1

Hamilton Circuit
Topics in Contemporary Mathematics
MA 103
Summer II, 2013

Hamilton Paths and Hamilton Circuits

In Euler paths and Euler circuits, the game was to find paths or circuits
that include every edge of the graph once (and only once).

In Hamilton paths and Hamilton circuits, the game is to find paths

and circuits that include every vertex of the graph once and only once.

HAMILTON PATHS & CIRCUITS

A Hamilton path in a graph is a path

that includes each vertex of the graph
once and only once. !
!A Hamilton circuit is a circuit that
includes each vertex of the graph
once and only once. (At the end, of
course, the circuit must return to the
starting vertex.) !
!

Example 1

Hamilton versus Euler

The figure shows a graph that

(1)has Euler circuits (the vertices are
all even) and
(2) has Hamilton circuits.
One such Hamilton circuit is A,
F, B, C, G, D, E, A there are
plenty more.

Example 1

Hamilton versus Euler

Note that if a graph has a Hamilton circuit, then it

automatically has a Hamilton path(the Hamilton circuit
can always be truncated into a Hamilton path by
dropping the last vertex of the circuit.)
For example, the Hamilton circuit A, F, B, C, G, D, E, A
can be truncated into the Hamilton path A, F, B, C, G, D, E.
Contrast this with the mutually exclusive relationship
between Euler circuits and paths: If a graph has an
Euler circuit it cannot have an Euler path and vice versa.

Example 1

Hamilton versus Euler

This figure shows a graph that

(1)has no Euler circuits but does have Euler paths (for
example C, D, E, B, A, D) and
(2)has no Hamilton circuits (sooner or later you have to go to
C, and then you are stuck) but does have Hamilton paths (for
example, A, B, E, D, C).
This illustrates that a graph can
have a Hamilton path but no
Hamilton circuit!

Example 1

Hamilton versus Euler

This figure shows a graph that

(1)has neither Euler circuits nor paths (it has four odd
vertices) and
(2) has Hamilton circuits (for example A, B, C, D, E, A
there are plenty more)
and consequently has
Hamilton paths (for example, A,
B, C, D, E).

Example 1

Hamilton versus Euler

This figure shows a graph that

(1) has no Euler circuits but has
Euler paths (F and G are the two
odd vertices) and
(2) has neither Hamilton circuits
nor Hamilton paths.

Example 1

Hamilton versus Euler

This figure shows a graph

that
(1) has neither Euler circuits nor
Euler paths (too many odd
vertices) and
(2) has neither Hamilton circuits
nor Hamilton paths.

10

Euler versus Hamilton

The lesson of Example 1 is that the existence of an
Euler path or circuit in a graph tells us nothing about
the existence of a Hamilton path or circuit in that
graph.

11

Euler versus Hamilton

Unlike Euler circuit and path, there exist no Hamilton
circuit and path theorems for determining if a graph has
a Hamilton circuit, a Hamilton path, or neither.
Determining when a given graph does or does not
have a Hamilton circuit or path can be very easy, but
it also can be very hardit all depends on the graph.

12

How Many Hamilton Circuits?

Sometimes the question, Does the graph have a
Hamilton circuit? has an obvious yes answer, and the
more relevant question turns out to be, How many
different Hamilton circuits does it have?

In this section we will answer this question for an

important family of graphs called complete graphs.

13

Complete Graph
A graph with N vertices in which every pair of
distinct vertices is joined by an edge is called
a complete graph on N vertices and is denoted by
the symbol KN .

14

How Many Hamilton Circuits?

One of the key properties of KN is that every vertex has
degree N 1.
This implies that the sum of the degrees of all the
vertices is N(N 1), and it follows from Eulers sum
of degrees theorem that the number of edges in KN is
N(N 1)/2.
For a graph with N vertices and no multiple edges or
loops, N(N 1)/2 is the maximum number of edges
possible, and this maximum can only occur when the
graph is KN.

15

NUMBER OF EDGES IN KN

KN has N(N 1)/2 edges. !

!Of all graphs with N vertices and no
multiple edges or loops, KN has the
most edges. !
!

Because KN has a complete set of edges (every

vertex is connected to every other vertex), it
also has a complete set of Hamilton circuits
you can travel the vertices in any sequence you
choose and you will not get stuck. !

16

Example 2

Hamilton Circuits in K4

If we travel the four ver.ces of K4 in an arbitrary order,

we get a Hamilton path. For example, C, A, D, B is a
Hamilton path [Fig. (a)]; D, C, A, B is another one [Fig.
(b)]; and so on.

17

Example 2

Hamilton Circuits in K4

Each of these Hamilton paths can be closed into a

Hamilton circuitthe path C, A, D, B begets the circuit C,
A, D, B, C [Fig. (c)]; the path D, C, A, B begets the circuit
D, C, A, B, D [Fig. (d)]; and so on.

18

Example 2

Hamilton Circuits in K4

It looks like we have an abundance of Hamilton

circuits, but it is important to remember that the
same Hamilton circuit can be wri@en in many
ways. For example, C, A, D, B, C is the same circuit
as A, D, B, C, A the gure describes either one
the only dierence is that in
the rst case we used C as
the reference point; in the
second case we used A as
the reference point.

