MarabouNetworkNNet¶

Top contributors (to current version):

Christopher Lazarus
Andrew Wu
Kyle Julian
Alex Usvyatsov

This file is part of the Marabou project. Copyright (c) 2017-2019 by the authors listed in the file AUTHORS in the top-level source directory) and their institutional affiliations. All rights reserved. See the file COPYING in the top-level source directory for licensing information.

MarabouNetworkNNet represents neural networks with piecewise linear constraints derived from the NNet format

class maraboupy.MarabouNetworkNNet.MarabouNetworkNNet(filename='', normalize=False)[source]¶

Bases: maraboupy.MarabouNetwork.MarabouNetwork

Class representing a Marabou network from an NNet file Inherits from MarabouNetwork class

numLayers[source]¶

The number of layers in the network

Type: int

layerSizes[source]¶

Layer sizes

Type: list of int

inputSize[source]¶

Size of the input

Type: int

outputSize[source]¶

Size of the output

Type: int

maxLayersize[source]¶

Size of largest layer

Type: int

inputMinimums[source]¶

Minimum value for each input

Type: list of float

inputMaximums[source]¶

Maximum value for each input

Type: list of float

inputMeans[source]¶

Mean value for each input

Type: list of float

inputRanges[source]¶

Range for each input

Type: list of float

outputMean[source]¶

Mean value of outputs

Type: float

outputRange[source]¶

Range of output values

Type: float

weights[source]¶

Network weight matrices, where the outer index corresponds to layer number

Type: list

biases[source]¶

Network bias vectors, where the outer index corresponds to layer number

Type: list

b_variables[source]¶

List of b variables

Type: list of int

f_variables[source]¶

List of f variables

Type: list of int

Parameters

filename (str) – Path to the .nnet file
normalize (bool) – True if network parameters should be adjusted to incorporate network input/output normalization. Otherwise, properties must be written with the normalization already incorporated.

addAbsConstraint(b, f)[source]¶

Function to add a new Abs constraint

Parameters

b (int) – Variable representing input of the Abs constraint
f (int) – Variable representing output of the Abs constraint

addDisjunctionConstraint(disjuncts)[source]¶

Function to add a new Disjunction constraint

Parameters: disjuncts (list of list of Equations) – Each inner list represents a disjunct

addEquality(vars, coeffs, scalar)[source]¶

Function to add equality constraint to network

\[\sum_i vars_i * coeffs_i = scalar\]

Parameters

vars (list of int) – Variable numbers
coeffs (list of float) – Coefficients
scalar (float) – Right hand side constant of equation

addEquation(x)[source]¶

Function to add new equation to the network

Parameters: x (Equation) – New equation to add

addInequality(vars, coeffs, scalar)[source]¶

Function to add inequality constraint to network

\[\sum_i vars_i * coeffs_i \le scalar\]

Parameters

vars (list of int) – Variable numbers
coeffs (list of float) – Coefficients
scalar (float) – Right hand side constant of inequality

addMaxConstraint(elements, v)[source]¶

Function to add a new Max constraint

Parameters

elements (set of int) – Variable representing input to max constraint
v (int) – Variable representing output of max constraint

addRelu(v1, v2)[source]¶

Function to add a new Relu constraint

Parameters

v1 (int) – Variable representing input of Relu
v2 (int) – Variable representing output of Relu

addSignConstraint(b, f)[source]¶

Function to add a new Sign constraint

Parameters

b (int) – Variable representing input of Sign
f (int) – Variable representing output of Sign

clear()[source]¶: Reset values to represent empty network

clearNetwork()[source]¶: Clears the network

createRandomInputsForNetwork()[source]¶

Create a random input for the network.

The value for each input variable is chosen uniformly at random between the lower and the upper bounds for that variable.

Returns: (list of float)

evaluate(inputValues, useMarabou=True, options=None, filename='evaluateWithMarabou.log')[source]¶

Function to evaluate network at a given point

Parameters

inputValues (list of np arrays) – Inputs to evaluate
useMarabou (bool) – Whether to use Marabou solver or TF/ONNX, defaults to True
options (Options) – Object for specifying Marabou options, defaults to None
filename (str) – Path to redirect output if using Marabou solver, defaults to “evaluateWithMarabou.log”

Returns

Values representing the output of the network or None if output cannot be computed

Return type

(np array)

evaluateNNet(inputs, first_layer=0, last_layer=- 1, normalize_inputs=False, normalize_outputs=False, activate_output_layer=False)[source]¶

Evaluate nnet directly (without Marabou) at a given point

Evaluates the network on input inputs between layer first_layer (input layer by default) and layer last_layer (output layer by default)

Parameters

inputs (list of float) – Network inputs to be evaluated
first_layer (int) – The initial layer of the evaluation
last_layer (int) – The last layer of the evaluation
normalize_inputs (bool) – If True and first_layer==0, normalization of inputs is performed
normalize_outputs (bool) – If True, normalization of output is undone
activate_output_layer (bool) – If True, the last layer is activated, otherwise it is not.

