-
Span
The span of v and w is the set of all their linear combinations.
av + bw , a and b vary over all real numbers.
-
linear dependent
u = av + bw, for some values of a and b.
-
basis
the basis of a vector space is a set of linearly independent vectors that spans the full space.
-
linear transformation
lines remain lines; origin stay at the same space.
grid lines remain parallel and evenly spaced.
matrix expression: 第一列代表 i 变换后的坐标,第二列表示 j 变换后的坐标,以此类推。左乘对应的矩阵可以得到向量变换后的结果。
matrix multiplication: 从右向左,分别是 transform 的顺序。不同个线性组合的复合。order matters!
-
determinant
the transformed area (2d) / volume (3d) are scaled by the determinant
negative: flipping the space / 右手定则变左手定则 $$ det(M_1M_2) = det(M_1)det(M_2) $$
-
rank of A
number of dimensions in the output
-
column space of A
set of all possible outputs Av / span of columns 矩阵的列张成的空间
因此,更精确的rank的定义是列空间的维数。
零向量永远都在列空间中,因为线性变换必须保持在原点位置不变且一直在原点。
满秩的矩阵只有原来的原点会变成零向量,非满秩可能有一条过原点的线/过原点的面。
-
Null space of A
变换后落在原点的向量的集合。
-
Gaussian elimination
-
row echelon form