Mathematical Modeling, Optimization, and Linear Programming Formulation

Fundamentals of Mathematical Modeling and Optimization

• Definition and Generalization of Profit Functions:

  • Profit is fundamentally expressed as a mathematical function of revenue and expenses:         $\text{Profit} = \text{Revenue} - \text{Expenses}$
  • More broadly, an operational model can generalize profit as a function of multiple independent variables or inputs:         $\text{Profit} = f(I_1, I_2, \dots, I_n)$         where $n$ represents the total number of system inputs influencing financial yield.

• Iterative Procedure of Model Development:

  • The creation and deployment of an operational optimization model is an iterative engineering process comprising several sequential steps:
    • Problem Identification: Clearly identifying the real-world operational challenge or production context.
    • Mathematical Formulation: Translating system requirements and physical relationships into formal mathematical formulas, objective functions, and constraints.
    • Model Analysis: Solving the equations and running computational routines to obtain numerical results.
    • Pre-Implementation Evaluation: Analyzing the output data critically to verify if the proposed numerical solution makes practical sense.
    • Iterative Feedback Loop: If solutions prove invalid or non-sensical, returning to examine the foundational assumptions and defined relationship formulas before modifying and re-solving the model.
    • Implementation: Deploying the finalized strategy into production once the output is thoroughly validated.

• Production Line Optimization Example:

  • In a manufacturing production line environment, an optimization model is constructed to determine optimal production volumes across multiple product lines, denoted as products $a$, $b$, $c$, and $d$, in order to maximize overall profitability.

Model Validation and Cognitive Factors in Decision Making

• Sanity Checks for Model Feasibility:

  • Mathematical models may generate results that are mathematically consistent with unconstrained formulas but physically impossible in practice.
  • For example, if an unconstrained production model produces an output of $-2$ units for product $a$ or product $b$, the result is invalid because negative quantities cannot be manufactured.
  • Such illogical outputs necessitate a direct investigation of which structural assumptions or formulated relationships within the model failed to reflect physical reality.

• Integration of Human Judgment:

  • Prior to implementation, computational solutions must be reviewed and tempered by human judgment to catch model blind spots.
  • Human judgment itself exhibits specific cognitive characteristics and biases that influence decision evaluation.

• Anchoring Effects in Estimation:

  • The anchoring effect is a cognitive bias wherein human estimates are heavily anchored to the initial presentation or sequence of information.
  • Sequential Multiplication Example: An individual asked to estimate the product of the sequence of integers from $1$ through $8$ ($1 \times 2 \times 3 \times 4 \times 5 \times 6 \times 7 \times 8$) will produce an initial estimate that varies significantly depending on whether the sequence is presented in ascending order ($1$ through $8$) or in an alternate presentation order.

• Decision Evaluation Under Probabilistic Risk:

  • Human decision-making frequently evaluates high-consequence opportunities governed by probabilistic rules.
  • Die Roll Hypothetical: A scenario granting an individual the opportunity to win a payout of \\$1{,}000{,}000 conditional on rolling a fair six-sided die and having it land on the outcome of $6$.

Structural Components of Linear Optimization Models

• System Harmonization:

  • A successful optimization model harmonizes four distinct operational dimensions:
    • What can be done (system capacities and potential actions).
    • What is known (parameters, input constants, and empirical data).
    • What is wanted (the designated operational objective).
    • The boundaries to which the system is subject (operational limits and restrictions).

• Decision Variables:

  • Decision variables represent the specific quantities or operational choices under direct managerial control within a problem.
  • In a commercial food preparation scenario, the primary decision variables correspond to package quantities:
    • $x_1$: Number of sandwich boxes to produce.
    • $x_2$: Number of salad boxes to produce.

• Objective Concept:

  • The objective represents the overarching operational target, which must be either maximized or minimized.
  • In profit-driven environments, the objective is to maximize total monetary return as much as possible.
  • Absent restrictions, higher production yields strictly greater profit; producing $1{,}000$ boxes generates significantly higher financial return than producing $500$ or $200$ boxes.

