The document proposes a new interactive multiobjective decision making technique called the Sequential Proxy Optimization Technique (SPOT). SPOT combines desirable features of existing methods like the Surrogate Worth Trade-off method and Multiattribute Utility Function method. It involves 3 steps: 1) Generate Pareto optimal solutions using €-constraint problems. 2) Obtain marginal rate of substitution values from the decision maker to determine direction of increasing utility. 3) Update local proxy preference functions, determine optimal step size, and maximize proxies to find preferred solutions while guaranteeing Pareto optimality at each iteration. A numerical example is provided to illustrate SPOT.
Aspect Extraction Performance With Common Pattern of Dependency Relation in ...Nurfadhlina Mohd Sharef
A. S., Shafie, Sharef, N. M., Murad, M. A. A., Azman, A., (2018), "Aspect Extraction Performance With Common Pattern of Dependency Relation in Multi Aspect Sentiment Analysis", 2018 Fourth International Conference on Information Retrieval and Knowledge Management (CAMP18), Kota Kinabalu, in press.
JOBS Act: TylerCap TopAUM on Regulation D BasicsJonathan Buffa
JOBS Act: TylerCap TopAUM on Regulation D Basics
JOBS Act, Hedge Fund, TylerCap.com, TopAUM.com, capital raising, investor relations, JOBS Act Rules, Hedge Fund Launch, on Regulation D Basics
Tyler Capital Group-Bio-Jonathan Buffa COOJonathan Buffa
Tyler Capital Group,LLC (TylerCap) is a leading global search firm for more than 30 of the largest 100 hedge funds1. The TylerCap search model is designed to produce significant p&l returns for the hedge fund elite by providing our clients with 24/7 execution and a team approach to all searches. Our track record, education, discipline, and infiltration capacity are the cornerstones enabling TylerCap to lead the search industry. Our dedication to market intelligence and targeted approach facilitates the delivery of the candidate pool’s top 10%.
coo,hedge fund,hedge fund executive search,jonathan buffa,tyler capital group, hedge fund jobs, hedge fund recruiter, capital raising, investor relations, topAUM.com, tylercap.com
Aspect Extraction Performance With Common Pattern of Dependency Relation in ...Nurfadhlina Mohd Sharef
A. S., Shafie, Sharef, N. M., Murad, M. A. A., Azman, A., (2018), "Aspect Extraction Performance With Common Pattern of Dependency Relation in Multi Aspect Sentiment Analysis", 2018 Fourth International Conference on Information Retrieval and Knowledge Management (CAMP18), Kota Kinabalu, in press.
JOBS Act: TylerCap TopAUM on Regulation D BasicsJonathan Buffa
JOBS Act: TylerCap TopAUM on Regulation D Basics
JOBS Act, Hedge Fund, TylerCap.com, TopAUM.com, capital raising, investor relations, JOBS Act Rules, Hedge Fund Launch, on Regulation D Basics
Tyler Capital Group-Bio-Jonathan Buffa COOJonathan Buffa
Tyler Capital Group,LLC (TylerCap) is a leading global search firm for more than 30 of the largest 100 hedge funds1. The TylerCap search model is designed to produce significant p&l returns for the hedge fund elite by providing our clients with 24/7 execution and a team approach to all searches. Our track record, education, discipline, and infiltration capacity are the cornerstones enabling TylerCap to lead the search industry. Our dedication to market intelligence and targeted approach facilitates the delivery of the candidate pool’s top 10%.
coo,hedge fund,hedge fund executive search,jonathan buffa,tyler capital group, hedge fund jobs, hedge fund recruiter, capital raising, investor relations, topAUM.com, tylercap.com
Culture is the driver of sustainable performance. Management board culture is not as elusive as often thought. It can be made concrete by evaluating management board performance, not only based on figures and strategic memos, but also on key cultural characteristics. It is time to rethink the role of non-executives in the boardroom.
Climate Information and Early Warning Systems Communications ToolkitUNDP Climate
This toolkit provides National HydroMeteorological Services (NHMS), policy makers, and media and communications for development practitioners with the tools, resources and templates necessary to design and implement an integrated communications strategy. These communications strategies include the effective issuance and packaging of early warnings as well as the creation of supportive communications products and outreach efforts that will support the long-term sustainability of investments in the climate information and services sector.
Research and Development in Roof-Top Solar Potentiality Using LiDAR Technology
Mr. Radhey Shyam Meena
M.Tech Scholar (Power System)
Student Member -The Institute of Engineering & Technology (IET), UACEE
Dept. Of Electrical Engineering
Sri Balaji College Of Engineering & Technology Jaipur Rajasthan Technical University Kota
4th International conference on “Advance Trend in Engineering, Technology and Research (ICATETR-2015)”
Date: 19-20 June-2015
Venue: Bal Krishna Institute of Technology, Kota IPC-15, RIICO Institutional Area, Ranpur Kota (Rajasthan) (India)
The 20th International Congress of Nutrition (ICN) hosted by the International Union of Nutritional Science (IUNS) took place on the 15th-20th September 2013, Granada, Spain. WCRF International held a 2-hour symposium on the Continuous Update Project (CUP) entitled ‘Food, Nutrition, Physical Activity and Cancer – Keeping the Evidence Current: WCRF/AICR Continuous Update Project (CUP).’ It included four presentations exploring the latest updates from the CUP.
Made by Group 7 - Fatimah Azzahra 1711011099 and Olinda Calista Yetri 1711011119
Dr. Nova Mardiana, S.E., M.M.
Human Resource Management
International Class
Faculty of Economic and Business
University of Lampung
Main points:
1. Equity Theory and Fairness
2. Market Pressures
3. Developing Pay Levels
4. Conflicts Between Market Pay Surveys and Job Evaluation
5. Monitoring Compensation Costs
6. Globalization, Geographic Region, and Pay Structures
7. The Important of Process: Participation and Communication
8. Current Challenges
9. Can The U.S. Labor Force Compete?
10. Executive Pay
11. Government Regulation of Employee Compensation
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
This is a presentation about the introduction to system and analysis design. The topic talks about what are the system development life cycle and how it works. It also talks about the professions or the team conducting a study.
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...James Anderson
Effective Application Security in Software Delivery lifecycle using Deployment Firewall and DBOM
The modern software delivery process (or the CI/CD process) includes many tools, distributed teams, open-source code, and cloud platforms. Constant focus on speed to release software to market, along with the traditional slow and manual security checks has caused gaps in continuous security as an important piece in the software supply chain. Today organizations feel more susceptible to external and internal cyber threats due to the vast attack surface in their applications supply chain and the lack of end-to-end governance and risk management.
