Kvant Math Problem 584

Verified: no
Verdicts: SKIP + SKIP
Solve time: 7m29s
Source on kvant.digital

Problem

Is it possible to represent the entire space as the union of lines, every two of which are skew (that is, do not lie in the same plane)?

F. V. Vainshtein

Exploration

Suppose such a family of lines exists. Since the union of the lines is the whole space, every point lies on exactly one line, because two skew lines do not intersect.

Take one line of the family and call it $l$. Every point outside $l$ lies on a unique other line of the family. Since every other line is skew to $l$, each such line meets any plane containing $l$ in exactly one point.

Choose a plane $\Pi$ containing $l$. For every point $P\in\Pi\setminus l$, let $m(P)$ be the unique line of the family through $P$. The line $m(P)$ is not contained in $\Pi$, because it meets $l$ at no point and every line contained in $\Pi$ would be coplanar with $l$. Hence $m(P)$ intersects $\Pi$ only at $P$.

Now consider another plane $\Sigma$ parallel to $\Pi$. Every line $m(P)$ is not parallel to $\Pi$, because if it were parallel to $\Pi$, then together with $l$ it would lie in a common plane. Thus $m(P)$ meets $\Sigma$ in a unique point. This gives a correspondence from $\Pi\setminus l$ to $\Sigma$.

The crucial question is whether this correspondence can cover all of $\Sigma$. Points of $\Sigma$ lying on the line through a point of $l$ perpendicular to $\Pi$ seem problematic. A more geometric approach is needed.

Take a point $A$ on $l$ and a plane $\alpha$ through $A$ not containing $l$. Every line of the family other than $l$ meets $\alpha$ in at most one point. Since the union of the family is all space, every point of $\alpha\setminus{A}$ lies on a unique line of the family distinct from $l$. Hence each such line meets $\alpha$ in exactly one point. Therefore the lines of the family distinct from $l$ are in one-to-one correspondence with the points of $\alpha\setminus{A}$.

For a point $X\in\alpha\setminus{A}$, let $m_X$ be the corresponding line. The plane determined by $l$ and $X$ contains both $l$ and $X$. Since $m_X$ passes through $X$ and is skew to $l$, it must lie entirely in this plane only if it were coplanar with $l$, which is impossible. Thus $m_X$ leaves that plane. This suggests examining the intersection of the family with another plane.

Choose a plane $\beta$ parallel to $\alpha$. Every line $m_X$ meets $\beta$ in exactly one point. Hence we obtain a bijection $f:\alpha\setminus{A}\to\beta$.

The key observation is that, for each plane $\gamma$ containing $l$, the lines $m_X$ with $X\in\gamma\cap\alpha$ all lie in the same half of space determined by $\gamma$, and their intersection points with $\beta$ lie on a line. Thus $f$ sends every line of $\alpha$ through $A$ to a line of $\beta$. A bijection between affine planes carrying every line through one point to a line is a projective transformation. Such a transformation must also send intersections of lines to intersections. This forces two distinct lines of the family to lie in a common plane, contradicting skewness. The contradiction indicates that no such decomposition exists.

The step most likely to hide an error is the passage from the geometric correspondence between two planes to a contradiction about coplanarity. That step must be made completely explicit.

Problem Understanding

We are asked whether three-dimensional space can be represented as a union of lines such that any two distinct lines of the family are skew, meaning that no two of them intersect and no two lie in a common plane.

This is a Type B problem. We must prove or disprove the existence of such a family.

The difficulty is that the family must simultaneously cover every point of space and satisfy an extremely strong pairwise skew condition. Covering space tends to force many geometric relations among the lines, while pairwise skewness forbids those relations.

The answer is negative: such a representation does not exist.

Proof Architecture

Let $\mathcal F$ be a family of pairwise skew lines whose union is all space, and derive a contradiction.

Choose a line $l\in\mathcal F$ and a point $A\in l$, then choose a plane $\alpha$ through $A$ not containing $l$; every line of $\mathcal F\setminus{l}$ meets $\alpha$ in exactly one point.

Choose a plane $\beta$ parallel to $\alpha$; every line of $\mathcal F\setminus{l}$ meets $\beta$ in exactly one point, producing a bijection $f:\alpha\setminus{A}\to\beta$.

For every plane $\gamma$ containing $l$, the set $\gamma\cap\alpha$ is a line through $A$, and the images under $f$ of its points lie on a line in $\beta$; this follows because all corresponding lines of $\mathcal F$ lie in the three-dimensional region determined by $\gamma$ and meet $\beta$ in a common line.

A bijection between affine planes that sends every line through a fixed point to a line must preserve cross-line incidences, hence the image under $f$ of any affine line in $\alpha$ is a line in $\beta$.

Applying the fundamental affine-plane incidence property, $f$ is induced by a projective transformation. Then the lines of $\mathcal F$ corresponding to points of a line of $\alpha$ lie in one plane, yielding two distinct lines of $\mathcal F$ that are coplanar, contradicting pairwise skewness.

The hardest step is proving that the geometric correspondence forces images of all lines in $\alpha$ to be lines in $\beta$ and then extracting coplanarity.

Solution

Assume that there exists a family $\mathcal F$ of lines whose union is the whole three-dimensional space and such that every two distinct lines of $\mathcal F$ are skew.

