The vector space of differentiable functions. Let C ∞ (R) denote the set of all infinitely differentiable functions f : R → R. Then C ∞ (R) is a vector space, using the usual notions of addition and scalar multiplication for functions. For instance, if f is the function f (x) = ex , and g is the function g(x) = sin(x), then 2f is the function 2f (x) = 2ex and f + g is the function (f + g)(x) = ex + sin(x).
Problem 1: Differentiation. Differentiation defines a function D : C ∞ (R) → C ∞ (R). This function D is defined by D(f ) = f 0 . (a) Calculate D(ex ), D(x3 + x), D(cos(x)).
(b) Show, by calculating both sides, that D(2ex + 3 cos(x)) = 2D(ex ) + 3D(cos(x)).
The computation in part b) shows that the function D preserves the linear combination 2ex + 3 cos(x). We’ll now show that D is a linear transformation of vector spaces, meaning it preserves all such linear combinations. (c) Let f and g be functions in C ∞ (R). Which rule for derivatives allows you to compute D(f + g)? Find a formula for D(f + g) in terms of D(f ) and D(g).
(d) Let f be a function in C ∞ (R), and let c ∈ R be a scalar. Which rule for derivatives allows you to compute D(cf )? Find a formula for D(cf ) in terms of c and D(f ).
Compare the formulas above to the rules for a linear transformation: you’ve just proven that the function D : C ∞ (R) → C ∞ (R) is a linear transformation from the vector space C ∞ (R) to itself! (e) If we compose two linear transformations, we get another linear transformation (we saw this for linear transformations T : Rn → Rm and S : Rm → Rp , but it’s true in general for the same reason). Example: D ◦ D : C ∞ (R) → C ∞ (R) is a linear transformation. Sometimes we write D2 instead of D ◦ D. Note that D2 sends each function f to its second derivative.
Calculate D2 (2ex + 3 cos(x)), D2 (ex ) and D2 (cos(x)).
Compare your answers above to see that
D2 (2ex + 3 cos(x)) = 2D2 (ex ) + 3D2 (cos x). This example shows that D2 preserves this particular linear combination; the fact that D2 is a linear transformation means that it actually preserves all such linear combinations.
Problem 2: Linear Independence in C ∞ (R):
In this problem, we’ll examine whether or not certain collections of functions are linearly inde- pendent in the vector space C ∞ (R). Question: Are ex , cos(x), and x3 + x linearly independent? (a) What would it mean if these functions were linearly dependent? Write an equation expressing your answer (your equation should involve three unknown scalars, c1 , c1 , and c3 , say). (b) Now, consider what happens if we apply the linear transformation D (differentiation) to the equation from part (a). Apply D to this equation to obtain an equation relating the derivatives of d x our functions (it’s better not to compute those derivatives quite yet; just write dx e or D(ex )).
Apply D again to the equation you just obtained to get an equation relating the second derivatives.
If ex , cos(x), and x3 + x are linearly dependent, what do the previous two equations tell you about the columns of the matrix below? (This matrix is called the Wronskian.) ex cos(x) x3 + x W (x) = D(ex ) D cos(x) D(x3 + x) D2 (ex ) D2 cos(x) D2 (x3 + x) ex cos(x) x3 + x (c) Compute the matrix W (x) = D(ex ) D cos(x) D(x3 + x) using the derivatives you D2 (ex ) D2 cos(x) D2 (x3 + x) calculated in 1a), and plug in x = 0 to get a matrix W (0).
(d) Explain the following statement: If ex , cos(x), and x3 +x were linearly dependent, then the columns ~a1 , ~a2 , ~a3 of the matrix W (0) that you computed above would also be linearly dependent (in other words, W (0) would fail to be invertible).
(e) Is W (0) invertible? Why or why not? What does this tell you about ex , cos(x), and x3 + x? Problem 3: Use an argument similar to that above to show that the functions ex , e−x , and e2x are linearly independent.
Problem 4: In the above discussion, we’ve seen that the Wronskian can be used to prove that certain collections of functions are linearly independent. In this problem, we’ll consider whether or not the Wronskian can be used to prove that collections of functions are dependent.
(a) Compute W (0) for the functions sin(x), sin(2x), cos(x). Is W (0) invertible? What do you think this says about linear independence or dependence of these functions?
(b) Compute W (π/4) for the same functions (sin(x), sin(2x), and cos(x)). Is W (π/4) invertible? What can you conclude about independence or dependence of these functions?
