Intervals and boxes

We will first only focus on static items that are not evolving with time. The vectors \(\mathbf{x}\) and matrices \(\mathbf{X}\), presented in the previous pages, have their interval counterpart. This page provides the main uses of these sets. They will be involved afterwards for solving problems.

Note

../../_images/logo_ibex.jpg

These interval variables come from the IBEX library. They are briefly presented here for the sake of consistency. For more information, please refer to the IBEX documentation for C++ use.

Intervals, boxes and interval matrices

  • Interval x(double lb, double ub) will define the interval \([x]\).
    It is made of its lower and upper bounds \([x^{-},x^{+}]\).
    x = Interval(0, 10)                         # [0,10]
    x = Interval(1, oo)                         # [1,∞]
    x = Interval(-oo, -10)                      # [-∞,-10]
    

    Some pre-defined values are also at hand:

    x = Interval()                              # [-∞,∞] (default value)
    x = Interval.EMPTY_SET                      # ∅
    x = Interval.PI                             # [π]
    x = Interval.TWO_PI                         # [2π]
    x = Interval.HALF_PI                        # [π/2]
    

    Note that the constant \([\pi]\) is a reliable enclosure of the \(\pi\) value, that cannot be exactly represented in a computer with a single floating-point value.

    x = Interval.PI                             # [π]
    # x = [3.141592653589793, 3.141592653589794]
    
  • IntervalVector x(int n) is used for \(n\)-d vectors of intervals, also called boxes.
    For instance:
    x = IntervalVector(2, [-1,3])               # creates [x]=[-1,3]×[-1,3]=[-1,3]^2
    y = IntervalVector([[3,4],[4,6]])           # creates [y]= [3,4]×[4,6]
    z = IntervalVector(3, Interval(0,oo))       # creates [z]=[0,∞]^3
    q = IntervalVector([x[1],y[0],z[0]])        # creates [q]=[-1,3]×[3,4]×[0,∞]
    w = IntervalVector(y)                       # creates a copy: [w]=[y]
    
    v = (0.42,0.42,0.42)                        # one vector (0.42;0.42;0.42)
    iv = IntervalVector(v)                      # creates one box that wraps v:
                                                #   [0.42,0.42]×[0.42,0.42]×[0.42,0.42]
    

    One can access vector components as we do for Vector objects:

    x[1] = Interval(0,10)                       # updates to [x]=[-1,3]×[0,10]
    

    The vector operations to handle Vector objects can also be used for boxes:

    n = x.size()                  # box dimension (number of components): 2
    x.resize(5)                   # updates [x] to [-1,3]×[0,10]×[-∞,∞]×[-∞,∞]×[-∞,∞]
    m = x.subvector(1,2)          # creates [m]=[0,10]×[-∞,∞]
    x.put(2,y)                    # updates [x] to [-1,3]×[0,10]×[3,4]×[4,6]×[-∞,∞]
    

    Lastly, the concatenation of two IntervalVector can be done with the cart_prod function:

    a = IntervalVector([[0,1],[2,3]])
    b = IntervalVector([[4,5],[6,7]])
    c = cart_prod(a,b)
    # c: ([0, 1] ; [2, 3] ; [4, 5] ; [6, 7])
    

    Note

    With Python, one can use NumPy arrays for building degenerated IntervalVector objects such as:

    x = IntervalVector(np.array([1,0,0]))       # [1,1]×[0,0]×[0,0]
    y = IntervalVector(np.array([[1],[0],[0]])) # [1,1]×[0,0]×[0,0]
    
  • IntervalMatrix is also available.
    One can refer to the documentation of IBEX for more information.

    Note

    With Python, one can use NumPy matrices for building degenerated IntervalMatrix objects such as:

    x = IntervalMatrix(np.eye(3,2))
    
    # Produces:
    #
    # ((<1, 1> ; <0, 0>)
    #  (<0, 0> ; <1, 1>)
    #  (<0, 0> ; <0, 0>))
    

The empty set

In mathematics, the empty set is the unique set having no elements; it corresponds to one entity while in Codac (as in IBEX) there exists one empty set representation for each class of domain.

Note

In our framework, empty sets correspond to domains that do not contain feasible solutions. This may be the result of a too restrictive definition of the problem, for instance due to some errors in the model or because of outliers in the dataset.

The empty set of an Interval object is given by:

x = Interval.EMPTY_SET                      # ∅

For boxes (interval vectors), we have to specify their dimension even in case of empty set. This differs from mathematical definitions, but allows simple operations when programming with boxes.

x = IntervalVector.empty(3) # ∅×∅×∅

Set operations

Set operations are available for Interval, IntervalVector and IntervalMatrix objects (see the official reference). In the following table, if \([x]\) is an interval object, \(d\) is a real value.