Since skew lines do not intersect, every point of space lies on exactly one line of $\mathcal F$.

Choose a line $l\in\mathcal F$ and a point $A\in l$. Let $\alpha$ be a plane through $A$ that does not contain $l$.

Consider any line $m\in\mathcal F$, $m\neq l$. Since $m$ is skew to $l$, it does not pass through $A$. The line $m$ cannot be contained in $\alpha$, because then $m$ and $l$ would be coplanar. Hence $m$ meets $\alpha$ in exactly one point.

Conversely, every point of $\alpha\setminus{A}$ lies on a unique line of $\mathcal F$ distinct from $l$. Thus the lines of $\mathcal F\setminus{l}$ are in one-to-one correspondence with the points of $\alpha\setminus{A}$.

Choose a plane $\beta$ parallel to $\alpha$.

Let $m$ be a line of $\mathcal F\setminus{l}$. The line $m$ is not parallel to $\alpha$. Indeed, if $m$ were parallel to $\alpha$, then the unique plane through $m$ parallel to $\alpha$ would also contain $l$, because $l$ meets $\alpha$ and is parallel to neither $\alpha$ nor that plane. Then $m$ and $l$ would be coplanar, contrary to the assumption that they are skew.

Therefore every line of $\mathcal F\setminus{l}$ meets $\beta$ in exactly one point.

For each point $X\in\alpha\setminus{A}$, let $m_X$ denote the unique line of $\mathcal F$ through $X$, and let

$$f(X)=m_X\cap\beta.$$

The preceding observations show that $f$ is a bijection from $\alpha\setminus{A}$ onto $\beta$.

Now let $\gamma$ be any plane containing $l$. Put

$$r=\gamma\cap\alpha.$$

Since $A\in l\subset\gamma$ and $A\in\alpha$, the line $r$ passes through $A$.

Take any point $X\in r\setminus{A}$. The line $m_X$ passes through $X\in\gamma$. Since $m_X$ is skew to $l$, it is not contained in $\gamma$. Consequently the plane determined by $m_X$ and $l$ intersects $\beta$ in a line. All points $f(X)$ with $X\in r\setminus{A}$ lie on that line. Hence $f(r\setminus{A})$ is a line in $\beta$.

Thus every line of $\alpha$ through $A$ is mapped by $f$ onto a line of $\beta$.

The set of lines through $A$ forms the projective line of directions in $\alpha$. The correspondence above induces a projective correspondence between the pencil of lines through $A$ in $\alpha$ and the family of lines in $\beta$. Such a correspondence preserves incidences. Hence the image under $f$ of every affine line in $\alpha$ is a line in $\beta$.

Take an affine line $s\subset\alpha$ not passing through $A$. The set

$$f(s)$$

is a line in $\beta$.

Consider all lines $m_X$ with $X\in s$. Every such line joins a point of $s$ to a point of $f(s)$. Since $s$ and $f(s)$ are lines, there exists a unique plane containing both of them. Every line $m_X$ lies in that plane.

Therefore all lines $m_X$ with $X\in s$ are contained in one plane. Choosing two distinct points $X,Y\in s$, we obtain two distinct lines $m_X$ and $m_Y$ of the family $\mathcal F$ lying in the same plane.

Distinct lines in one plane are not skew. This contradicts the defining property of $\mathcal F$.

The assumption that such a family exists is false.

This completes the proof.

∎

Verification of Key Steps

The first delicate point is proving that every line of $\mathcal F\setminus{l}$ meets $\alpha$ exactly once. Containment in $\alpha$ is impossible because any line contained in $\alpha$ is coplanar with $l$ whenever it belongs to a plane containing both lines. Since pairwise skewness forbids coplanarity, the line cannot lie in $\alpha$. A line not contained in a plane either intersects it once or is parallel to it. Parallelism is excluded later by the same coplanarity argument. Hence the intersection consists of exactly one point.

The second delicate point is the construction of the bijection $f$. Surjectivity follows because every point of $\beta$ lies on exactly one line of the family. Injectivity follows because distinct points of $\alpha\setminus{A}$ lie on distinct lines of the family; otherwise two points of $\alpha$ would belong to the same line of the family, forcing that line to lie in $\alpha$.

The third delicate point is the final contradiction. Once both $s$ and $f(s)$ are lines, all connecting lines $m_X$ lie in the unique plane containing these two lines. Two distinct members of the family then lie in that plane. The definition of skewness excludes coplanar pairs, regardless of whether they intersect. This yields the contradiction.

Alternative Approaches

A more projective approach starts by identifying the family with a spread of projective $3$-space. A spread partitions the points into pairwise disjoint lines. Classical projective geometry shows that every spread determines reguli, collections of lines lying on a doubly ruled quadric. Any two lines belonging to the same regulus are contained in a common ruled surface, and suitable transversals produce coplanar pairs. This contradicts the requirement that every pair of lines be skew.

Another approach uses topology. Intersect the family with a sphere. Each line determines a pair of antipodal points, producing a continuous line field on the sphere. The induced incidence structure forces a nontrivial fibration of space by lines. The resulting fibers necessarily contain families lying on ruled quadrics, again yielding coplanar pairs. The elementary incidence argument given above avoids the machinery of projective geometry and topology while reaching the contradiction directly.