The values of f(t) at the odd integers are obtained directly from the first line of the definition.
f(2n−1)=(−1)n+12⇒f(1)=2,f(3)=−2,f(5)=2,f(7)=−2,…
Between two successive odd integers 2n−1 and 2n+1 the function is a straight-line joining these end-values, i.e. it is linear on every interval [2n−1,2n+1].
The integral required is g(x)=∫1xf(t)dt,x>1.
Because g′(x)=f(x), the right-hand derivative of g at x=1 equals f(1):
β=x→1+limx−1g(x)=g′(1+)=f(1)=2.(1)
Next we look for the zeros of g(x) in (1,8].
Observe first that on each interval [2n−1,2n+1] the end-values of f are equal in magnitude and opposite in sign, so the average value of f over the full length 2 is zero. Hence
∫2n−12n+1f(t)dt=0⟹g(2n+1)=g(2n−1).
Starting from g(1)=0, this gives successively
g(3)=0,g(5)=0,g(7)=0,…
Thus x=3,5,7 are certainly zeros lying in (1,8].
To see whether any additional zeros occur inside the open sub-intervals, evaluate g(x) on each of them.
Case 1: x∈[1,3]
The straight-line expression for f here is f(t)=2−2(t−1).
Integrating,
g(x)=∫1x[2−2(t−1)]dt=(2x−(x−1)2)−2=−x2+4x−3.
This quadratic satisfies g(1)=g(3)=0 and g(2)=1>0,
so no other root occurs in (1,3).
Case 2: x∈[3,5]
Here f(t)=−2+2(t−3). Put y=x−3(0≤y≤2). Then
g(x)=∫3x[−2+2(t−3)]dt=−2y+y2=y(y−2).
Hence g(3)=g(5)=0 and g(x)<0 for x∈(3,5).
Again, no extra root.
Case 3: x∈[5,7]
With u=x−5(0≤u≤2) one obtains
g(x)=2u−u2=u(2−u)>0 in (5,7), and only the end-points give zeros.
Case 4: x∈[7,8]
Let v=x−7. The integral gives
g(x)=−2v+v2=v(v−2)<0 for v∈(0,1], so no zero arises in (7,8] except the already counted one at x=7.
Collecting all solutions in (1,8]:
x=3,5,7⇒α=3.(2)
Finally, from \eqref1 and \eqref2
α+β=3+2=5.
Hence the required value equals 5.