During takeoff, or while performing a high alpha/low speed pass showing off the aircraft's unique high-alpha capabilities in an airobatics display. Closing throttles would have been little more than a way to select the area of impact. Considering how close to the central axis the engines are in twin fighter jets going full throttle was probably the least bad attempt to save the plane.
Or landing. A famous case is the death of Kara Hultgreen, who suffered a compressor stall and subsequent asymmetric thrust while trying to land a F-14 on an aircraft carrier.
> The radar intercept officer in the rear seat, Lt. Matthew Klemish, initiated ejection for himself and Hultgreen as soon as it was apparent the aircraft was becoming uncontrollable. First in the automated ejection sequence, Klemish survived. However, by the time Hultgreen's seat fired 0.4 seconds later, the plane had exceeded 90 degrees of roll, and she was ejected downward into the water, killing her instantly.
It seems that she was just 0.4 seconds too late, otherwise she could have survived like Klemish.
Well, the plane was. When either co-pilot or pilot ejects, the other is ejected as well shortly after (on pretty much all planes). The delay is likely there to make sure it goes smoothly.
Multi-engine aircraft have a known minimal control speed, below the speed you cannot maintain directional control in case of an engine failure with full power from the other engine.
The theory is that you can only be below that during the takeoff roll and in case of a problem you have to abort the takeoff (close both throttles) to maintain control.
I've never flown military jets, but I expect they have this same concept. Then going full afterburner while below minimum control speed is what put him in an uncontrollable roll.