• Role of Constraints and Bottlenecks:

  • Production cannot expand toward infinity because operations are constrained by finite resources.
  • The primary operational constraint restricting output volume is available preparation time, alongside physical ingredient supplies.

Formulation of the Sandwich and Salad Production Model

• Mathematical Representation of Decision Variables:

  • Let $x_1$ be the unknown quantity of sandwich boxes.
  • Let $x_2$ be the unknown quantity of salad boxes.

• Objective Function Formulation:

  • Unit profit contributions:
    • Each sandwich box yields a profit of \\$8.
    • Total profit contribution from sandwiches is represented by $8x_1$.
    • Each salad box yields a profit of \\$6.
    • Total profit contribution from salads is represented by $6x_2$.
  • Total system profit is denoted by $z$.
  • The complete objective function to maximize is:         $\text{Maximize } z = 8x_1 + 6x_2$

• Time Resource Constraint Formulation:

  • Unit time consumption values:
    • Preparation of one sandwich box requires $4\,\text{minutes}$.
    • Total time required to prepare all sandwich boxes is $4x_1$.
    • Preparation of one salad box requires $2\,\text{minutes}$.
    • Total time required to prepare all salad boxes is $2x_2$.
  • Demonstration Scenario ($x_1 = 5$, $x_2 = 10$):
    • Time required for $5$ sandwich boxes:             $5 \times 4 = 20\,\text{minutes}$
    • Time required for $10$ salad boxes:             $10 \times 2 = 20\,\text{minutes}$
    • Total combined preparation time consumed:             $20 + 20 = 40\,\text{minutes}$
  • Total time consumed cannot exceed the total operational time available. The algebraic time constraint is formulated as:         $4x_1 + 2x_2 \le T$         where $T$ represents the upper limit of available operational minutes.

• Ingredient Resource Constraint Formulation:

  • Unit ingredient usage values:
    • Preparation of each sandwich box requires $3\,\text{units}$ of ingredients.
    • Total ingredient volume consumed by sandwiches is $3x_1$.
    • Preparation of each salad box consumes $4\,\text{units}$ of ingredients, yielding an aggregate salad ingredient consumption of $4x_2$.
  • The combined usage of ingredients cannot exceed the maximum inventory available. The algebraic ingredient constraint is formulated as:         $3x_1 + 4x_2 \le I$         where $I$ represents the upper limit of total available ingredient units.

Domain Specification and Non-Negativity Constraints

• Need for Computational Domain Bounds:

  • Computing systems and automated solvers possess no intrinsic common sense regarding the physical reality of manufactured products.
  • Without explicit parameter restrictions, a solver could produce mathematically viable but physically absurd values, such as negative quantities of food boxes.

• Non-Negativity Constraints:

  • Variables must be mathematically bounded to non-negative values to ensure physical feasibility:         $x_1 \ge 0$$x_2 \ge 0$

• Integer Domain Specification ($\mathbb{Z}^+$):

  • Because sandwich boxes and salad boxes are discrete units, a more precise mathematical representation requires variables to belong to the set of non-negative integers ($\mathbb{Z}^+ \cup \{0\}$):         $x_1, x_2 \in \{0, 1, 2, 3, 4, \dots\}$
  • Specifying this discrete domain establishes the final structural boundary condition of the mathematical program.

Questions & Discussion

• Verification of Formulation Principles:
  • Query: How is the base profit metric structured mathematically?
    • Explanation: Profit is formally defined as revenue minus expenses and generalized into a multi-input function representing system parameters.
  • Query: What occurs if total resource consumption exceeds available capacities?
    • Explanation: If combined production requires resources that exceed defined limits (such as available preparation minutes or inventory units), the operational plan is infeasible and cannot be produced.
  • Open Floor Inquiries: Solicitations for audience questions regarding model components, domain boundaries, and parameter constraints before proceeding.