Amply allows you to load and manipulate AMPL data as Python data structures.

Amply only supports a specific subset of the AMPL syntax:

• set declarations
• set data statements
• parameter declarations
• parameter data statements

Typically, problems expressed in AMPL consist of two parts, a model section and a data section. Amply is only designed to parse the parameter and set statements contained within AMPL data sections. However, in order to parse these statements correctly, information that would usually be contained within the model section may be required. For instance, it may not be possible to infer the dimension of a set purely from its data statement. Therefore, Amply also supports set and parameter declarations. These do not have to be put in a separate section, they only need to occur before the corresponding data statement.

The declaration syntax supported is extremely limited, and does not include most elements of the AMPL programming language. The intention is that this library is used as a way of loading data specified in an AMPL-like syntax.

Furthermore, Amply does not perform any validation on data statements.

This document is intended as a guide to the syntax supported by Amply, and not as a general AMPL reference manual. For more in depth coverage see the GNU MathProg manual, Chapter 5: Model data or the following links:

• Sets in AMPL
• Parameters in AMPL

>>> from amply import Amply

Import the class:

>>> from amply import Amply

A simple set. Sets behave a lot like lists.

>>> data = Amply("set CITIES := Auckland Wellington Christchurch;")
>>> print data.CITIES
<SetObject: ['Auckland', 'Wellington', 'Christchurch']>
>>> print data['CITIES']
<SetObject: ['Auckland', 'Wellington', 'Christchurch']>
>>> for c in data.CITIES: print c
...
Auckland
Wellington
Christchurch
>>> print data.CITIES[0]
Auckland
>>> print len(data.CITIES)
3

Data can be integers, reals, symbolic, or quoted strings:

>>> data = Amply("""
...   set BitsNPieces := 0 3.2 -6e4 Hello "Hello, World!";
... """)
>>> print data.BitsNPieces
<SetObject: [0.0, 3.2000000000000002, -60000.0, 'Hello', 'Hello, World!']>

Sets can contain multidimensional data, but we have to declare them to be so first.

>>> data = Amply("""
... set pairs dimen 2;
... set pairs := (1, 2) (2, 3) (3, 4);
... """)
>>> print data.pairs
<SetObject: [(1, 2), (2, 3), (3, 4)]>

Sets themselves can be multidimensional (i.e. be subscriptable):

>>> data = Amply("""
... set CITIES{COUNTRIES};
... set CITIES[Australia] := Adelaide Melbourne Sydney;
... set CITIES[Italy] := Florence Milan Rome;
... """)
>>> print data.CITIES['Australia']
['Adelaide', 'Melbourne', 'Sydney']
>>> print data.CITIES['Italy']
['Florence', 'Milan', 'Rome']

Note that in the above example, the set COUNTRIES didn't actually have to exist itself. Amply does not perform any validation on subscripts, it only uses them to figure out how many subscripts a set has. To specify more than one, separate them by commas:

>>> data = Amply("""
... set SUBURBS{COUNTRIES, CITIES};
... set SUBURBS[Australia, Melbourne] := Docklands 'South Wharf' Kensington;
... """)
>>> print data.SUBURBS['Australia', 'Melbourne']
['Docklands', 'South Wharf', 'Kensington']

Slices can be used to simplify the entry of multi-dimensional data.

>>> data=Amply("""
... set TRIPLES dimen 3;
... set TRIPLES := (1, 1, *) 2 3 4 (*, 2, *) 6 7 8 9 (*, *, *) (1, 1, 1);
... """)
>>> print data.TRIPLES
<SetObject: [(1, 1, 2), (1, 1, 3), (1, 1, 4), (6, 2, 7), (8, 2, 9), (1, 1, 1)]>
>

Set data can also be specified using a matrix notation. A '+' indicates that the pair is included in the set whereas a '-' indicates a pair not in the set.

>>> data=Amply("""
... set ROUTES dimen 2;
... set ROUTES : A B C D :=
...            E + - - +
...            F + + - -
... ;
... """)
>>> print data.ROUTES
<SetObject: [('E', 'A'), ('E', 'D'), ('F', 'A'), ('F', 'B')]>

Matrices can also be transposed:

>>> data=Amply("""
... set ROUTES dimen 2;
... set ROUTES (tr) : E F :=
...                 A + +
...                 B - +
...                 C - -
...                 D + -
... ;
... """)
>>> print data.ROUTES
<SetObject: [('E', 'A'), ('F', 'A'), ('F', 'B'), ('E', 'D')]>

Matrices only specify 2d data, however they can be combined with slices to define higher-dimensional data:

>>> data = Amply("""
... set QUADS dimen 2;
... set QUADS :=
... (1, 1, *, *) : 2 3 4 :=
...              2 + - +
...              3 - + +
... (1, 2, *, *) : 2 3 4 :=
...              2 - + -
...              3 + - -
... ;
... """)
>>> print data.QUADS
<SetObject: [(1, 1, 2, 2), (1, 1, 2, 4), (1, 1, 3, 3), (1, 1, 3, 4), (1, 2, 2, 3), (1, 2, 3, 2)]>

Parameters are also supported:

>>> data = Amply("""
... param T := 30;
... param n := 5;
... """)
>>> print data.T
30
>>> print data.n
5

Parameters are commonly indexed over sets. No validation is done by Amply, and the sets do not have to exist. Parameter objects are represented as a mapping.