The software team must secure its software delivery process to avoid vulnerability and security breaches. This needs to be achieved with existing tool chains and without extensive rework of the delivery processes. This talk will present strategies and techniques for providing visibility into the true risk of the existing vulnerabilities, preventing the introduction of security issues in the software, resolving vulnerabilities in production environments quickly, and capturing the deployment bill of materials (DBOM).
Speakers:
Bob Boule
Robert Boule is a technology enthusiast with PASSION for technology and making things work along with a knack for helping others understand how things work. He comes with around 20 years of solution engineering experience in application security, software continuous delivery, and SaaS platforms. He is known for his dynamic presentations in CI/CD and application security integrated in software delivery lifecycle.
Gopinath Rebala
Gopinath Rebala is the CTO of OpsMx, where he has overall responsibility for the machine learning and data processing architectures for Secure Software Delivery. Gopi also has a strong connection with our customers, leading design and architecture for strategic implementations. Gopi is a frequent speaker and well-known leader in continuous delivery and integrating security into software delivery.
More Related Content
Similar to Tyler Capital Group: Sequential Proxy Optimization Tools
Culture is the driver of sustainable performance. Management board culture is not as elusive as often thought. It can be made concrete by evaluating management board performance, not only based on figures and strategic memos, but also on key cultural characteristics. It is time to rethink the role of non-executives in the boardroom.
Climate Information and Early Warning Systems Communications ToolkitUNDP Climate
This toolkit provides National HydroMeteorological Services (NHMS), policy makers, and media and communications for development practitioners with the tools, resources and templates necessary to design and implement an integrated communications strategy. These communications strategies include the effective issuance and packaging of early warnings as well as the creation of supportive communications products and outreach efforts that will support the long-term sustainability of investments in the climate information and services sector.
Research and Development in Roof-Top Solar Potentiality Using LiDAR Technology
Mr. Radhey Shyam Meena
M.Tech Scholar (Power System)
Student Member -The Institute of Engineering & Technology (IET), UACEE
Dept. Of Electrical Engineering
Sri Balaji College Of Engineering & Technology Jaipur Rajasthan Technical University Kota
4th International conference on “Advance Trend in Engineering, Technology and Research (ICATETR-2015)”
Date: 19-20 June-2015
Venue: Bal Krishna Institute of Technology, Kota IPC-15, RIICO Institutional Area, Ranpur Kota (Rajasthan) (India)
The 20th International Congress of Nutrition (ICN) hosted by the International Union of Nutritional Science (IUNS) took place on the 15th-20th September 2013, Granada, Spain. WCRF International held a 2-hour symposium on the Continuous Update Project (CUP) entitled ‘Food, Nutrition, Physical Activity and Cancer – Keeping the Evidence Current: WCRF/AICR Continuous Update Project (CUP).’ It included four presentations exploring the latest updates from the CUP.
Made by Group 7 - Fatimah Azzahra 1711011099 and Olinda Calista Yetri 1711011119
Dr. Nova Mardiana, S.E., M.M.
Human Resource Management
International Class
Faculty of Economic and Business
University of Lampung
Main points:
1. Equity Theory and Fairness
2. Market Pressures
3. Developing Pay Levels
4. Conflicts Between Market Pay Surveys and Job Evaluation
5. Monitoring Compensation Costs
6. Globalization, Geographic Region, and Pay Structures
7. The Important of Process: Participation and Communication
8. Current Challenges
9. Can The U.S. Labor Force Compete?
10. Executive Pay
11. Government Regulation of Employee Compensation
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
This is a presentation about the introduction to system and analysis design. The topic talks about what are the system development life cycle and how it works. It also talks about the professions or the team conducting a study.
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...James Anderson
Effective Application Security in Software Delivery lifecycle using Deployment Firewall and DBOM
The modern software delivery process (or the CI/CD process) includes many tools, distributed teams, open-source code, and cloud platforms. Constant focus on speed to release software to market, along with the traditional slow and manual security checks has caused gaps in continuous security as an important piece in the software supply chain. Today organizations feel more susceptible to external and internal cyber threats due to the vast attack surface in their applications supply chain and the lack of end-to-end governance and risk management.
The software team must secure its software delivery process to avoid vulnerability and security breaches. This needs to be achieved with existing tool chains and without extensive rework of the delivery processes. This talk will present strategies and techniques for providing visibility into the true risk of the existing vulnerabilities, preventing the introduction of security issues in the software, resolving vulnerabilities in production environments quickly, and capturing the deployment bill of materials (DBOM).
Speakers:
Bob Boule
Robert Boule is a technology enthusiast with PASSION for technology and making things work along with a knack for helping others understand how things work. He comes with around 20 years of solution engineering experience in application security, software continuous delivery, and SaaS platforms. He is known for his dynamic presentations in CI/CD and application security integrated in software delivery lifecycle.
Gopinath Rebala
Gopinath Rebala is the CTO of OpsMx, where he has overall responsibility for the machine learning and data processing architectures for Secure Software Delivery. Gopi also has a strong connection with our customers, leading design and architecture for strategic implementations. Gopi is a frequent speaker and well-known leader in continuous delivery and integrating security into software delivery.
Generating a custom Ruby SDK for your web service or Rails API using Smithyg2nightmarescribd
Have you ever wanted a Ruby client API to communicate with your web service? Smithy is a protocol-agnostic language for defining services and SDKs. Smithy Ruby is an implementation of Smithy that generates a Ruby SDK using a Smithy model. In this talk, we will explore Smithy and Smithy Ruby to learn how to generate custom feature-rich SDKs that can communicate with any web service, such as a Rails JSON API.
Elevating Tactical DDD Patterns Through Object CalisthenicsDorra BARTAGUIZ
After immersing yourself in the blue book and its red counterpart, attending DDD-focused conferences, and applying tactical patterns, you're left with a crucial question: How do I ensure my design is effective? Tactical patterns within Domain-Driven Design (DDD) serve as guiding principles for creating clear and manageable domain models. However, achieving success with these patterns requires additional guidance. Interestingly, we've observed that a set of constraints initially designed for training purposes remarkably aligns with effective pattern implementation, offering a more ‘mechanical’ approach. Let's explore together how Object Calisthenics can elevate the design of your tactical DDD patterns, offering concrete help for those venturing into DDD for the first time!
Epistemic Interaction - tuning interfaces to provide information for AI supportAlan Dix
Paper presented at SYNERGY workshop at AVI 2024, Genoa, Italy. 3rd June 2024
https://alandix.com/academic/papers/synergy2024-epistemic/
As machine learning integrates deeper into human-computer interactions, the concept of epistemic interaction emerges, aiming to refine these interactions to enhance system adaptability. This approach encourages minor, intentional adjustments in user behaviour to enrich the data available for system learning. This paper introduces epistemic interaction within the context of human-system communication, illustrating how deliberate interaction design can improve system understanding and adaptation. Through concrete examples, we demonstrate the potential of epistemic interaction to significantly advance human-computer interaction by leveraging intuitive human communication strategies to inform system design and functionality, offering a novel pathway for enriching user-system engagements.
Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024Tobias Schneck
As AI technology is pushing into IT I was wondering myself, as an “infrastructure container kubernetes guy”, how get this fancy AI technology get managed from an infrastructure operational view? Is it possible to apply our lovely cloud native principals as well? What benefit’s both technologies could bring to each other?
Let me take this questions and provide you a short journey through existing deployment models and use cases for AI software. On practical examples, we discuss what cloud/on-premise strategy we may need for applying it to our own infrastructure to get it to work from an enterprise perspective. I want to give an overview about infrastructure requirements and technologies, what could be beneficial or limiting your AI use cases in an enterprise environment. An interactive Demo will give you some insides, what approaches I got already working for real.
Transcript: Selling digital books in 2024: Insights from industry leaders - T...BookNet Canada
The publishing industry has been selling digital audiobooks and ebooks for over a decade and has found its groove. What’s changed? What has stayed the same? Where do we go from here? Join a group of leading sales peers from across the industry for a conversation about the lessons learned since the popularization of digital books, best practices, digital book supply chain management, and more.
Link to video recording: https://bnctechforum.ca/sessions/selling-digital-books-in-2024-insights-from-industry-leaders/
Presented by BookNet Canada on May 28, 2024, with support from the Department of Canadian Heritage.
Smart TV Buyer Insights Survey 2024 by 91mobiles.pdf91mobiles
91mobiles recently conducted a Smart TV Buyer Insights Survey in which we asked over 3,000 respondents about the TV they own, aspects they look at on a new TV, and their TV buying preferences.
Essentials of Automations: Optimizing FME Workflows with ParametersSafe Software
Are you looking to streamline your workflows and boost your projects’ efficiency? Do you find yourself searching for ways to add flexibility and control over your FME workflows? If so, you’re in the right place.
Join us for an insightful dive into the world of FME parameters, a critical element in optimizing workflow efficiency. This webinar marks the beginning of our three-part “Essentials of Automation” series. This first webinar is designed to equip you with the knowledge and skills to utilize parameters effectively: enhancing the flexibility, maintainability, and user control of your FME projects.
Here’s what you’ll gain:
- Essentials of FME Parameters: Understand the pivotal role of parameters, including Reader/Writer, Transformer, User, and FME Flow categories. Discover how they are the key to unlocking automation and optimization within your workflows.
- Practical Applications in FME Form: Delve into key user parameter types including choice, connections, and file URLs. Allow users to control how a workflow runs, making your workflows more reusable. Learn to import values and deliver the best user experience for your workflows while enhancing accuracy.
- Optimization Strategies in FME Flow: Explore the creation and strategic deployment of parameters in FME Flow, including the use of deployment and geometry parameters, to maximize workflow efficiency.
- Pro Tips for Success: Gain insights on parameterizing connections and leveraging new features like Conditional Visibility for clarity and simplicity.
We’ll wrap up with a glimpse into future webinars, followed by a Q&A session to address your specific questions surrounding this topic.
Don’t miss this opportunity to elevate your FME expertise and drive your projects to new heights of efficiency.
Securing your Kubernetes cluster_ a step-by-step guide to success !KatiaHIMEUR1
Today, after several years of existence, an extremely active community and an ultra-dynamic ecosystem, Kubernetes has established itself as the de facto standard in container orchestration. Thanks to a wide range of managed services, it has never been so easy to set up a ready-to-use Kubernetes cluster.
However, this ease of use means that the subject of security in Kubernetes is often left for later, or even neglected. This exposes companies to significant risks.
In this talk, I'll show you step-by-step how to secure your Kubernetes cluster for greater peace of mind and reliability.
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered QualityInflectra
In this insightful webinar, Inflectra explores how artificial intelligence (AI) is transforming software development and testing. Discover how AI-powered tools are revolutionizing every stage of the software development lifecycle (SDLC), from design and prototyping to testing, deployment, and monitoring.
Learn about:
• The Future of Testing: How AI is shifting testing towards verification, analysis, and higher-level skills, while reducing repetitive tasks.
• Test Automation: How AI-powered test case generation, optimization, and self-healing tests are making testing more efficient and effective.
• Visual Testing: Explore the emerging capabilities of AI in visual testing and how it's set to revolutionize UI verification.
• Inflectra's AI Solutions: See demonstrations of Inflectra's cutting-edge AI tools like the ChatGPT plugin and Azure Open AI platform, designed to streamline your testing process.
Whether you're a developer, tester, or QA professional, this webinar will give you valuable insights into how AI is shaping the future of software delivery.
The Art of the Pitch: WordPress Relationships and SalesLaura Byrne
Clients don’t know what they don’t know. What web solutions are right for them? How does WordPress come into the picture? How do you make sure you understand scope and timeline? What do you do if sometime changes?
All these questions and more will be explored as we talk about matching clients’ needs with what your agency offers without pulling teeth or pulling your hair out. Practical tips, and strategies for successful relationship building that leads to closing the deal.
Connector Corner: Automate dynamic content and events by pushing a buttonDianaGray10
Here is something new! In our next Connector Corner webinar, we will demonstrate how you can use a single workflow to:
Create a campaign using Mailchimp with merge tags/fields
Send an interactive Slack channel message (using buttons)
Have the message received by managers and peers along with a test email for review
But there’s more:
In a second workflow supporting the same use case, you’ll see:
Your campaign sent to target colleagues for approval
If the “Approve” button is clicked, a Jira/Zendesk ticket is created for the marketing design team
But—if the “Reject” button is pushed, colleagues will be alerted via Slack message
Join us to learn more about this new, human-in-the-loop capability, brought to you by Integration Service connectors.
And...
Speakers:
Akshay Agnihotri, Product Manager
Charlie Greenberg, Host
Connector Corner: Automate dynamic content and events by pushing a button
Tyler Capital Group: Sequential Proxy Optimization Tools
1. NOT FOR QUOTATION
WITHOUT PERMISSION
OF THE AUTHOR
INTERACTIVE MULTIOBJECTIVE DECISION
MAKING BY THE SEQUENTIAL PROXY
OPTIMIZATION TECHNIQUE: SPOT
Masatoshi Sakawa
April 1980
WP-80-66
Working Papers are interim reports on work of the
International Institute for Applied Systems Analysis
and have received only limited review. Views or
opinions expressed herein do not necessarily repre-
sent those of the Institute or of its National Member
Organizations.
INTERNATIONAL INSTITUTE FOR APPLIED SYSTEMS ANALYSIS
A-2361 Laxenburg, Austria
2. PREFACE
Methodologies for decision making with conflicting multiple objectives
has attracted increasing attention since the early period of IIASA activity.
