ALAN: Alan Hoffman
IRV: Irvin Lustig
IRV
Were you using computers for data collection to solve these problems?
ALAN
We certainly were. We were using SEAC, a computer now long dead. It had 1024 memory - that's it. It used mercury tubes and cathode ray tubes, and it wasn't very fast.
Now, the way we did the simplex method was that we had two tapes and you would read in the last iteration from the tape into the computer. Then you would do your transformation and it would read out on the other tape. The tapes were in bays behind glass and they would cascade down from the machine. You could watch it for hours. You could also see the iterations take place by watching the flashing red dots and so forth, and you'd know when the cycle of calculations was reaching a certain stage. That was loads of fun.
IRV
So, you were working with tapes at that time. Was it cumbersome?
ALAN
Let me tell you an interesting anecdote. Another group, not the optimizing group, was doing calculations for Los Alamos. Every night the person doing the calculations would take off the input and output Teletype tapes, roll them up, and mail them to Los Alamos. One night the programming team discovered they were out of tape. They didn't want to lose the night's work so they put a roll of toilet paper in the machine, printed out the calculations, and mailed it to Los Alamos the next day. I think the man who thought of that should've gotten a medal.
IRV
Was there some connection between the development of computers at that time and what was happening with linear programming?
ALAN
Yes. I alluded to the fact that the financing of SEAC was largely done with air force money because they had it. At that time, the air force was doing the best work in planning anywhere in the world. It is absolutely a thrill to realize retrospectively how great it was. I confess that at that time I didn't realize it. I just thought it was fun. But it really was an astounding achievement.
|