or transpose(A) if the conj is returned
If there are more unknowns than
equations in a homogeneous linear
system, then it has nontrivial solutions
only for +ve powers
Use for Ax=b when
det(A) is non 0 (unique
solutions)
Note: fr A(B-C)=0, not necessary fr B=C (consider det)
obv A is not 0
count pivot cols
must be unique, can't be
combis of other vectors
intersecting sets: look for vectors dependent on both sets, so
in the rref look fr non-pivots
EROs: preserve row and null space, not column space
Rank: leading entry count = dim(Row(A)) = dim(Col(A)) = Rank of a square matrix - but they preserve the lin r/s btwn cols
- row(A), col(A), null(A) are subspaces that can be
Equivalent statements: described using basis
A is invertible (RREF(A) = I) Basis stuff
Ax = 0 has only the trivial solution (otherwise there is linear dependence i.e. no basis fr Rn - row: pivots/non-0s in rref (the pivots in the original
A can be written as the product of elementary matrices are lin dep)
Rows & Columns of A are basis of Rn - cols: pivots in original
Non-zero determinant
rank(A) = n
- full rank = rank(A) = min{col count, row count}
Nullity: No. of non-pivot columns in RREF(A), this gives the solution to Ax=0
Rank(A) + nullity(A) = n where A is mxn
A is an m x n matrix For A = PB where A, B are 4x5 and P is invertible,
Col(A): subspace of Rm nullity(A)=nullity(B) since both are row equivalent
Basis: Cols of A corresponding to RREF
Rank(A) = pivot cols count in RREF
Row(A): subspace of Rn
Basis: Non-zero rows in RREF
Rank(A) = number of non-zero rows in RREF
Null(A): subspace of Rn
Basis: Vectors in general solution to Ax=0
Nullity(A) = number of non-pivots in RREF
Full rank: when rank(A) = number of cols/rows
Null(A-transpose A) = Null(A) = {0};
At*A is invertible, A has a left inverse
A*At is invertible, A has a right inverse
v1 = u1;
v2 = u2 - dot(u2,v1)/dot(v1,v1)*v1;
v3 = u3 - dot(u3,v1)/dot(v1,v1)*v1 - dot(u3,v2)/dot(v2,v2)*v2;
v4 = u4 - dot(u4,v1)/dot(v1,v1)*v1 - dot(u4,v2)/dot(v2,v2)*v2 - dot(u4,v3)/dot(v3,v3)*v3;
MATLAB Commands
A = sym([ ]);
[P D] = eig(A)
ok since both are in
same E space