In the System and Decision Sciences Area of IIASA, decision making processes
with conflicting objectives as well as multiobjective optimization are main
projects and many techniques have been developed. This paper intends to pro-
vide a modest approach to such a research direction for decision sciences.
The author is thankful to Professor A. Wierzbicki, Chairman of the System
and Decision Sciences Area, for providing him with the opportunity to visit
IIASA and to work on this project. The author expresses his gratitude to
Professor F. Seo for discussions and comments, and is indebted to Professor
Y. Sawaragi of Kyoto University for his constant encouragement. The numerical
results were obtained while the author was at the Systems Engineering Depart-
ment of Kobe University in Japan, and the author wishes to thank Mrs. T. Sasakura
and K. Tazumi for their cooperation in this study.
3. ABSTRACT
In this paper, we propose a new interactive multiobjective decision
making technique, which we call the Sequential Proxy Optimization Technique
( SPOT ), in order to overcome the drawbacks of the conventional multiobjec-
tive decision making methods. Our method combines the desirable features
of both the Surrogate Worth Trade-off ( SWT ) method and the Multiattribute
Utility Function ( MUF ) method. We can interactively derive the preferred
solution of the decision maker efficiently by assessing his marginal rate
of substitution and maximizing sequentially the local proxy preference func-
tion. A numerical example illustrates the feasibility and efficiency of
the proposed method.
4. 1. Introduction
The development of decision making methodologies under multiple conflicting
objectives has been one of the most active areas of research in recent years.
Several techniques have been developed ; among them two rival methods, namely,
the multiattribute utility function ( MUF ) method [ I ] and the surrogate worth
trade-off ( SWT ) method [2,31 use global and local utility ( preference )
modelling respectively.
The MUF method developed by Keeney et al., global utility function model-
Ling, uses two assumptions of preference independence and utility independence to
limit the utility function to specialized forms - additive or multiplicative.
Once the form is selected, a few assessments determine the free parameters.
These global functions are mathematically simple and convenient, but they have
disadvantages. Their assumptions are reasonable locally, but when assumed
globally, they are very restrictive and may force the decision maker ( DM )
to fit a function not truly representing his or her preferences.
The SWT method developed by Haimes et al., local utility function modelling,
provides an alternative approach that avoids restrictive assumptions. Instead
of specifying the utility function globally, their procedures construct a
sequence of local preference approximations of it.
The general philosophy taken in the interactive approach using the local
utility function modelling is that the multiobjective decision making process
should follow the following 3-step procedure.
Step 1. Generate Pareto optimal solutions
Step 2. Obtain meaningful information to interact with the DM
Step 3. Use information obtained in step 2 to interact with the DM and select
5. t h e f i n a l s o l u t i o n b a s e d on t h e DM'S preference response.
The i n t e r a c t i v e Frank-Wolfe ( IFW ) method d e v e l o p e d by G e o f f r i o n e t a l .
[ 4 ] , p u t s p e c i a l emphasis on s t e p s 2 and 3. I n s t e p 2, t h e DM i s simply
s u p p l i e d w i t h t h e c u r r e n t v a l u e s of t h e o b j e c t i v e f u n c t i o n s t o which t h e DM
r e s p o n d s by p r o v i d i n g t h e m a r g i n a l r a t e of s u b s t i t u t i o n ( M S ) v a l u e s between
R
two o b j e c t i v e s . T h i s i n f o r m a t i o n i s t h e n used t o modify t h e o b j e c t i v e f u n c t i o n
f o r g e n e r a t i n g a new p o i n t i n s t e p 1 of t h e n e x t i t e r a t i o n by a p p l y i n g Frank-
Wolfe a l g o r i t h m . U n f o r t u n a t e l y , t h i s method d o e s n o t g u a r a n t e e t h a t t h e
generated s o l u t i o n i n each i t e r a t i o n w i l l b e P a r e t o optimal.
The S T method u s e s t h e € - c o n s t r a i n t problem a s a means o f g e n e r a t i n g
W
Pareto optimal solutions. O b j e c t i v e t r a d e - o f f s , whose v a l u e s c a n b e e a s i l y
o b t a i n e d from t h e v a l u e s of some s t r i c t l y p o s i t i v e Lagrange m u l t i p l i e r s from
s t e p 1 a r e used a s t h e i n f o r m a t i o n c a r r i e r i n s t e p 2. And i n s t e p 3 , t h e
DM r e s p o n d s by e x p r e s s i n g h i s d e g r e e o f p r e f e r e n c e o v e r t h e p r e s c r i b e d t r a d e -
o f f s by a s s i g n i n g n u m e r i c a l v a l u e s t o e a c h s u r r o g a t e w o r t h f u n c t i o n . This
method g u a r a n t e e s t h e g e n e r a t e d s o l u t i o n i n e a c h i t e r a t i o n t o b e P a r e t o o p t i m a l
and t h e DM c a n s e l e c t h i s p r e f e r r e d s o l u t i o n from among P a r e t o o p t i m a l s o l u t i o n s .
However, t h e o r i g i n a l v e r s i o n o f t h e S T method i s n o n i n t e r a c t i v e and some
W
improvement, p a r t i c u l a r l y i n t h e way t h e i n f o r m a t i o n from t h e DM i s u t i l i z e d ,
must b e made.
R e c e n t l y , Chankong and Haimes [5,61 and Simizu e t a 1 . ,[ 7 1 independently
proposed a n i n t e r a c t i v e v e r s i o n of t h e S T method on t h e b a s i s of t h e S T
W W
and t h e IFW methods. T h e i r methods f o l l o w a l l t h e s t e p s of t h e S T method
W
up t o t h e p o i n t where a l l t h e s u r r o g a t e worth v a l u e s c o r r e s p o n d i n g t o t h e
P a r e t o opt2mal s o l u t i o n a r e o b t a i n e d from t h e DM. An i n t e r a c t i v e o n - l i n e
scheme was c o n s t r u c t e d i n such a way t h a t t h e v a l u e s o f e i t h e r t h e s u r r o g a t e
6. worth function or the MRS are used to determine the direction in which the
utility function, although unknown, increases most rapidly. In their method,
however, the DM must assess his preference at each trial solution in order to
determine the step size. Such a requirement is very difficult for the DM,
since he does not know the explicit form of his utility function.
On the other hand, in 1978, Oppenheimer proposed a proxy approach [8] to
multiobjective decision making. He introduced the local proxy preference
functions in the IFW method. In his procedure the local proxy preference
function is updated at each iteration by assessing a new MRS vector. Then
the proxy is maximized to find a better point. Unfortunately, this method,
like the IFW method, does not guarantee the generated solution in each itera-
tion to be Pareto optimal. Furthermore, the systematic procedure to maximize
the proxies is not mentioned, so it seems to be very difficult to do so in
practice.
