Free cheat sheets

Precalculus / Chapter 12

Matrices

Cheat sheet

Matrix Operations

  • Matrix dimensions are rows × columns. Addition requires matching dimensions and acts entrywise; scalar multiplication multiplies every entry. Matrix multiplication is different: each output entry is a row-column dot product.
  • If A is m×nm\times n and B is n×pn\times p, then AB is m×pm\times p. Reversing order may change the result or make multiplication undefined.
  • For A=[2103]A=\begin{bmatrix}2&1\\0&3\end{bmatrix}, A[45]=[1315]A\begin{bmatrix}4\\5\end{bmatrix}=\begin{bmatrix}13\\15\end{bmatrix}. Do not treat a matrix product as entrywise multiplication.

The Inverse and Determinant of a Matrix

  • For A=[abcd]A=\begin{bmatrix}a&b\\c&d\end{bmatrix}, det⁡A=ad−bc\det A=ad-bc. If this is nonzero, A−1=1ad−bc[d−b−ca]A^{-1}=\frac1{ad-bc}\begin{bmatrix}d&-b\\-c&a\end{bmatrix} and AA−1=A−1A=IAA^{-1}=A^{-1}A=I.
  • For A=[2101]A=\begin{bmatrix}2&1\\0&1\end{bmatrix}, A−1=[1/2−1/201]A^{-1}=\begin{bmatrix}1/2&-1/2\\0&1\end{bmatrix}. The inverse is not the entrywise reciprocal. Verify by multiplication.
  • A determinant of zero means singular: the map loses information and has no inverse on the whole plane. Area scales by ∣det⁡A∣\lvert\det A\rvert; a negative determinant reverses orientation. Area itself is never negative.

Linear Transformations and Matrices

  • A linear map preserves addition and scalar multiplication: T(u+v)=T(u)+T(v)T(u+v)=T(u)+T(v) and T(cu)=cT(u)T(cu)=cT(u). It sends zero to zero, but that test alone is insufficient; (x,y)↦(x2,y)(x,y)\mapsto(x^2,y) passes the zero test and is not linear.
  • For column vectors, an m×nm\times n matrix maps Rn\mathbb R^n to Rm\mathbb R^m. Its columns are the images of the input basis vectors. If T(1,0)=(2,1)T(1,0)=(2,1) and T(0,1)=(−1,3)T(0,1)=(-1,3), then A=[2−113]A=\begin{bmatrix}2&-1\\1&3\end{bmatrix}.
  • A nonzero translation is affine, not linear. In a matrix map, each output is a sum of constant multiples of the input coordinates, without an added nonzero constant.

Matrices as Functions

  • Counterclockwise rotation by θ uses Rθ=[cos⁡θ−sin⁡θsin⁡θcos⁡θ]R_\theta=\begin{bmatrix}\cos\theta&-\sin\theta\\\sin\theta&\cos\theta\end{bmatrix}. A quarter-turn sends (x,y)(x,y) to (−y,x)(-y,x) and preserves lengths and area.
  • With column vectors, A followed by B is BA: the rightmost matrix acts first. Stretching x by 2 and then rotating a quarter-turn sends (1,1)(1,1) to (−1,2)(-1,2); reversing the order gives (−2,1)(-2,1).
  • If both maps are invertible, (BA)−1=A−1B−1(BA)^{-1}=A^{-1}B^{-1}. A projection (x,y)↦(x,0)(x,y)\mapsto(x,0) cannot be uniquely undone because many inputs share one output.

Matrices Modeling Contexts

  • Using column states, sn+1=Psns_{n+1}=Ps_n: column j describes destinations from current state j. In a closed transition model, entries are nonnegative and every column sums to 1. Counts keep their total; normalized probability states keep total 1.
  • For P=[0.80.30.20.7]P=\begin{bmatrix}0.8&0.3\\0.2&0.7\end{bmatrix}, starting from (100,0)(100,0) gives (80,20)(80,20), then (70,30)(70,30). The entry 0.3 is the fraction moving from B to A. Fractional counts are expectations, not fractions of individual objects.
  • A steady state satisfies Ps=sPs=s. With total 100, this P has steady state (60,40)(60,40) because 0.2A=0.3B0.2A=0.3B. Its repeated updates converge to that balance; not every transition matrix converges. Fixed transfer rates and a closed total are model assumptions.
  • If P is invertible, sn−1=P−1sns_{n-1}=P^{-1}s_n. Here P−1=[1.4−0.6−0.41.6]P^{-1}=\begin{bmatrix}1.4&-0.6\\-0.4&1.6\end{bmatrix}. Current (70,30)(70,30) recovers (80,20)(80,20), but current (10,90)(10,90) recovers (−40,140)(-40,140) and has no nonnegative previous state under the model.

Sign in