optimization is a decision making method:

1. identify a performance measure and a space of possible strategies to try
2. run a bunch of simulations given a particular strategy, and measuring the performance
3. try strategies with the goal of maximizing the performance measured

Importantly: model is not used to guide the search, it is only used to run simulations to evaluate performance.

Disadvantage (or advantage)

does not take a advantage of the structure of the problem

AA222/CS361

Applications

Example Objectives

• efficiency
• safety
• accuracy

Example Constraints

• cost
• weight
• (structural) integrity

Why this is hard?

• high dimensional search spaces
• multiple competing objectives
• model uncertainty

Queen Dido’s Problem

• you are granted a plot of land
• how do we surround it with the smallest amount of string?

Design Workflow

“Optimization: how do we ‘change design’ in response to evaluation”

Logistics

aa222.stanford.edu

Overview

Lectures

Derivatives, Bracketing, Descent, and Approximations

Topics: Derivatives, Bracking, Descent, and Approximation

In the event your domain knowledge can help you make decisions about how spark load-balances or stripes data across worker nodes.

Persistence

“you should store this data in faster/slower memory”

MEMORY_ONLY, MEMORY_ONLY_SER, MEMORY_AND_DISK, MEMORY_AND_DISK_SER, DISK_ONLY

rdd.persist(StorageLevel.MEMORY_AND_DISK)
# ... do work ...
rdd.unpersist()

Parallel Programming

options are derivatives which gives you the permission to make a transaction at a particular date.

There are two main types of options:

• call: gives permission to buy a security on or before the “exercise” date
• puts: gives permission to sell a security on or before the “exercise” date

For this article, we will define \(S_{t}\) to be the stock price at the time \(t\), \(K\) as the option’s strike price, \(C_{t}\) to be the price of the “call” option, and \(P_{t}\) to be the price of the “put” option at strike price \(K\); lastly \(T\) we define as the maturity date.

an Option (MDP) represents a high level collection of actions. Big Picture: abstract away your big policy into \(n\) small policies, and value-iterate over expected values of the big policies.

Markov Option

A Markov Option is given by a triple \((I, \pi, \beta)\)

• \(I \subset S\), the states from which the option maybe started
• \(S \times A\), the MDP during that option
• \(\beta(s)\), the probability of the option terminating at state \(s\)

one-step options

You can develop one-shot options, which terminates immediate after one action with underlying probability