In this paper, we propose a new interactive multiobjective decision making
technique, which we call the sequential proxy optimization technique ( SPOT )
incorporating the desirable features of the conventional multiobjective decision
making methods. In our interactive on-line scheme, after solving the €-constraint
problem, the values of the MRS assessed by the DM are used to determine the
direction in which the utility function increases most rapidly and the local
proxy preference function is updated .to determine the optimal step size and
Pareto optimality of the generated solution is guaranteed. A numerical example
illustrates the feasibility and efficiency of the proposed method.
7. 2. Multiobjective Decision Making Problem
2.1 Preliminaries
The multiobjective optimization problem ( MOP ) is represented as
MOP
min
X
subject to
XEX= (XI N
X E E , gi(x)zO, i=1,2 ,...,m l (2
where x is an N-dimensional vector of decision variables, f l , ...,fn are n district
objective functions of the decision vector x, gl, ...,gm are a set of inequality
constraints and X is the constrained set of feasible decisions.
Such a vector-valued index usually induces a partial ordering on the set
of alternatives, one cannot speak of optimal solutions. Fundamental to the
MOP is the Pareto optimal concept, also known as a noninferior solution.
Qualitatively, a Pareto optimal solution of the MOP is one where any improvement
of one objective function can be achieved only at the expense of another.
Mathematically, a formal definition of a Pareto optimal solution is given below:
Definition 1.
A decision x* is said to be a Pareto optimal solution to the MOP, if and
only if there does not exist another x so that fJ. ( y- ) f~(x*) , j=1,2,. .. ,n, with
J
.
strict inequality holding for at least one j.
Usually, Pareto optimal solutions consist of an infinite number of points,
and some kinds of subjective judgement should be added to the quantitative
analyses by the DM. The DM must select his preferred solution from among
Pareto optimal solutions.
Definition 2.
8. A preferred solution is a Pareto optimal solution which is chosen as the
final decision through the preference ordering relation given by the DM.
Thenwe can state the multiobjective decision making problem ( MDMP )
we wish to solve
MDMP
max U(fl(x),f2(~)y.--yfn(~))
X
P
subject to XEX (4)
P
where X is the set of Pareto optimal solutions of the MOP and U '
() is the
N
DM'S overall utility function defined on F)
!-
{f
(
x ( xEE and is assumed to exist
and is known only implicitly to the DM.
One way of obtaining Pareto optimal solutions to the MOP is to solve
€-constraint problem Pk(€ -k )
min fk(x)
subject to xEXn (E-~)
E-k Ek
where A
E - ~ ( ..
E ~ , . ,Ek-l~Ek+l Y . •
n1
YE
(E-~) {x 1 f.(XI (cj , j=l,. . . ,n, jfk 1
J
Ek&{ E-k I (E-~) f$ 1
The following theorem is well known [5,61.
Theorem 1.
A unique solution Pk (E -k) , for any l ~ k ~ n , a Pareto optimal solution of
is
the MOP. Conversely, any Pareto optimal solution of the MOP solves ( not
necessarily uniquely ) Pk(~-k) for some E
-k
E E and for all k=l,
k
...,n.
9. 2.2 Multiobjective Decision Making in Objective Space
Let us assume that x*(E-~), an optimal solution to the P (E ) , be unique
k -k
for the given E And let AEk be a set of such
E - ~ that all the €-constraint
-k'
(9) is active, that is
-k)) = L j , j=l,. .. ,n, jZk 1 (11)
Define the following active €-constraint problem APk (E -k)'
min fk(x) (12)
X
subject to XEX~X~() E
-k
€-k AEk
If the Kuhn-Tucker condition for problem APk (E -k) is satisfied, the
Lagrange multiplier Akj(~-k) associated with the jth active constraint can be
represented as follows:
A
kj
= - { afk(~-~) / { af. ( E ) 1
1 J -j
j=l,. ..,n, jZk
The optimal values of the original decision variables, X(
*E ) , and the
-k
corresponding values of the primal objective, f [x*(E ) I , determines the
k -k
trade-off surface by the repeated solution of AP (& -k)y with various values
of the secondary objectives, c j=l,... ,n, j#k.
j'
Substituting the optimal solutions of the AP (E
k - ) x*(E_~), given desired
k'
levels of the secondary objectives, E
j'
...,n, jZk, the MODM
j=l, can be restated
as follows:
max .. .
U ( E ~ , , ~ ~ - ~ , f ~ [ ~ * ( ~ k + l yl
y -~)
E
-k
No constraints are involved in equation (16), since all constraints were
considered in the solution of the active epsilon-constraint problem AP (E ) .
k -k
10. Restricting the MDMP to the Pareto optimal region simplifies the approach
considerably. The decision variables are now the desired levels of the
objectives,
j'
j=l,E ...,
n, j#k, rather than the original decision variables, x.
n-1
The optimization is carried on in the objective function space, E , not in
the decision variable space, E ~ . This is of course, a clear advantage since
in most realistic problems, N >>n.
Throughout this paper we make the following.
Assumption 1 : U : F + R exists and is known only implicitly to the DM.
Moreover, it is assumed to be concave, a strictly decreasing
and continuously differentiable function on F.
Assumption 2 : All fi, i=l, ...,n
and all g j=l,
j'
...,
m are convex and twice
continuously differentiable in their respective domains and
constraint set X is compact.
Assumption 3 : For every feasible E E A E the solution to AP (E ) exists
~
-k k -k
and is finite.
Under Assumptions 1-3, the following theorem holds.
Theorem 2.
Under Assumptions 1-3, the utility function U ( rl, . . . , E ~ - ~ , ~ ~ [ x * ( E - ~ ) ] , E ~ + ~ ,
...,En ) is concave with respect to E -k E A E ~ .
Proof
By the convexity of fi and X, the setE is convex and the function
k
fk[~*(~-k)] is convex with respect to E Furthermore, by the monotonicity
-k '
and concavity of U with respect to fk, the following relations hold for any
-
E E E AE and OZe_<l.
-kY -k k
( (l-e)SkYfk( (l-e)<k) )
11. >
- u ( er-k + (l-e)Bk , efk(~-k) + (l-e)fk(;-k)
- -
=u ( e('-k,fk(8-k) ) + (-)
I' ( 'kf(-)
-,k'k )
>
- eu ( E-k,fk(~-k) ) + (I-') (-,kFk
'kf(-) )
where
n
fk(~-k) f k k * ( E - ~ )I
u (E-~,~~(E-~)E~,
) ( , € k-1' fk [x*('-~)] ,Ek+l,...,En )
Thus U ( E - ~ , ~ ~ ( E i s concave w i t h r e s p e c t t o E - ~ E A E ~ .
-k ) )
12. 3. Marginal Rate of Substitution
Now, before formulating the gradient, aU(.)/a~, of utility function U,
we introduce the concept of the marginal rates of substitution ( MRS ) of the
DM.