>>> data = Amply("""
... param COSTS{PRODUCTS};
... param COSTS :=
...   FISH 8.5
...   CARROTS 2.4
...   POTATOES 1.6
... ;
... """)
>>> print data.COSTS
<ParamObject: {'POTATOES': 1.6000000000000001, 'FISH': 8.5, 'CARROTS': 2.3999999999999999}>
>>> print data.COSTS['FISH']
8.5

Parameter data statements can include a default clause. If a '.' is included in the data, it is replaced with the default value:

>>> data = Amply("""
... param COSTS{P};
... param COSTS default 2 :=
... F 2
... E 1
... D .
... ;
... """)
>>> print data.COSTS['D']
2.0

Parameter declarations can also have a default clause. For these parameters, any attempt to access the parameter for a key that has not been defined will return the default value:

>>> data = Amply("""
... param COSTS{P} default 42;
... param COSTS :=
... F 2
... E 1
... ;
... """)
>>> print data.COSTS['DOES NOT EXIST']
42.0

Parameters can be indexed over multiple sets. The resulting values can be accessed by treating the parameter object as a nested dictionary, or by using a tuple as an index:

>>> data = Amply("""
... param COSTS{CITIES, PRODUCTS};
... param COSTS :=
...  Auckland FISH 5
...  Auckland CHIPS 3
...  Wellington FISH 4
...  Wellington CHIPS 1
... ;
... """)
>>> print data.COSTS
<ParamObject: {'Wellington': {'FISH': 4.0, 'CHIPS': 1.0}, 'Auckland': {'FISH': 5.0, 'CHIPS': 3.0}}>
>>> print data.COSTS['Wellington']['CHIPS'] # nested dict
1.0
>>> print data.COSTS['Wellington', 'CHIPS'] # tuple as key
1.0

Parameters support a slice syntax similar to that of sets:

>>> data = Amply("""
... param COSTS{CITIES, PRODUCTS};
... param COSTS :=
...  [Auckland, * ]
...   FISH 5
...   CHIPS 3
...  [Wellington, * ]
...   FISH 4
...   CHIPS 1
... ;
... """)
>>> print data.COSTS
<ParamObject: {'Wellington': {'FISH': 4.0, 'CHIPS': 1.0}, 'Auckland': {'FISH': 5.0, 'CHIPS': 3.0}}>

Parameters indexed over two sets can also be specified in tabular format:

>>> data = Amply("""
... param COSTS{CITIES, PRODUCTS};
... param COSTS: FISH CHIPS :=
...  Auckland    5    3
...  Wellington  4    1
... ;
... """)
>>> print data.COSTS
<ParamObject: {'Wellington': {'FISH': 4.0, 'CHIPS': 1.0}, 'Auckland': {'FISH': 5.0, 'CHIPS': 3.0}}>

Tabular data can also be transposed:

>>> data = Amply("""
... param COSTS{CITIES, PRODUCTS};
... param COSTS (tr): Auckland Wellington :=
...            FISH   5        4
...            CHIPS  3        1
... ;
... """)
>>> print data.COSTS
<ParamObject: {'Wellington': {'FISH': 4.0, 'CHIPS': 1.0}, 'Auckland': {'FISH': 5.0, 'CHIPS': 3.0}}>

Slices can be combined with tabular data for parameters indexed over more than 2 sets:

>>> data = Amply("""
... param COSTS{CITIES, PRODUCTS, SIZE};
... param COSTS :=
...  [Auckland, *, *] :   SMALL LARGE :=
...                 FISH  5     9
...                 CHIPS 3     5
...  [Wellington, *, *] : SMALL LARGE :=
...                 FISH  4     7
...                 CHIPS 1     2
... ;
... """)
>>> print data.COSTS
<ParamObject: {'Wellington': {'FISH': {'SMALL': 4.0, 'LARGE': 7.0}, 'CHIPS': {'SMALL': 1.0, 'LARGE': 2.0}}, 'Auckland': {'FISH': {'SMALL': 5.0, 'LARGE': 9.0}, '

All functionality is contained within the Amply class.

load_string(string)

Parse string data.

load_file(file)

Parse contents of file or file-like object (has a read() method).

from_file(file)

Alternate constructor. Create Amply object from contents of file or file-like object.

The parsed data structures can then be accessed from an Amply object via attribute lookup (if the name of the symbol is a valid Python name) or item lookup.

from pulp import Amply

data = Amply("set CITIES := Auckland Hamilton Wellington")

# attribute lookup
assert data.CITIES == ['Auckland', 'Hamilton', 'Wellington']

# item lookup
assert data['CITIES'] == data.CITIES

Note that additional data may be loaded into an Amply object simply by calling one of its methods. A common idiom might be to specify the set and parameter declarations within your Python script, then load the actual data from external files.

from pulp import Amply

data = Amply("""
  set CITIES;
  set ROUTES dimen 2;
  param COSTS{ROUTES};
  param DISTANCES{ROUTES};
""")

for data_file in ('cities.dat', 'routes.dat', 'costs.dat', 'distances.dat'):
    data.load_file(open(data_file))

Many thanks to Johannes Ragam (@thet), former custodian of the "amply" project on PyPi. Johannes graciously transferred the project to this. Thanks!