1: \begin{abstract}

2: %Numerical optimization is indispensable

3: %in many disciplines of engineering.

4: %In order to be applied effectively, 

5: %it is necessary to determine

6: %the rate at which a given optimization

7: %algorithm will converge.

8: %Conventional approaches to derive 

9: %bounds on the convergence rate are

10: %non-systematic, requiring theorists

11: %to analyze optimization algorithms

12: %on a case by case basis. Approaches

13: %from control theory may be applied to determine the

14: %convergence rate of particular classes of

15: %algorithms in a systematic manner by solving

16: %a series of semidefinite programs. While

17: %this has proven effective, there is no guarantee

18: %that the produced bounds on convergence rate are

19: %tight. The progress to date has shown that

20: %one of the most prominent approaches for the analysis

21: %of optimization algorithms following the general control theoretic

22: %strategy mentioned previously fails to produce tight bounds. This was done

23: %by developing an approach

24: %that achieves better bounds on the convergence rate. The goal

25: %for the remainder of the project will be to attempt to 

26: %use this method to provide tight bounds by automatically

27: %generating functions that attain the worst case convergence rate.

28: %% The goal of this project is

29: %% to explore modifications to the control theoretic

30: %% approaches to convergence rate

31: %% analysis of first-order optimization

32: %% algorithms on smooth, strongly convex 

33: %% functions in order to produce

34: %% tighter bounds on the rate of convergence. 

35: %\end{abstract}

36: