Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and...
Transcript of Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and...
![Page 1: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/1.jpg)
Optimum decision making under
uncertainty and risk
Presented by: Gautam Mitra
Acknowledgements:Francis Ellison, Cormac Lucas,
(late)Ken Darby-Dowman, Enza Messina, Patrick Valente,
Christian Valente, Victor Zverovich, Nico Di Domenica,
Katharina Schwaiger, Leela Mitra, and others…
![Page 2: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/2.jpg)
Outline Background
Decision making under uncertainty
Algebraic representation
Computational framework
Modelling framework
Asset and Liability Management (ALM):
Introduction
An example
Stochastic and Robust optimisation comes of age
Why we need high performance solvers?
Conclusions
![Page 3: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/3.jpg)
Background: < the message >
Linear and integer optimisation have been around
since 1960.
But only after 1980s with spreadsheet optimisation
could you show proof of concepts
(Business school … toy models)
As
1. Modelling systems (AMPL, GAMS, OPL…)
supported the creation of complex models
2. Solver systems reached high performance
(CPLEX, …., ….)
Construction of Optimisation based Decision
Support Systems became a reality.
![Page 4: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/4.jpg)
Background:
“The time has come ,” the walrus said “to talk of
many things of gold and spice and mica and mice
and how to give them wings” – Lewis Carol
We are able to not only model
1. Optimum decision problems under uncertainty
and risk
2. We are also able to process, solve and evaluate
them for decision makers…
We have given SP wings…SP can now fly….!
Let us see how….!!
![Page 5: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/5.jpg)
Background :Optimisation – What…Why…How
![Page 6: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/6.jpg)
Background :Optimisation – What…Why…How
![Page 7: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/7.jpg)
Background: Leading applications
Logistics
Manufacturing logistics
Retail logistics
Scheduling Transport scheduling (airline, bus, railways)
Personnel scheduling
Energy systems
Refinery operations scheduling
Resource allocation Capital budgeting – Project planning
Financial planning
![Page 8: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/8.jpg)
Outline Background
Decision making under uncertainty
Algebraic representation
Computational framework
Modelling framework
Asset and Liability Management (ALM):
Introduction
An example
Stochastic and Robust optimisation comes of age
Why we need high performance solvers?
Conclusions
![Page 9: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/9.jpg)
Decision Making: Role of uncertainty and time
Any planning process which takes into account future outcomes has two
important determinants which are:
a) Uncertainty
b) Time
Traditional approach accounts for these by net present value (NPV)
In optimisation based approach planning decisions are broken down to:
a) Robust strategic decision followed by
b) Rolling contingency plans which respond to different future outcomes.
Stochastic Programming (SP) breaks down the decision making into two
parts more or less along these two lines. In SP we make:
a) First stage decisions
b) Corrective (recourse) actions (decisions) as future events unfold.
![Page 10: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/10.jpg)
Decision making:optimisation models with uncertainty representation
Notation:
Let (,F,P) denote a probability space, where denotes a
particular realisation of the uncertain parameters and p(), the
corresponding probability.
Let us denote the realisations of A, b, c for a given as
mnnm RbRxcRAwhere
x
bAxtosubject
cxZ
;,;
0
min
orcbA ,,
![Page 11: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/11.jpg)
Decision making:Alternative models of stochastic programming
• Single stage stochastic programs …(Markowitz)
• Two stage stochastic programs with recourse
• Multi stage stochastic programs with recourse
• Chance constrained programming problem
• Integrated Chance constrained programming problem
• Robust optimisation
![Page 12: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/12.jpg)
Decision making:The classical stochastic linear program with recourse
The stochastic linear program with recourse (SLPR) is stated as
where
.0
min,
y
xBdyDtosubject
yfxQ
,0
,min
x
bAxtosubject
xQEcxZ
![Page 13: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/13.jpg)
Decision making :
Two stage stochastic linear program
Below is a two-stage staircase problem that is transformed
into a two-stage stochastic linear program:
;0,
min
yx
dyDxB
bAxtosubject
fyEcxZ
![Page 14: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/14.jpg)
Decision making:
Two stage stochastic programming
Define as an index set ={1,...,K}, meaning that the parameter may take on K different values.