Definition 3.
At any f, the amount of fi that the DM is willing to sacrifice to acquire
an additional unit of f is called the MRS. Mathematically, the MRS is the
j
negative slope of the indifference curve at f:
where each indifference curve is a locus of points among which the DM is
indifferent.
The decision analyst assesses MRS by presenting the following prospects
to the DM
for a small fixed Af small enough so the indifference curve is approximately
j'
linear but large enough so the increment is meaningful, the analyst varies Af
i
until the DM is indifferent between f and f'. At this level, m
ij (f)rAfi/Af j '
in Fig.l,df.=-Afi and df = Af
I
. j j
.
13. curve
Figure 1. Assessing the Marginal Rate of Substitution.
4. Gradient Method in Objective Space
Now, we can formulate the gradient aU(.)/a~. of utility function U'.
()
J
Applying the chain rule
Using the relations (15) and (171, we have the following
From the strict monotonicity of U with respect to fk, k=l, ...,n, aU(-)/afk
is always negative. Therefore - (mkj - Akj ) . . . ,n, j#k
( j=l, ) decide a
direction improving the values of U(-) at a current point.
Under the assumptions 1-3, the optimality conditions for a maximization
point are
E - ~ aU(.)/a~-~ = 0, that is
q. A, j=l,. . . ,n, j#k (20)
J= j
This is a well known result that at the optimum the MRS of the DM must
be equal to the trade-off rate.
14. I f t h e o p t i m a l i t y c o n d i t i o n (20) i s n o t s a t i s f i e d a t t h e Rth i t e r a t i o n ,
R R
t h e o p t i m a l d i r e c t i o n of s e a r c h s and t h e corresponding d i r e c t i o n of bf a r e
j k
given by :
Then, we must d e t e r m i n e t h e o p t i m a l s t e p s i z e a which maximizes
U (
R R
+ a b ~ f + abf )
-kY k k
R
.. R 11 R
~ ~. ..
R
a l o n g t h e d i r e c t i o n bf' = ( A E ~ , . , b ~ ~ - ~ , b f ~ , b, B E n ) + ~ ,
To s o l v e t h i s l i n e a r s e a r c h problem, t h e f o l l o w i n g two problems a r i s e .
Problem 1.
R R
The DM must a s s e s s h i s p r e f e r e n c e a t each t r i a l s o l u t i o n (E' + a b c -k' f k + a b f k )
-k R
f o r s e v e r a l v a l u e s of a , i n o r d e r t o d e t e r m i n e t h e b e s t s t e p s i z e . Such
requirement i s v e r y d i f f i c u l t f o r t h e DM, s i n c e he does n o t know t h e e x p l i c i t
form of h i s u t i l i t y f u n c t i o n .
Problem 2.
Even i f i t i s p o s s i b l e f o r t h e DM t o a s s e s s t h e u t i l i t y v a l u e , t h e r e remains
R R
a problem. I n F i g . 2, new t r i a l p o i n t f + a b f R , where bf is a direction
v e c t o r , i s n o t a P a r e t o o p t i m a l s o l u t i o n f o r any a s a t i s f y i n g O5a5a2 ; t h a t i s ,
--
f o r a=a t h e r e is a Pareto optimal point P which i s obviously b e t t e r than T
1' 1 1'
Furthermore, f o r any a s a t i s f y i n g a > a t h e t r i a l p o i n t , l i k e p o i n t T3,becomes
2'
i n f e a s i b l e and t h e assessment a t such a p o i n t i s meaningless.
R R R
I n o r d e r t o r e s o l v e problem 2 , we Adopt (E' +an& f ( ~ - ~ + a A t - ~ ) )
-k -kY k
R 11
a s a t r i a l p o i n t i n t h e p r o c e s s of l i n e a r s e a r c h i n s t e a d of (E' +%A€' - k Yf k + a A f k ) .
-k
15. ( Trade-off curve
Figure 2. Gradient Method in Objective Space.
Our trial point becomes a Pareto optimal solution by solving the active epsilon
constraint problem AP k (E -k)* Although it is necessary to solve APk(~ ) for
-k
several values of a, the generated solution in each iteration becomes a Pareto
optimal solution and the DM can select his preferred solution from among Pareto
optimal solutions.
Concerning the problem 1, it is necessary to.construct some kind of
utility ( preference ) function, so we introduce local proxy preference
functions like Oppenheimerismethod [ 8 ] as explained in the following.
16. 5. Local Proxy Preference Functions
Using the deterministic additive independence condition P (f) = 1 p . (fi)
1
of Keeney et al. [I], together with assumptions about a marginal rate of
substitution variation, Barrager and Keelin [g], [lo] derive the following
three global utility functions:
(1) sum-of-exponentials
1f [-amij (f)/afjl /mij(f) = uj then Pf
() = - 1 aiexp (-uifi) (23)
It implies that if the DM is indifferent between any f1 and f2, then he
1 2
is also indifferent between f +A and f +A where A = (A,A, ...,A).
(2) sum-of-powers (a.#o)
J a
i
If [-amij(f)/afj]/mij(f) = (l+a.)/f. thenP(f)=-1 aifi (24)
J J
It implies that as the DM accumulates more of each attribute, he becomes
less sensitive to substitutions among them.
(3) sum-of-logarithms
If [-amij(f)/afjl /mij(f) = 1/(M-f . ) then P(f) = lail,(~-fi) (25)
J
where M is a sufficiently large positive number.
This utility function can be viewed as the additive form of the Cobb-
Douglass function. It implies that if the DM is indifferent between any
M-f 1 and M-f 2 , then he is also indifferent between b(M-f 1) and b(M-f 2) for
any positive constant b.
Although these utility functions are very restrictive globally, they
are reasonable when assumed locally. We use these utility functions as
local proxy preference functions to determine the optimal step size because
they seem to be a very good model locally.
17. In the following, we deal with the sum-of-exponentials as an example of
a local proxy preference function, but for other types of proxy functions
similar discussions can be made.
For the sum-of-exponentials function, the constant al can arbitrarily be set
equal to one in P ( f ) = - 1 aiexp(-w 1f1) .
. . The remaining parameters, a2,. . . ,an,
w~,w~,...,w can be calculated from MRS assessment. At any f, there are n-1
n
MRS at each of two points plus a single MRS at third point are required to fit
the 2n-1 parameters.
The numerical MRS actually assessed relate to the sum-of-exponential
parameters ai and wi in the following way;
f )I
mkj (f) = [a~(f)/af.]/[ap(f)/af~]
J
= [w.a.exp(-w.f .)l(ukakexp(-w
J J J J k k (26)
.
j=l,.. ,n, j#k
By taking the logarithm and solving a set of linear equations, the proxy
parameters a and w are uniquely determined from the 20-1 assessment of r k ( )
njf.
