Linear Algebra (Pt. 2): 3x3
Last updated
Was this helpful?
Last updated
Was this helpful?
It feels I have written a lot, but still have yet to cover 3x3 problems (fml) 👵.
This is also to highlight the non-linear increase in complexity to solve these equations with every increase in dimension.
In a 3x3 problem, we will require 3 equations with 3 variables (unknowns) to proceed. The following problem is extracted from video lecture 1 of MIT 18.06 on YouTube.
Sticking to the abbreviations used by Strang, the linear equations can be viewed as:
As mentioned previously, the equations for the row picture is left intact. Hence, the coefficient matrix A is structured in a way such that each row are the coefficients of each linear equation.
As now there are 3 unknowns (variables), we would require an additional axis and the picture should be a 3-D one.
You will have to excuse my drawing here (despite having a computer lol), as I am not Rembrandt as well 🤪. We can observe that
Each equation has solutions that are essentially a 2-D plane
While considering 2 planes, its simultaneous solution is essentially a line
While considering 3 planes, its simultaneous solution is essentially a point,
all while bearing that each solution to an equation should not be parallel.
As you see, we cannot easily solve for the point, the solution that simultaneously satisfy all 3 equations. Thus, the row picture proves to be reliable only up till 2x2 problems.
As mentioned in the column picture, the structure of each equations are broken up into its columns (variables). Therefore, the coefficient matrix will describe each unknown (variable), instead of individual equations. The column shorthand is as of follows:
We know the solution is probably z = 1 as Dr. Strang has specially designed z's coefficient matrix to equate to the result. This is purely illustrating for the purpose of understanding instead of solving (use of elimination method @ next part 👵).
Similar to the column picture of 2x2, the coefficient matrices merely represents vectors in 3-D space. Since Col 3 (z) equals to Col 4 (b or RHS), there is no need to go through the whole trouble of finding the linear combination of vectors like 2x2 (thank god 🥵).
