180 likes | 370 Views
Local and Local-Global Approximations. Local algebraic approximations Variants on Taylor series Local-Global approximations Variants on “fudge factor”. Local algebraic approximations. Linear Taylor series Intervening variables Transformed approximation Most common: y i =1/x i.
E N D
Local and Local-Global Approximations • Local algebraic approximations • Variants on Taylor series • Local-Global approximations • Variants on “fudge factor”
Local algebraic approximations • Linear Taylor series • Intervening variables • Transformed approximation • Most common: yi=1/xi
Beam example • Tip displacement • Intervening variables yi=1/Ii
Reciprocal approximation • It is often useful to write the reciprocal approximation in terms of the original variables x instead of the reciprocals y
Conservative-convex approximation • At times we benefit from conservative approximations • All second derivatives of gCare non-negative • Convex linearization obtained by applying the approximation to both objective and constraints
Stress constraint on member C • Stress in terms of areas • Stress constraint • Using non-dimensional variables • What assumption on stress?
Problems local approximations • What are intervening variables? There are also cases when we use “intervening function” in order to improve the accuracy of a Taylor series approximation. Can you give an example? • What is conservative about the conservative approximation? Why is that a plus? Why is it useful that it is convex?
Local Approximations pros and cons • Derivative based local approximations have several advantages • Derivatives are often computationally inexpensive • Derivatives are needed anyhow for optimization algorithms • These approximations allow rigorous convergence proofs • There are some disadvantages too • They can have very small region of acceptable accuracy • They do not work well with noisy functions
Global approximations • Can be based on more approximate mathematical model • Can be based on same mathematical model with coarser discretization • Can be based on fitting a meta-model (surrogate, response surface) to a number of simulations • Pro and cons complement those of local approximations: Wider range, noise tolerance, but more expensive, and less amenable to math proofs
Combining local and global approximations • Can use derivatives to combine the two models • The combined approximation matches the value and slope at x0.
Example • Approximating the sine function as a quadratic polynomial
Problems local-global • For cantilever beam (Ex. 6.1.1) use a global approximation of a uniform beam with average moment of inertia. The exact solution can be simplified to • Given the function y=sinx • What is the linear approximation about x=0? • What is the reciprocal approximation about x=pi/2? • What is the global-local approximation about x=0, given the global approximation y=2x/pi?