to\infty.

Fix the height

Write P=(t,t2)P=(t,t^2), −1≤t≤1-1\le t\le1, and Q=(u,0)Q=(u,0), 0≤u≤10\le u\le1. Then

R=(ku+tk,t2k).R=\left(ku+\frac tk,\frac{t^2}{k}\right).

Fix y=p∈[0,1/k]y=p\in[0,1/k]. The possible tt are ±kp\pm\sqrt{kp}, so the possible x-values are the union of the intervals

[−p/k, k−p/k]and[p/k, k+p/k].\left[-\sqrt{p/k},\ k-\sqrt{p/k}\right] \quad\text{and}\quad \left[\sqrt{p/k},\ k+\sqrt{p/k}\right].

They overlap exactly when p≤k3/4p\le k^3/4. Thus their union has length

Lk(p)={k+2p/k0≤p≤k3/4,2kp≥k3/4,L_k(p)=\begin{cases}k+2\sqrt{p/k}&0\le p\le k^3/4,\\2k&p\ge k^3/4,\end{cases}

restricted in both cases to p≤1/kp\le1/k. Every point of each interval occurs because uu ranges over a full interval; hence this is the actual cross-section, not just an outer bound.

Integrate the union length

The switching height lies within the vertical range exactly when k≤2k\le\sqrt2. Therefore

S(k)=∫0k3/4(k+2p/k) dp+∫k3/41/k2k dp=2−k412(0<k≤2).\begin{aligned} S(k)&=\int_0^{k^3/4}(k+2\sqrt{p/k})\,dp+\int_{k^3/4}^{1/k}2k\,dp\\ &=2-\frac{k^4}{12}\qquad(0<k\le\sqrt2). \end{aligned}

For k≥2k\ge\sqrt2, the intervals overlap at every height, so

S(k)=∫01/k(k+2p/k) dp=1+43k2.S(k)=\int_0^{1/k}(k+2\sqrt{p/k})\,dp =1+\frac4{3k^2}.

Consequently,

S(k)={2−k4/120<k≤2,1+4/(3k2)k≥2.\boxed{S(k)=\begin{cases}2-k^4/12&0<k\le\sqrt2,\\1+4/(3k^2)&k\ge\sqrt2. \end{cases}}

The plus sign in the second expression is essential; the source's minus sign would make the formulas disagree at k=2k=\sqrt2. Both correct expressions give 5/35/3 there.

The limits are

lim⁡k→0+S(k)=2,lim⁡k→∞S(k)=1.\boxed{\lim_{k\to0+}S(k)=2,\qquad\lim_{k\to\infty}S(k)=1.}

Geometrically, when kk is small, most horizontal sections are two separate intervals of length kk, spread over height 1/k1/k; their overlap correction tends to zero, giving area two. When kk is large, the sections overlap throughout and have base length kk, height 1/k1/k, and a vanishing width correction, giving area one. These are checks on the calculation rather than substitutes for its interval analysis. The swept-region guide develops the same coordinate-fixing method.

Problem 4: three cubic roots with the middle root above one

Let a>0a>0 and f(x)=x3−3a2xf(x)=x^3-3a^2x. Find and sketch all (a,b)(a,b) such that f(x)=bf(x)=b has three distinct real roots α<β<γ\alpha<\beta<\gamma, with β>1\beta>1.

Solution

Since f′(x)=3(x−a)(x+a)f'(x)=3(x-a)(x+a), the local maximum is f(−a)=2a3f(-a)=2a^3 and the local minimum is f(a)=−2a3f(a)=-2a^3. Three distinct roots exist exactly when

−2a3<b<2a3.-2a^3<b<2a^3.

Cubic graph with local maximum 2a cubed at -a and local minimum -2a cubed at a.

The middle root lies on the strictly decreasing branch (−a,a)(-a,a). It is greater than one exactly when a>1a>1 and f(1)>bf(1)>b, or b<1−3a2b<1-3a^2.

For a>1a>1, the latter upper bound is already below 2a32a^3. Also 1−3a2>−2a31-3a^2>-2a^3, because their difference is (a−1)2(2a+1)>0(a-1)^2(2a+1)>0. Hence

a>1,−2a3<b<1−3a2.\boxed{a>1,\qquad -2a^3<b<1-3a^2.}

All boundaries are excluded.

Open parameter region between b=-2a cubed and b=1-3a squared for a greater than one.

Separating the horizontal level bb is the key simplification. See root placement and parameter separation.

Problem 5: reflect a point in a circle tangent

Let C be the unit circle centered at zero in the complex plane. Take P=z∈CP=z\in C, with z≠1z\ne1, and reflect A=1A=1 in the tangent to C at P to obtain Q=uQ=u. Put w=1/(1−u)w=1/(1-u).