0.110
0...,,,
...min
21
2222
1111
2211
pandp
yyyx
dyDxB
dyDxB
dyDxB
bAxtosubject
fypfypfypcxZ
K
KKKK
KK
![Page 15: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/15.jpg)
In this tree x represents the first stage (Here-and-Now) decision vector. =1,…,K are the possible future realisations (scenarios); the
associated probabilities and demands are p, d respectively. As future unfolds for a given scenario a corrective recourse action is
taken; this is represented by the vector y.
(B, D, d, f,) 1
(B, D, d, f) 2
(B, D, d, f) K
x
Here-and-Now
y1
y2
yK
=1,…,K future realisations
Decision making:Two Stage Stochastic Linear Programming
![Page 16: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/16.jpg)
Optimum decision
model and constraints
Model of randomness - Scenario Generator
Stochastic Programming
Decision making : the SP paradigm
Stochastic programming seen as a combination of optimisation decision models and models of
randomness
![Page 17: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/17.jpg)
Decision making:Stochastic processes and scenario generation
![Page 18: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/18.jpg)
FansMathematical
structureTrees
Decision making: Stochastic processes and scenario generation
Stochastic processes generate data in form of
fans/time series but SP needs trees
Mathematical structure
![Page 19: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/19.jpg)
Outline Background
Decision making under uncertainty
Algebraic representation
Computational framework
Modelling framework
Asset and Liability Management (ALM):
Introduction
An example
Stochastic and Robust optimisation comes of age
Why we need high performance solvers?
Conclusions
![Page 20: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/20.jpg)
Introducing ALM
![Page 21: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/21.jpg)
Introducing ALM
Application areas:
Banks <credit portfolios>
Bond issuance <corporate as well as sovereign>
University endowments
Hedge funds and mutual funds
Wealthy individuals
Insurance companies and
Pension funds
![Page 22: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/22.jpg)
Introducing ALM
Fixed Mix Strategy
Dynamic Asset Only Strategy
Asset and Liability
Management
Asset and Liability
Management under
Uncertainty
![Page 23: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/23.jpg)
ALM example: Pension fund cash flows
2323
• Investment: portfolio of fixed income and cash
Cash
Injection
![Page 24: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/24.jpg)
Asset & Liability Management: ALM
Optimum
Decision Models
Models of
Randomness
Asset Liability Asset and
Liability
Management
under
Uncertainty:
•SP
•CCP
•ICCP
![Page 25: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/25.jpg)
Surplus Wealth = assets – PV(liabilities) – PV(goals)
Inflows
Wealth
Outflows
carry
Inflows
Wealth
Outflows
Inflows
Wealth
Outflows
carry
t=1 t=2..T-1 t=T
ALM : a stochastic programming model
![Page 26: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/26.jpg)
ALM: Model Components
NOTATION ENTITY
INDEX (SET)
i Assets i=1..n
s Scenarios s=1…S
t Timeperiod t=1..T
K RiskGroup k=1..K
PARAMETER
Pist Prices[assets, scenarios, timeperiod]
Lts Liability[timeperiod,secenario]
s Probability[scenario]
Hi0 Initial Holdings[assets, timeperiod=0]
Ft Funding[timeperiod]
1+tr Transaction Buy
1- tr Transaction Sell
Rk RiskGroup Holdings
DECISION VARIABLE
Hist Amounthold[assets, scenarios,timeperiod]
Bist Amountbuy[assets, scenarios,timeperiod]
Sist Amountsell[assets, scenarios,timeperiod]
![Page 27: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/27.jpg)
ALM: Model Components
Name
i asset
s scenario
PistPrices[assets, scenarios,
timeperiod]
Lts Liability[timeperiod,scenario]
s prob[scenario]
tr Sales transaction cost (%)
Hist Amounthold[assets,
scenarios,timeperiod]
Bist Amountbuy[assets, scenarios,timeperiod]
Sist Amountsell[assets,
scenarios,timeperiod]
SETSset asset;
scenarioset scenario;
VARIA
BLES
var hold{assets, scenarios, time};
var buy{assets, scenarios, time};
var sell{assets, scenarios, time};
PARAME
TERS
random param prices{assets, ... };
param liability{timep,scenario};
probability param prob{scenario};
param tsell;
SAMPL syntax
![Page 28: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/28.jpg)
ALM: stochastic programming model…cont..
+Hi0 +(1-tr)Si1sPi1s +F1
-L1s
WEALTH WEALTH
+Ft +(1-tr)Si tsPi ts
-(1+tr)Bi tsPi ts -Lts
WEALTH
+(1-tr)SiTsPiTs
-LTs
t=1 t=2...T-1 t=T
-G1 s -Gt s -GTs
-(1+tr)Bi1sPi1s -(1+tr)BiTsPiTs
Surplus Wealth = assets – PV(liabilities) – PV(goals)
subject to fundbalance1{t in 2..T-1,s in scen}:
sum{a in assets} amountbuy[t,a,s]*price[t,a,s]*tbuy
-sum{a in assets} amountsell[t,a,s]*price[t,a,s]*tsell=
income[t,s]-liabilities[t,s];
SAMPL syntax
![Page 29: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/29.jpg)
Outline Background
Decision making under uncertainty
Algebraic representation
Computational framework
Modelling framework
Asset and Liability Management (ALM):
Introduction
An example
Stochastic and Robust optimisation comes of age
Why we need high performance solvers?
Conclusions
![Page 30: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/30.jpg)
Name Affiliation System Name Type
JJ Bisshop, et al. Paragon Decision
Tech.
AIMMS Modelling System
A Meeraus, et al. GAMS GAMS Modelling System
B Kristjansson Maximal Software MPL Modelling System
R Fourer, et al. Northwestern
University
AMPL Modelling System
MAH Dempster, et al. Cambridge University STOCHGEN Modelling System
E Fragniere, et al. University of Geneva SETSTOCH Modelling System
A King, et al. IBM/COIN-OR OSL/SE, SMI Solver
HI Gassmann, et al. Dalhousie University MSLiP Solver
G Infanger et. Al. Stanford University DECIS Solver
P Kall, et al. University of Zürich SLP-IOR Modelling System /
Solver
G Mitra, et al. Brunel University AMPLDev SP Edition Modelling System /
Solver
A. Ramos U P Comillas MSP & SIP Modelling System /
Solver
A. Gavronsky Norwegian Univ of
Tech
ALM Modelling System
SP comes of age: SP software tools
![Page 31: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/31.jpg)
SP comes of age: modelling paradigms
• Descriptive Models as defined by a set of mathematical relations,which simply predicts how a physical, industrial or a social systemmay behave.
• Normative Models constitute the basis for (quantitative) decisionmaking by a superhuman following an entirely rational that is,logically scrupulous set of arguments. Hence quantitative decisionproblems and idealised decision makers are postulated in order todefine these models.
• Prescriptive Models involve systematic analysis of problems ascarried out by normally intelligent persons who apply intuition andjudgement. Two distinctive features of this approach are uncertaintyanalysis and preference (or value or utility) analysis.
• Decision Models are in some sense a derived category as theycombine the concept underlying the normative models andprescriptive models.
![Page 32: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/32.jpg)
Role of uncertainty and time
Any planning process which takes into account future outcomes has two
important determinants which are:
a) Uncertainty
b) Time
Traditional approach accounts for these by net present value (NPV)
Also planning decisions have been broken down to:
a) Robust strategic decision followed by
b) Rolling contingency plans which respond to different future outcomes.
Stochastic Programming (SP) breaks down the decision making into two
parts more or less along these two lines. In SP we make:
a) First stage decisions
b) Corrective (recourse) actions as future events unfold.