Returns

the result of the evaluation

Return type

(list of float)

evaluateWithMarabou(inputValues, filename='evaluateWithMarabou.log', options=None)[source]¶

Function to evaluate network at a given point using Marabou as solver

Parameters

inputValues (list of np arrays) – Inputs to evaluate
filename (str) – Path to redirect output if using Marabou solver, defaults to “evaluateWithMarabou.log”
options (Options) – Object for specifying Marabou options, defaults to None

Returns

Values representing the output of the network or None if system is UNSAT

Return type

(np array)

evaluateWithoutMarabou(inputValues)[source]¶

Evaluate network directly (without Marabou) at a given point

Parameters: inputValues (list of np array) – Input to network
Returns: Output of the network
Return type: (np array)

findError(inputValues, options=None, filename='evaluateWithMarabou.log')[source]¶

Function to find error between Marabou solver and TF/Nnet at a given point

Parameters

inputValues (list of np arrays) – Input values to evaluate
options (Options) –
filename (str) – Path to redirect output if using Marabou solver, defaults to “evaluateWithMarabou.log”

Returns

Values representing the error in each output variable

Return type

(np array)

getBoundsForLayer(layer, b=True)[source]¶

Returns a tuple of two lists, the lower and upper bounds for the variables corresponding to the given layer

If b=True, b variable Otherwise, f variable

Parameters

layer (int) – Layer number
b (bool) – If True, the b variable is considered, otherwise the f variable

Returns

list of lower bounds, list of upper bounds

Return type

(tuple of two lists of float)

getInputMaximum(input_var)[source]¶

Get maximal value for given input variable

Parameters: input_var (int) – index of input variable
Returns: (float)

getInputMinimum(input_var)[source]¶

Get minimum value for given input variable

Parameters: input_var (int) – Index of input variable
Returns: (float)

getLowerBound(layer, node, b=True)[source]¶

Get lower bound for the variable corresponding to layer, node

If b=True, b variable Otherwise, f variable

Parameters

layer (int) – Layer number
node (int) – Node index within layer
b (bool) – If True, the b variable is considered, otherwise the f variable

Returns

(float)

getLowerBoundsForLayer(layer, b=True)[source]¶

Returns a list of lower bounds for the given layer

If b=True, b variables Otherwise, f variables

Parameters

layer (int) – Layer number
b (bool) – If True, the b variable is considered, otherwise the f variable

Returns

(list of float)

getMarabouQuery()[source]¶

Function to convert network into Marabou InputQuery

Returns: InputQuery

getUpperBound(layer, node, b=True)[source]¶

Get upper bound for the variable corresponding to layer, node

If b=True, b variable Otherwise, f variable

Parameters

layer (int) – Layer number
node (int) – Node index within layer
b (bool) – If True, the b variable is considered, otherwise the f variable

Returns

(float)

getUpperBoundsForLayer(layer, b=True)[source]¶

Returns a list of upper bounds for the given layer

If b=True, b variables Otherwise, f variables

Parameters

layer (int) – Layer number
b (bool) – If True, the b variable is considered, otherwise the f variable

Returns

(list of float)

getVariable(layer, node, b=True)[source]¶

Get variable number corresponding to layer, node

If b=True, b variable Otherwise, f variable If layer is 0 (input layer), always returns the f variable

Parameters

layer (int) – Layer number
node (int) – Node index within layer
b (bool) – If True, the b variable is considered, otherwise the f variable

Returns

(int)

lowerBoundExists(x)[source]¶

Function to check whether lower bound for a variable is known

Parameters: x (int) – Variable to check

numberOfVariables()[source]¶

Get total number of variables in network

Returns: (int)

resetNetworkFromParameters(weights: list, biases: list, normalize=False, inputMinimums=[], inputMaximums=[], inputMeans=[], inputRanges=[], outputMean=0, outputRange=1)[source]¶

Recompute the attributes of the network from basic arguments

Parameters

weights (list of list of list of float) – Outer index corresponds to layer number
biases (list of list of float) – Outer index corresponds to layer number
inputMinimums (list of float) –
inputMaximums (list of float) –
inputMeans (list of float) – Mean value for each input
inputRanges (list of float) – Range for each input
outputMean (float) – Mean value of outputs
outputRange (float) – Range of output values

saveQuery(filename='')[source]¶

Serializes the inputQuery in the given filename

Parameters: filename – (string) file to write serialized inputQuery

setLowerBound(x, v)[source]¶

Function to set lower bound for variable

Parameters

x (int) – Variable number to set
v (float) – Value representing lower bound

setUpperBound(x, v)[source]¶

Function to set upper bound for variable

Parameters

x (int) – Variable number to set
v (float) – Value representing upper bound

solve(filename='', verbose=True, options=None)[source]¶

Function to solve query represented by this network

Parameters

filename (string) – Path for redirecting output
verbose (bool) – If true, print out solution after solve finishes
options (Options) – Object for specifying Marabou options, defaults to None

Returns

tuple containing:

vals (Dict[int, float]): Empty dictionary if UNSAT, otherwise a dictionary of SATisfying values for variables
stats (Statistics): A Statistics object to how Marabou performed

Return type

(tuple)

upperBoundExists(x)[source]¶

Function to check whether upper bound for a variable is known

Parameters: x (int) – Variable to check

writeNNet(file_name: str)[source]¶

Write network data into an .nnet file

Parameters: file_name (str) – Path to the file to which the network will be written