  1. Express uu and wˉ/w\bar w/w as polynomials in zz, and calculate ∣w+wˉ−1∣/∣w∣|w+\bar w-1|/|w|.
  2. Find the locus of ww when ∣z∣=1|z|=1 and Re⁡z≤1/2\operatorname{Re}z\le1/2.

Part 1: translate reflection into two conditions

The vector from A to Q is normal to the tangent, hence parallel to OP. Thus u−1=λzu-1=\lambda z for a nonzero real λ\lambda. Also P is equidistant from A and Q:

∣z−1∣=∣z−u∣=∣(1−λ)z−1∣.|z-1|=|z-u|=|(1-\lambda)z-1|.

Squaring and using zzˉ=1z\bar z=1 gives

λ2−2λ+λ(z+zˉ)=0.\lambda^2-2\lambda+\lambda(z+\bar z)=0.

Therefore λ=2−z−zˉ\lambda=2-z-\bar z, and

u=−z2+2z.\boxed{u=-z^2+2z.}

This is the actual reflection: the tangent's normal direction gives the perpendicular relation, and equal distances place its midpoint on the tangent. Since z≠1z\ne1, the reflection is nontrivial.

Now 1−u=(z−1)21-u=(z-1)^2, so w=1/(z−1)2≠0w=1/(z-1)^2\ne0. As zˉ=1/z\bar z=1/z,

wˉw=(z−1)2(zˉ−1)2=z2.\boxed{\frac{\bar w}{w}=\frac{(z-1)^2}{(\bar z-1)^2}=z^2.}

Finally,

∣w+wˉ−1∣∣w∣=∣1+wˉw−1w∣=∣1+z2−(z−1)2∣=2.\frac{|w+\bar w-1|}{|w|} =\left|1+\frac{\bar w}{w}-\frac1w\right| =|1+z^2-(z-1)^2|=\boxed{2}.

Part 2: obtain the parabola and its exact interval

Write w=x+iyw=x+iy. The last identity becomes

∣2x−1∣=2x2+y2,x=14−y2.|2x-1|=2\sqrt{x^2+y^2}, \qquad x=\frac14-y^2.

To determine which part is attained, write z=eiθz=e^{i\theta} with π/3≤θ≤5π/3\pi/3\le\theta\le5\pi/3. Using z−1=2isin⁡(θ/2)eiθ/2z-1=2i\sin(\theta/2)e^{i\theta/2},

w=−e−iθ4sin⁡2(θ/2)=14−14cot⁡2θ2+i2cot⁡θ2.w=-\frac{e^{-i\theta}}{4\sin^2(\theta/2)} =\frac14-\frac14\cot^2\frac\theta2 +\frac i2\cot\frac\theta2.

As θ/2\theta/2 increases from π/6\pi/6 to 5π/65\pi/6, its cotangent decreases continuously from 3\sqrt3 to −3-\sqrt3. Therefore every value of yy in [−3/2,3/2][-\sqrt3/2,\sqrt3/2] occurs, giving

x=14−y2,−32≤y≤32.\boxed{x=\frac14-y^2,\qquad-\frac{\sqrt3}{2}\le y\le\frac{\sqrt3}{2}.}

Closed parabola arc x=1/4-y squared, from (-1/2,-square root of three over two) to (-1/2,square root of three over two).

The equality of distances also identifies a parabola with focus zero and directrix x=1/2x=1/2, but the parameter restriction is still needed to select this finite arc.

Problem 6: a union of three rounded tubes

Let

O=(0,0,0),A=(1,0,0),B=(1,1,0),C=(1,1,1),12<r<1.O=(0,0,0),\quad A=(1,0,0),\quad B=(1,1,0),\quad C=(1,1,1), \qquad\frac12<r<1.

For a moving point on OA, AB, or BC, take the closed ball of radius rr centered at that point. The three swept solids are V1,V2,V3V_1,V_2,V_3, respectively.

  1. Find the heights tt for which the plane y=ty=t meets both V1V_1 and V3V_3, and describe their sections in that plane.
  2. Find the condition on rr for V1∩V3⊆V2V_1\cap V_3\subseteq V_2.
  3. Under that condition, let S=Vol⁡(V1)S=\operatorname{Vol}(V_1) and T=Vol⁡(V1∩V2)T=\operatorname{Vol}(V_1\cap V_2). Express the volume of V1∪V2∪V3V_1\cup V_2\cup V_3 using S and T.
  4. Calculate S, T, and the union volume.

