FoodOptimizer

Optimize daily food intake using nutrient requirements, food-specific limits, energy-share constraints (E%), and custom linear dietary constraints.

Installation

Clone the repository and activate the project environment:

julia --project=.

Run interactively:

using Revise
using FoodOptimizer

main()

Or run directly:

julia --project=. -e 'using FoodOptimizer; FoodOptimizer.main()'

Configuration

All file locations, optimisation settings, objectives and output options are configured through config.yml.

Typical configuration options include:

model:
  objective: RI
  energy_unit: kJ
  energy_intake: 2500 kcal
  food_item_limit: 10
  threshold: 0.01

Input Files

The optimiser uses the following input files (in the data folder):

File	Purpose
matvaretabellen_2026.csv	Main food database containing nutrient content per 100 g edible portion.
matvaretabellen_extra.csv	Additional user-defined foods not present in the main database.
food_limits.csv	Optional lower bounds, upper bounds and recommended amounts for individual foods.
nutrient_limits.csv	Lower bounds, upper bounds and recommended intake levels for nutrients.
energy_limits.csv	Lower bounds, upper bounds and recommended energy shares (E%).
custom_constraints.csv	Additional custom dietary constraints expressed as linear equations or inequalities.

Main Food Database

Each row in the food database represents a food identified by a unique Matvare ID.

Nutrient values are specified per 100 g edible portion.

Additional Foods

Additional foods may be added through matvaretabellen_extra.csv.

Example:

Matvare ID,Matvare,Kalsium (Ca) (mg),Jod (I) (µg)
99.001,"Svensk salt, bordsalt, jodert, Jozo",10,5000

Columns missing from matvaretabellen_extra.csv are automatically added during import.

Nutrient Limits

Nutrient requirements are configured in nutrient_limits.csv:

Description,Lower,Upper,Recommended
Protein (g),52.8,135,66.4
Vitamin B12 (kobalamin) (µg),2,,4
Salt (NaCl) (g),0,5.75,

Rules:

• Lower defines a minimum intake.
• Upper defines a maximum intake.
• Recommended is used only by the RI objective.
• Empty cells disable the corresponding constraint.

Food Limits

Food-specific limits are configured in food_limits.csv:

Matvare ID,Matvare,Lower,Upper,Recommended
99.001,"Svensk salt, bordsalt, jodert, Jozo",,,0

Rules:

• Lower defines a minimum amount.
• Upper defines a maximum amount.
• Recommended is used by the RI objective.
• Amounts are expressed in model units of 100 g.

Example:

1.5 = 150 g/day

Energy Share Constraints (E%)

Energy-share constraints are defined in energy_limits.csv.

Example:

Description,Lower,Upper,Recommended
Fett (E%),25,40,32.5
Karbohydrat (E%),45,60,52.5
Protein (E%),10,20,15

FoodOptimizer calculates energy shares from fat, carbohydrates, protein and alcohol using configurable energy factors:

Nutrient	kcal/g	kJ/g
Fat	9	37
Carbohydrate	4	17
Protein	4	17
Alcohol	7	29

The unit is selected through:

model:
  energy_unit: kJ

or

model:
  energy_unit: kcal

Custom Constraints

Additional constraints may be defined in custom_constraints.csv.

Example:

Description,Constraint
Andel umettede fettsyrer,[Enumettede fettsyrer (g)]+[Flerumettede fettsyrer (g)]>=2/3*[Fett (g)]
Tilsatt og fritt sukker,[Sukker, tilsatt (E%)]+[Sukker, fritt (E%)]<10

Supported operators:

<=
>=
<
>
=
==
≤
≥

Supported syntax:

[Nutrient]
coefficient*[Nutrient]
constant
fractions such as 2/3

Energy-share variables may be referenced using:

[Fett (E%)]
[Protein (E%)]

The special variable

[total_energy]

is also supported.

Mixing E% terms and ordinary nutrient variables within the same custom constraint is not supported.

Objective Functions

Cost Minimisation

model:
  objective: cost

Minimises total food cost while satisfying all constraints.

Recommended Intake (RI)

model:
  objective: RI

Minimises normalised absolute deviation from:

• nutrient recommendations in nutrient_limits.csv
• food recommendations in food_limits.csv
• energy share recommendations in energy_limits.csv

while satisfying all lower and upper bounds.

Limiting the Number of Foods

The number of selected foods can be restricted:

model:
  food_item_limit: 10
  threshold: 0.01

When food_item_limit > 0, binary decision variables are introduced and the optimiser can select at most the specified number of foods.

Foods with amounts below threshold are treated as not selected.

Outputs

The optimiser can generate:

• optimal food amounts
• energy-share summary
• nutrient summary
• scaled nutrient contribution per selected food
• binding constraint report
• total daily cost
• reduced costs for selected foods (LP models only)

Results may optionally be exported to Excel:

io:
  write_results_to_excel: true

Sheets:

• food_amounts
• energy_summary
• nutrient_summary
• scaled_foods

Creating the Matvaretabellen CSV

From the Matvaretabellen 2026 Excel workbook, export the sheet:

Matvarer (alle næringsstoffer)

Remove category header rows such as:

Grønnsaker
Drikke
Frukt

and export the remaining data to CSV.

If values are stored with decimal commas, convert them prior to export. One approach is to create a helper sheet using:

=NUMBERVALUE('Matvarer (alle næringsstoffer)'!C2, ",", " ")

and copy the formula over all numeric cells.

Disclaimer

This software is intended for educational, research and informational purposes only.

It is not a substitute for professional dietary, nutritional or medical advice.

References

1. Helsedirektoratet. Referanseverdier for energi og næringsstoffer. https://www.helsedirektoratet.no/rapporter/referanseverdier-for-energi-og-naeringsstoffer
2. Matvaretabellen 2026. Mattilsynet. https://www.matvaretabellen.no