HW12 prob. 8.14 (shouldn't ramp slope of PWM be negative?)

In HW12 prob. 8.14, shouldn't the ramp slope of PWM be negative, so that when vg(t) is high, the duty cycle is low? But then if the ramp slope is negative, will it not make the loop gain negative...but don't we need the loop gain to be positive for negative feedback (because va=ig(t) is already being subtracted from the vref1)?
Any comments would be appreciated.
Nitish