ConvexOptimizationII-Lecture16

ConvexOptimizationII-Lecture16 -...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
ConvexOptimizationII-Lecture16 Instructor (Stephen Boyd) :That’s okay though. I see it puts a nice emphasis on, I mean, it’s probably okay to say here’s our problem, actually, it’s an approximation. Here’s the relaxation. Then when we run the relaxation, of course, on the real thing you get one thing, but then – actually, when you run it on the true system the relaxation thing does better than the other one. That’s not uncommon. Student: Yeah. Instructor (Stephen Boyd) :Perfectly okay. Student: Yeah. Student: It’s Kreloff. Instructor (Stephen Boyd) :It’s what? Student: Kreloff. Instructor (Stephen Boyd) :Kreloff. I just spent five days with a bunch of Russians. I guess I’ll have to remember that for next year. The left monitor’s going to be right to you that I’m on. Does that mean I’m on? Something about this room, I don’t know. At least I haven’t said anything too horrible while being taped or whatever. Okay. Today we’re jumping around a different order of topics, but this is maybe, I think, the next topic that some people are working on, it’s obviously too late for them for their projects, but we can at least cover the material and for people who are doing this at least it’ll make a lot of sense. For other people, it’s actually very, very good stuff to know about. It’s widely, widely used. So it’s called Model Predictive Control. In fact, I’ve been reading a lot about it the last couple of days. To sit through very long airplane flights, read a couple more books on it. It has got tons of different names, all different. Basically all the different areas doing this don’t know about the others. Often the notation and stuff is absolutely horrible. So, anyway, here’s the basic idea. We’ll get this in a minute. It’s basically, I mean, there’s a bigger area. Model Predictive Control is the name that comes from a relatively small community. In fact, I read one book and heard, I guess from some sort of member of this community, as referred to as the control community. That was as opposed to some others. Other names would be used. One that is a very good name in a sense is embedded optimization, real time optimization. Actually, the way that fits in the future, I think, is very, very interesting. A lot of people have their view of optimization locked into a model that’s from 1969. Either because they learned it in 1969 or they learned it from someone who learned it in 1969. They think of optimization as this big, complicated, very big thing that requires several Ph.D.’s, a bunch of super computers, and things like that. You really wouldn’t use
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
optimization except maybe to, I don’t know, design a nuclear weapon, design the newest airplane, schedule United Airlines for tomorrow, or who knows what else. But these are the kinds of things you – or price derivative in the basement of Goldman Sachs. These are the things people think of when you think of optimization. You all ready know that a lot of these things actually are extremely reliable. They’re reliable enough to just work always, period, end of story. They’re also very, very
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 04/09/2010 for the course EE 360B taught by Professor Stephenboyd during the Fall '09 term at Stanford.

Page1 / 22

ConvexOptimizationII-Lecture16 -...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online