If the equations are linearly dependent, an additional assessment at the third
point is required.
18. 6. Consistency of Marginal Rate of Substitution Assessment
We have assumed up to this point that at each iteration the DM provides
MRS consistent with a continuously differentiable deterministic utility func-
tion, But, it is a question whether the DM can respond precise and consistent
values of MRS through the whole searching process. In this section, we relax
the assumption and examine techniques for checking MRS consistency following
Oppenheimer [ 8 ] .
Two types of consistency tests are employed, the first testing MRS consis-
tency at a single point, and the second testing consistency at successive points.
The single point test requires a second set of assessments at each point
and checks whether the MRS of the DM satisfies the chain rule, i.e.
mkj =mkimij
i,j1 , . .n , i ,k ,k j . Since only n-1 unique MRS among the objectives
exist at any point, the second set can be used to measure the discrepancy E:
E = [ (nfk/nfj) - (f/f)
A,~, (Afi/Af .) ] / (Afk/Af .) (2) (27)
J J
Certainly we would not expect exact agreement. Instead, we set a reasonable
tolerance level; if the discrepancy exceeds the tolerance, the analyst should
explain the inconsistency to the DM and reassess the MRS until the discrepancy
is resolved.
The second test checks for decreasing marginal rates of substitution of
the proxy. In assumption 1, we assumed the utility function U ( f ) is
strictly concave, satisfying a strictly decreasing marginal rate of substitution.
So, we must check the concavity and monotonicity of the proxy P f .
() The
necessary and sufficient condit.ion for the three types of proxy P(f) to be
strictly decreasing and concave can be shown using the parameter values con-
dition.
19. The following theorem can be easily proven by constructing the Hessian
matrix of P f .
()
Theorem 3.
(1) The sum-of-exponentials proxy P(f) is strictly decreasing and concave if
and only if all the parameters ai and wi are strictly positive, i.e.,
a. > 0 and w. > 0,
1 1
i=l,...,n (28)
(2) The sum-of-powers proxy P(f) is strictly decreasing and concave if and
only if
a. > 0;
1
ai > 1 i=l,. .. ,n (29)
(3) The sum-of-logarithms proxy P(f) is strictly decreasing and concave if
and only if
a. > 0
1
.
i=l,.. ,n (30)
20. 7. Algorithm of the SPOT
Following the above discussions, we can now construct the algorithm
of the sequential proxy optimization technique ( SPOT ) in order to obtain
the preferred solution of the DM for the 'MDMP.
Step 1 Choose initial point E A E and set R=1.
-k k
R R R
Step 2 Set E - ~= E solve an active €-constraint problem AP (E ) for E
-k ' k -k -k
and obtain a Pareto optimal solution x*(rR ) , a Pareto optimal value
-k
R R R R R
f = (E-~,~~[x*(E-~)]and corresponding Lagrange multiplier A
)
kj ( j=
.
I,.. ,n, j#k 1.
Step 3 Assess the MRS of the DM at fey where Af
j
..
( j=l,. ,n, j#k ) must be
fixed small enough so the indifference curve is approximately linear
but large enough so the increment is meaningful.
Step 4 For MRS at fey evaluate discrepancy E. If E < go to step 5, where
the tolerance is a prescribed sufficiency small positive number.
If E exceeds the tolerance, the DM reassess the MRS until the tolerance
condition is satisfied.
Step 5 R
If I mkj - A
R
kj
..
1 < 62 for j=l,. ,n, j#k, stop, where the tolerance 62
is a prescribed sufficiency small positive number. Then a Pareto
R R R
optimal solution (E-~,~~[x*(E-~)]is the preferred solution of the
)
DM. Otherwise, determine the direction vector AE
R by
-k
R R R R
s
j
= - (mkj-Ak.) = AE
J j
( j=l,.. . ,n, j#k )
21. 2 R
Step 6 O b t a i n two P a r e t o o p t i m a l p o i n t s I f R and f i n t h e neighbourhood
of f R and a s s e s s n-1 M S nfj
R a t a p o i n t I f R p l u s a s i n g l e MRS a t a
t h i r d point 2fR. I f t h e c o n s i s t e n c y check a t s t e p 4 i s p a s s e d ,
s e l e c t t h e form of t h e proxy f u n c t i o n t h a t w i l l b e used a t each
i t e r a t i o n by t h e measure about M S v a r i a t i o n .
R I f t h e parameter v a l u e
c o n d i t i o n s of theorem 3 a r e passed go t o t h e n e x t s t e p . Otherwise,
t h e DM r e a s s e s s e s t h e M S u n t i l t h e parameter v a l u e c o n d i t i o n s a r e
R
satisfied.
Step 7 Determine t h e s t e p s i z e u which maximizes t h e proxy p r e f e r e n c e
L R R R R
function P ( E + a A ~ - ~ , f ~ [ x * ( ~ - ~ + a A ) - ~ )a) a s follows.
~ P( l
-k
Change t h e s t e p s i z e , o b t a i n c o r r e s p o n d i n g P a r e t o o p t i m a l v a l u e s
and s e a r c h f o r t h r e e a v a l u e s a A , a B and a which s a t i s f y
C
a <a <a
A B C
p (aA) 'p (aB) ' P(ac)
Then a l o c a l maximum of P ( a ) i s i n t h e neighbourhood of a = a
B'
R
I f u(fa+') > f ) where
R R R R R R R
f R + l = ( E +a A E - ~ , ~ ~ [ x * ( E - ~ + ] ), A e t - R )= R + l and r e t u r n t o
~ ~ sE ~
-k B
s t e p 2. Otherwise r e d u c e u t o be ?, 1
B
1
z...
u n t i l improvement i s
achieved.
Remark. Under t h e assumption of i d e a l DM, t h e proposed SPOT a l g o r i t h m i s
n o t h i n g b e s i d e s a f e a s i b l e d i r e c t i o n method t o s o l v e MDMP. Thus, t h e conver-
gence o f t h e SPOT c a n b e demonstrated by t h e convergence o f t h e modified
f e a s i b l e d i r e c t i o n method.
22. 8. An Illustrative Example
We now demonstrate the interaction processes of the SPOT by means of an
illustrative example which is designed to test the method under the assumption
of an ideal DM.
Consider the following multiobjective decision making problem.
min f (x) = (fl (x) ,f2 (XI,f3 (4
X
subject to
2 2
XEX= {X I x1+x2+x: 5 100, O~x1.x2,x3=lo 1
<
where
2 2 2
fl(x) = 565 (x1 + ~ ~ + 1 0 x ~ + ~ ~ - 1 2 0 x ~ + 8 0 0 )
2 2
f (x) = (x1+40) +(x2-224) +(x3+40) 2
2
f3 (x) = (xl-224) ' ( 2 4 )
+x+0 +(x3+40) 2
For illustrative purposes, we shall assume that the DM'S structure of
preference can be accurately represented by the utility function U f 1 f2 ,f3)
( '
where
U (f) = -180f - (f2-40000) - (f3-45000) 2 (36)
However, it should be stressed that the explicit form of utility function as
in (36) is used in this example purely for simulating values of MRS.
