Preimage of Submodule under Linear Transformation is Submodule
Theorem
Let $\struct {R, +_R, \times_R}$ be a ring.
Let $\struct {G, +_G, \circ_G}_R$ and $\struct {H, +_H, \circ_H}_R$ be $R$-modules.
Let $\phi: G \to H$ be a linear transformation.
Let $N$ be a submodule of $H$.
Then $\phi^{-1} \sqbrk N$ is a submodule of $G$.
Proof
Let $M = \phi^{-1} \sqbrk N$ be the preimage of $N$ under $\phi$.
Aiming for a contradiction, suppose $M$ is not a submodule of $G$.
This means that $M$ does not fulfil all the module axioms.
First suppose that:
- $(1): \quad \struct {M, +_G}$ is not a subgroup of $G$.
Then $M$ is not a group.
Then by the One-Step Subgroup Test:
- $\exists x, y \in M: x +_G \paren {-y} \notin M$
Then:
\(\ds \exists x, y \in M: \, \) | \(\ds x +_G \paren {-y}\) | \(\notin\) | \(\ds M\) | |||||||||||
\(\ds \leadsto \ \ \) | \(\ds \map \phi {x +_G \paren {-y} }\) | \(\notin\) | \(\ds N\) | |||||||||||
\(\ds \leadsto \ \ \) | \(\ds \map \phi x +_H \map \phi {-y}\) | \(\notin\) | \(\ds N\) |
But by definition of $\phi$, both $\map \phi x \in N$ and $\map \phi {-y} \in N$.
Hence $\struct {N, +_G}$ is not closed.
By Group Axiom $\text G 0$: Closure as applied to $N$, it follows that $N$ is not a group.
Hence $N$ is not a subgroup of $H$.
From this contradiction it follows that $(1)$ is false.
Thus $\struct {M, +_G}$ is a subgroup of $G$.
$\Box$
We have by hypothesis that $M$ is not a submodule of $G$.
We have shown that $\struct {M, +_G}$ is a subgroup of $G$.
Hence it must be the case that $M$ is not closed for scalar product:
\(\text {(2)}: \quad\) | \(\ds \exists \lambda \in R, x \in M: \, \) | \(\ds \lambda \circ_G x\) | \(\notin\) | \(\ds M\) | ||||||||||
\(\ds \leadsto \ \ \) | \(\ds \map \phi {\lambda \circ_G x}\) | \(\notin\) | \(\ds N\) | |||||||||||
\(\ds \leadsto \ \ \) | \(\ds \lambda \circ_H {\map \phi x}\) | \(\notin\) | \(\ds N\) |
But by definition of $\phi$, both $\map \phi x \in N$.
Hence $N$ is not closed for scalar product.
From this contradiction it follows that $(2)$ is false.
Thus $\struct {M, +_G}$ is closed for scalar product.
$\Box$
We have shown that $M$ is a subgroup of $G$ which is closed for scalar product.
That is, $M$ is a submodule of $G$.
$\blacksquare$
Sources
- 1965: Seth Warner: Modern Algebra ... (previous) ... (next): Chapter $\text {V}$: Vector Spaces: $\S 28$. Linear Transformations: Theorem $28.2$