Part 1: two capsule sections

The planes meeting V1V_1 have −r≤t≤r-r\le t\le r, while those meeting V3V_3 have 1−r≤t≤1+r1-r\le t\le1+r. Their common range is

1−r≤t≤r.\boxed{1-r\le t\le r.}

Put

at=r2−t2,bt=r2−(1−t)2.a_t=\sqrt{r^2-t^2},\qquad b_t=\sqrt{r^2-(1-t)^2}.

In the xz-plane at that height, the section of V1V_1 is the closed capsule of radius ata_t about the horizontal segment from (0,0)(0,0) to (1,0)(1,0). The section of V3V_3 is the closed capsule of radius btb_t about the vertical segment from (1,0)(1,0) to (1,1)(1,1).

For t≤1/2t\le1/2, at≥bta_t\ge b_t; for t≥1/2t\ge1/2 the inequality reverses. The illustrations show the corresponding two shapes.

Horizontal capsule wider than the vertical capsule in the first height range.

Vertical capsule wider than the horizontal capsule in the second height range.

Part 2: include the intersection in the middle tube

Throughout the relevant height range, 0<t<10<t<1, so the section of V2V_2 is the disk (x−1)2+z2≤r2(x-1)^2+z^2\le r^2.

The farthest points of the two-capsule intersection from (1,0)(1,0) include (1−bt,at)(1-b_t,a_t), at squared distance at2+bt2a_t^2+b_t^2. This is also the maximum. Indeed, in the upper-left quadrant relative to (1,0)(1,0), the capsule constraints give 1−x≤bt1-x\le b_t and z≤atz\le a_t; in the remaining quadrants one of the radius-ata_t or radius-btb_t end disks bounds the distance by a smaller value. The corner lies in both rectangular portions because at,bt<1a_t,b_t<1.

Thus inclusion of the whole section is equivalent to

at2+bt2≤r2⟺r2≤t2+(1−t)2=2(t−1/2)2+1/2.a_t^2+b_t^2\le r^2 \quad\Longleftrightarrow\quad r^2\le t^2+(1-t)^2 =2(t-1/2)^2+1/2.

The allowed interval contains t=1/2t=1/2, where the right side has minimum 1/21/2. Therefore

12<r≤12.\boxed{\frac12<r\le\frac1{\sqrt2}.}

Part 3: inclusion–exclusion

The three individual volumes equal S, and the two adjacent intersections have volume T by congruence. The nonadjacent intersection V1∩V3V_1\cap V_3 equals the triple intersection under the inclusion condition. Those two terms cancel in inclusion–exclusion:

Vol⁡(V1∪V2∪V3)=3S−2T.\boxed{\operatorname{Vol}(V_1\cup V_2\cup V_3)=3S-2T.}

The source has S and T reversed in one displayed answer; the correct expression agrees with its final calculation.

Part 4: calculate the two building blocks

A radius-r ball swept along a unit segment forms a cylinder of radius r and length one, plus two hemispherical caps. Hence

S=πr2+4π3r3.\boxed{S=\pi r^2+\frac{4\pi}{3}r^3.}

For the adjacent intersection V1∩V2V_1\cap V_2, cut at z=pz=p, −r≤p≤r-r\le p\le r. Put ρ=r2−p2\rho=\sqrt{r^2-p^2}. The intersection consists of a square of side ρ\rho in the upper-left quadrant about A and three quarter-disks of radius ρ\rho in the other quadrants. The segment lengths are one and ρ≤r<1\rho\le r<1, so these pieces lie within the relevant tube sections.

Adjacent-tube intersection section comprising one square and three quarter-disks.

Consequently,

T=∫−rr(1+3π4)(r2−p2) dp=(π+43)r3.\begin{aligned} T&=\int_{-r}^{r}\left(1+\frac{3\pi}{4}\right)(r^2-p^2)\,dp\\ &=\boxed{\left(\pi+\frac43\right)r^3.} \end{aligned}

The final volume is

3πr2+(2π−83)r3,12<r≤12.\boxed{3\pi r^2+\left(2\pi-\frac83\right)r^3, \qquad\frac12<r\le\frac1{\sqrt2}.}

Using unit-radius cap volumes would be incorrect: all radius powers must remain in S and T.

Reviewing this paper

The original author's historical practice targets were approximately 90–100 points for the strongest medical-track candidates and 70–80 for other science-track candidates. These are personal strategy suggestions, not official cutoffs. The main review themes are controlling derivative signs, finding uncancelled prime factors, integrating actual unions of intervals, locating a middle root on a monotone branch, proving a locus is fully attained, and applying inclusion–exclusion to solids.

Related articles

About the author

Koki Saito

Physician and graduate of the University of Tokyo Faculty of Medicine. Drawing on his entrance-exam experience, he explains how to choose mathematical methods and construct complete solutions.

Full profile →

Have a question or spotted an error? Contact us.