To be more specific, mkj will be obtained through the following expression:
( f = [au(f)lafj] 1 [au(f)lafkl j=I,... ,n, j#k (37
obtained this way are as if they had been obtained from the ideal DM
%
directly.
Let us now choose fl(x) as our primary objective and formulate the
corresponding €-constraint problem P1 (E -1)-
23. min fl(x)
X
subject to x E X n X1
where X1 ( E - ~ ) = {X 1 fJ. (x) 5 E j'
-
j=2,3 1 (40)
In applying the SPOT, (36) will be used to simulate our imaginary DM.
To be more specific, m12 and m 13 at the ith iteration will be obtained through
(37).
To understand how the SPOT actually work, we give the following description
of the iterations.
Iteration 1
Choose initial =E,:
(:E) = (54000,50000) and solving P (E ) by the
-1 1 -1
generalized reduced gradient ( GRG ) algorithm [Ill yields
1 1 1
(f2,f3) = (54000,50000), fl = 203889.082 and Xi2 = 76.321, Xi3 = 206.654.
From this information, the DM, by giving values (ml m1 ) = (155.555,55 -555)
12' 13
( as simulated by (36)) determines the direction of search (s
s:
:)
, to be
Update (c2,c3) by the formula
Construct the proxy preference function P(f) to determine the optimal step size.
Adopt the following sum-of-exponentials.
24. Obtain two Pareto optimal solutions ' l
f, 2f1 in the neighbourhood of f1 and
assess MRS yields
Using these values together with fl, mi2 and rnl
13 , P(f) becomes
where all the parameters are positive, the parameter values condition is satis-
fied. Calculate Pareto optimal solutions and corresponding P(f) for each
step size a = 1,2,4,8,16,24 yields the following.
a Pareto optimal solutions P(f)
This result shows the maximization point of P(f) is in the neighbourhood of
a = 16, set optimal step size a1 = 16 and go to the next iteration where
The same procedure continues in this manner. In this example, at the
5th iteration optimality condition is satisfied.
In the following, we roughly show the main results for each iteration.
Iteration 2
25. Pf =
() - exp (0.167094~16~f~)-0.123431x10-~ex~
(0.643129~10-~f
2)
- 0.40928xl0-~~ex~(0.129212x10-~f
3)
a2 = 16
Iteration 3
3
8,
2 = (,
E: )
:
8 = (51847.671,52428.516), f = 76325.022
Iteration 4
€4, = (
E: ,
)E
: = (51565.217.52885.796) , f4 = 70717.521
4
s2= - 4 4
(m12-X12)= -0.5598, s3 =
4 - (m13-X13) =
4 4 - 5.0054256
Pf =
() 0
- exp (0.726055~10-~f - 0.136163~10exp(0.220701~10-~f
2)
- 0.959876~10-~exp 3)
(0.119558~10-~f
cr4 = 12
Iteration 5
= (:
E ,E
): = (51558.499,52825.731) , f5 = 76622.752
Optimality test ( 6 = 2 ) is satisfied at this iteration at which
2
f5 = (76622.752,51558.499,52825.731) , x5 = (3.785588,6.185648,6.885281),
U5 = -2.0863306~10
8 . This result compares favorable with the true optimum
8
which is x*= (3.870271,6.136885,6.881835) and U*= -2.08624446~10.
26. 9. Conclusion
In this paper, we have proposed an interactive multiobjective decision
making technique, which we call SPOT, incorporating the desirable features of
both the SWT and the M F method.
U In our interactive on-line scheme, after
solving the epsilon constraint problem, the values of MRS assessed by the DM
were used to determine the direction and the local proxy preference function
was updated to determine the optimal step size. Of course, Pareto optimality
of the generated solution in each iteration is guaranteed in our technique.
An illustrative example demonstrated the feasibility and efficiency of the
SPOT.
Although we have assumed the convexity of the objective functions and
the constraint set, it is possible to extend our technique to the nonconvex
problems by introducing the concept of local Pareto optimality. Furthermore,
extentions to the non-smooth Pareto surface is also possible by utilizing the
directional derivatives. Applications of the SPOT to environmental systems
will be reported elsewhere.
27. References
[I] Keeney, R.L., and Raiffa, H., Decision Analysis with Multiple Conflicting
Objectives : Preference and Value Tradeoffs, John Wiley & Sons, New York,
New York, 1976.
[2.]Haimes, Y.Y., Hall, W.A., and Freedman, H.T., Multiobjective Optimization
in Water Resources Systems : The Surrogate Worth Trade-off Method, Elsevier
Scientific Publishing Company, Amsterdam, Amsterdam, 1975.
[31 Haimes, Y.Y., Hierarchical Analyses of Water Resources Systems, McGraw-Hill,
New York, New York, 1977.
[4] Geoffrion, A.M., Dyer, J.S., and Feinberg, A., An Interactive Approach for
Multicriterion Optimization with an Application to the Operation of an
Academic Department, Management Science, Vol. 19, No. 4, 1972.
[51 Chankong , V. , and Haimes , Y .Y ., The Interactive Surrogate Worth Trade-off
(ISWT) Method for Multiobjective Decision-Making, Multiple Criterion Problem
Solving, Edited by S. Zionts, Springer-Verlag, Berlin, Berlin, 1977.
[61 Haimes, Y .Y., and Chankong, V., Kuhn Tucker Multipliers as Trade-offs in
Multiobjective Decision-Making Analysis, Automatica, Vol. 15, No. 1, 1979.
[71 Shimizu, K., Kawabe, H., and Aiyoshi, E., A Theory for Interactive Preference
Optimization and Its Algorithm - Generalized SWT Method - , The Transactions
of the Institute of Electronics and Communication Engineering of Japan,
Vol., J61-A, No., 11, 1978.
[81 Oppenheimer, K.R., A Proxy Approach to Multi-Attribute Decision Making,
Management Science, Vol. 24, No. 6, 1978.
28. [9] Barrager, S.M., Preferences for Dynamic Lotteries: Assessment and Sensitivity,
Stanford University, PhD Thesis, 1975.
[lo] Keelin, T.W., A Protocol and Procedure for Assessing Multi-Attribute
Preference Functions, Stanford University, PhD Thesis, 1976.
[I11 Lasdon, L.S., Fox, R.L., and Ratner, M.W., Nonlinear Optimization Using
the Generalized Reduced Gradient Method, Revue Francaise dfAutomatique,
Informatique et Recherche Operationnelle, 8 annee, V-3, 1974.