![Page 33: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/33.jpg)
Time
0 1 2 3 … T
Ex ante decision Ex post evaluation
(simulation)
SP comes of age: scope of models
Data Model
Decision Model: Constrained optimisation
Descriptive Model: Simulation and Evaluation
![Page 34: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/34.jpg)
SP comes of age: Uncertainty models
SP Problems
Distribution Problems
Wait and See
Expected Value
Recourse Problems
Distribution based
Scenario based
Problems with Chance
Constraints
Problems with ICC
Robust Optimisation
Dynamic Programming
![Page 35: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/35.jpg)
SP comes of age:
inter-related modelling paradigms
The modelling paradigms revisited
Stage 0: Analyse historical data [ data model ]
Stage 1: Create data paths for random parameter
values [descriptive models]
Stage 2: Make decision using SP or DP
[decision models] (data processes and decision
processes are inter-twined)
Stage 3: Test the decisions using simulation Back
testing, stress testing, out of sample [descriptive
models]
![Page 36: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/36.jpg)
Models of (parameter)
randomness
Scenario
Generator 1
Scenario
Generator 2
Simulation
and
Decision evaluation
Performance measures
Statistical measures: mean, variance, skewness, kurtosis
Stochastic measures: EVPI, VSS
Risk measures: VaR, CVaR, standard deviation
Performance measures: Solvency ratio, Sharpe ratio, Sortino ratio
Ex-ante decision models
Two-stage SP - recourse
Multistage SP- recourse
Chance-constrained SP
Expected value LP
Integrated chance constraints
Robust optimisation
SP comes of age: decision…simulation…evaluation engine
![Page 37: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/37.jpg)
Two stage SP
Multi stage SP
Chance Constrained
Integrated CCPs
ROBUST
Soyster’s
ROBUST
Bertsimas
Benders
Nested Benders
Level
ICCP cutting plane
Deterministic Equivalent
ROBUST
Ben-Tal
QP
MIP
LP
SOCP
SP comes of age: Solution methods
![Page 38: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/38.jpg)
Outline Background
Decision making under uncertainty
Algebraic representation
Computational framework
Modelling framework
Asset and Liability Management (ALM):
Introduction
An example
Stochastic and Robust optimisation comes of age
Why we need high performance solvers?
Conclusions
![Page 39: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/39.jpg)
Why high performance solvers: Problem structure
• Structure of deterministic equivalent makes it suitable for
decomposition methods
• IPM can be often more appropriate than Simplex due to large size and high sparsity
A
T1
TK
…
W1
WK
A
T
T
T
T
T
T
W
W
W
W
W
W
![Page 40: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/40.jpg)
Why hp solver: …Solver Architecture
FortSP.exe
FortSP.dll
cplex*.dll FortMP.dllOsiClp.dll
libClp.lib
FortSP
External solvers
Benders, Level,
…, DEQ
![Page 41: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/41.jpg)
Why hp solvers: …Test problems
![Page 42: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/42.jpg)
Test problems
![Page 43: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/43.jpg)
Why hp solvers: Test results
![Page 44: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/44.jpg)
Outline Background
Decision making under uncertainty
Algebraic representation
Computational framework
Modelling framework
Asset and Liability Management (ALM):
Introduction
An example
Stochastic and Robust optimisation comes of age
Why we need high performance solvers?
Conclusions
![Page 45: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/45.jpg)
Conclusions: Assertions and Commercials
The models and tools for decision making under
uncertainty, optimum risk decisions have matured.
For more information and to gain SP skills attend
OptiRisk Workshops…..!
The domains of supply chain logistics, energy systems
planning and operation … especially financial
planning are ready for SP
Quants may use
SP modelling support for rapid prototyping
Special purpose solvers to scale up prototypes
![Page 46: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/46.jpg)
Conclusions: Assertions and Commercials
A particularly valuable aspect of scenario based SP is
the dual paradigm of
Ex ante optimisation decisions followed by
Ex poste decision evaluation
OptiRisk is a certified partner of IBM in the UK and in
India
OptiRisk and IBM together are ready for business…!
We will be waiting for you to call or email…!!
![Page 47: Optimum decision making under uncertainty and risk · Optimum decision making under uncertainty and risk ... future unfolds for a given scenario a corrective recourse action is ...](https://reader031.fdocuments.us/reader031/viewer/2022022506/5ac3b4a97f8b9aa0518ca9a4/html5/thumbnails/47.jpg)
Thank you....
Thank you for your attention
Comments and Questions please
{ not difficult or awkward ones…! }