Code

Meaning

x==y

\([x]=[y]\)

x!=y

\([x]\neq [y]\)

x.is_empty()

\([x]=\varnothing\)

x.is_unbounded()

true iff \([x]\) has one of its bounds infinite

x.is_subset(y)

\([x]\subseteq [y]\)

x.is_strict_subset(y)

\([x]\subseteq [y]\wedge [x]\neq [y]\)

x.is_superset(y)

\([x]\supseteq [y]\)

x.is_strict_superset(y)

\([x]\supseteq [y]\wedge [x]\neq [y]\)

x.contains(p)

\(d\in [x]\)

x.intersects(y)

\([x]\cap [y]\neq\varnothing\)

x.is_disjoint(y)

\([x]\cap [y]=\varnothing\)

x.overlaps(y)

\(\mathring{[x]}\cap \mathring{[y]}\neq\varnothing\)

Where \(\mathring{[x]}\) denotes the interior of \([x]\).
In addition of these test functions, operations on sets are available:

Code

Meaning

x&y

\([x]\cap [y]\)

x|y

\([x]\sqcup[y]\)

x.set_empty()

\([x]\leftarrow \varnothing\)

x=y

\([x]\leftarrow [y]\)

x&=y

\([x]\leftarrow ([x]\cap [y])\)

x|=y

\([x]\leftarrow ([x]\sqcup[y])\)

Finally, one can also access properties of the sets. First for Interval:

Return type

Code

Meaning

double

x.lb()

\(\underline{x}\), the lower (left) bound of \([x]\)

double

x.ub()

\(\overline{x}\), the upper (right) bound of \([x]\)

double

x.diam()

diameter, \(|\overline{x}-\underline{x}|\)

double

x.rad()

radius, half of the diameter

double

x.mid()

the midpoint, (\((\underline{x}+\overline{x})/2\))

Interval

x.inflate(eps)

an interval with the same midpoint and radius increased by eps

Then for IntervalVector:

Return type

Code

Meaning

Vector

x.lb()

lower-left corner (vector of lower bounds of \([x]\))

Vector

x.ub()

upper-right corner (vector of upper bounds of \([x]\))

Vector

x.diam()

vector of diameters, \(|\overline{x_i}-\underline{x_i}|\)

double

x.min_diam()

minimal diameter, among all components of [x]

double

x.max_diam()

maximal diameter, among all components of [x]

Vector

x.rad()

vector of radii (halves of diameters)

Vector

x.mid()

the midpoint, (\((\underline{x}+\overline{x})/2\))

double

x.volume()

the volume of the box

bool

x.is_flat()

true if the volume is null (one dimension is degenerated)

IntervalVector

x.inflate(eps)

new box: same midpoint and each radius increased by eps

bool

x.is_unbounded()

true iff \([x]\) has one of its bounds infinite

Interval arithmetic

Interval analysis is based on the extension of all classical real arithmetic operators. Consider two intervals \([x]\) and \([y]\) and an operator \(\diamond\in\left\{+,-,\cdot,/\right\}\). We define \([x]\diamond[y]\) as the smallest interval containing all feasible values for \(x\diamond y\), assuming that \(x\in[x]\) and \(y\in[y]\).

\[\begin{split}[x]\diamond[y]&=&\left[\left\{x\diamond y\in\mathbb{R} \mid x\in[x],y\in[y]\right\}\right],\\ \left[x\right]\diamond\varnothing&=&\varnothing.\end{split}\]

Dealing with closed intervals, most of the operations can rely on their bounds. It is for instance the case of addition, difference, union, etc.:

\[\begin{split}\begin{eqnarray} [x]+[y]&=&\left[\underline{x}+\underline{y},\overline{x}+\overline{y}\right],\\ \left[x\right]-\left[y\right]& = &\left[\underline{x}-\overline{y},\overline{x}-\underline{y}\right],\\ \left[x\right]\sqcup\left[y\right]& = &\left[\min\left(\underline{x},\underline{y}\right),\max\left(\overline{x},\overline{y}\right)\right],\\ \left[x\right]\cap\left[y\right]& = &\left[\max\left(\underline{x},\underline{y}\right),\min\left(\overline{x},\overline{y}\right)\right] \textrm{if} \max\left\{\underline{x},\underline{y}\right\}\leqslant\min\left\{\overline{x},\overline{y}\right\},\nonumber\\ ~ & = & \varnothing \textrm{ otherwise}. \end{eqnarray}\end{split}\]

Low-level libraries upon which Codac has been built provide functionalities for computing arithmetic on intervals, involving basic operations as well as non-linear functions. The following functions can be used:

Code

Meaning

sqr(x)

\([x]^2\)

sqrt(x)

\(\sqrt{[x]}\)

pow(x,n)

\([x]^n\)

pow(x,y)

\([x]^{[y]} = e^{[y]\log([x])}\)

root(x,n)

\(\sqrt[n]{[x]}\)

exp(x)

\(\exp([x])\)

log(x)

\(\log([x])\)

cos(x)

\(\cos([x])\)

sin(x)

\(\sin([x])\)

tan(x)

\(\tan([x])\)

acos(x)

\(\textrm{acos}([x])\)

asin(x)

\(\textrm{asin}([x])\)

atan(x)

\(\textrm{atan}([x])\)

atan2(y,x)

\(\textrm{atan2}([y],[x])\)

The use on intervals is transparent:

a = Interval(-2,4) * Interval(1,3)                              # a = [-6,12]
b = Interval(-2,4) & Interval(6,7)                              # b = [empty] (intersection)
c = max(Interval(2,7), Interval(1,9))                           # c = [2,9]
d = max(Interval.EMPTY_SET, Interval(1,2))                      # d = [empty]
e = Interval(-1,3) / Interval(0,oo)                             # e = [-oo,oo]
f = (Interval(1,2) * Interval(-1,3)) \
  + max(Interval(1,3) & Interval(6,7), Interval(1,2))           # f = [empty]

If intervals and boxes are used to handle static variables, tubes provide a way to deal with trajectories.