19

Example 2

Hamilton Circuits in K4

There are two additional sequences that describe this

same Hamilton circuit: D, B, C, A, D (with reference point
D) and B, C, A, D, B (with reference point B). Taking all
this into account, there are six different Hamilton
circuits in K4, as shown in the
Table on the next slide (the
table also shows the four
different ways each circuit can
be written).

20

Example 2

Hamilton Circuits in K4

21

Example 2

Hamilton Circuits in K5

Lets try to list all the Hamilton circuits in K5.

For simplicity, we will write each circuit just once, using a common
reference point say A. (As long as we are consistent, it doesnt really
matter which reference point we pick.)
Each of the Hamilton circuits will be described by a sequence that starts
and ends with A, with the letters B, C, D, and E sandwiched in between
in some order.
There are 4 3 2 1 = 24 different ways to shuffle the letters B, C,
D, and E, each producing a different Hamilton circuit.

22

Example 2

Hamilton Circuits in K5

The complete list of the 24 Hamilton circuits in K5 is

shown in the table on the next slide. The table is laid out
so that each of the circuits in the table is directly
opposite its mirror-image circuit (the circuit with
ver.ces listed in reverse order).

Although they are close rela.ves, a circuit and its mirror
image are not considered the same circuit.

23

Example 2
Hamilton
Circuits in K5

The complete list of the

24 Hamilton circuits in
K5.

24

Example 2

Hamilton Circuits in K5

Here are three of the 24 Hamilton circuits in K5.

25

Number of Hamilton Circuits

What is the number of Hamilton circuits in KN ? boils
down to the equivalent ques.on,
How many dierent ways are there to rearrange the
(N 1) verMces?

The answer is given by the number
1 2 3 (N 1), called the factorial of (N 1)
and wriRen as (N 1)! for short.

26

NUMBER OF HAMILTON
CIRCUITS IN KN
! KN has N(N 1)/2 edges. !
!Of all graphs with N vertices and no
multiple edges or loops, KN has the
most edges. !

27

Number of Hamilton Circuits

The table on the next slide shows the number

of Hamilton circuits in complete graphs with up
to N = 20 ver.ces.

No.ce that as we increase the number of
ver.ces, the number of Hamilton circuits in the
complete graph goes through the roof.

Even a relaMvely small graph such as has K8
more than ve thousand Hamilton circuits.
Double the number of verMces to K16 and the
number of Hamilton circuits exceeds 1.3 trillion.

28

Traveling Salesman Problem

Any problem that shares these common elements:
1)a traveler,
2)a set of sites,
3)a cost function for travel between pairs of sites,
4)a need to tour all the sites, and
5)a desire to minimize the total cost of the tour
is known as a traveling salesman problem, or
TSP. !

Routing School Buses

A school bus (the traveler) picks up children in the
morning and drops them o at the end of the day at
designated stops (the sites).

On a typical school bus route there may be 20 to 30 such
stops. With school buses, total .me on the bus is always
the most important variable (students have to get to
school on .me), and there is a known .me of travel (the
cost) between any two bus stops.

Since children must be picked up at every bus stop, a
tour of all the sites (starMng and ending at the school) is
required. Since the bus repeats its route every day
during the school year, nding an opMmal tour is crucial.

Delivering Packages
Package delivery companies such as UPS and FedEx deal
with TSPs on a daily basis.

Each truck is a traveler that must deliver packages to a
specic list of delivery des.na.ons (the sites). The travel
.me between any two delivery sites (the cost) is known
or can be es.mated. Each day the truck must deliver to
all the sites on its list, so a tour is an implied part of the
requirements. Since one can assume that the driver
would rather be home than out delivering packages, an
op.mal tour is a highly desirable goal.

Modeling a TSP
Every TSP can be modeled by a weighted graph, that is, a
graph such that there is a number associated with each
edge (called the weight of the edge).


The beauty of this approach is that the model always has
the same structure: The verMces of the graph are the
sites of the TSP, and there is an edge between X and Y if
there is a direct link for the traveler to travel from site X
to site Y. Moreover, the weight of the edge XY is the cost
of travel between X and Y.

Modeling a TSP
In this seXng a tour is a Hamilton circuit of the graph,
and an op.mal tour is the Hamilton circuit of least total
weight.

In all the applica.ons and examples we will be
considering, we will make the assumpMon that there is
an edge connecMng every pair of sites, which implies that
the underlying graph model is always a complete
weighted graph. The following is a summary of the
preceding observa.ons.

GRAPH MODEL OF A TRAVELING

SALESMAN PROBLEM

Sites g vertices of the graph. !

!Costs g weights of the edges.
!Tour g Hamilton circuit.
!Optimal tour g Hamilton circuit of
least total cost !
!

Example 3

A Tour of Five Cities

Willy a traveling salesman is pondering his upcoming

sales trip. He wants to tour ve ci.es; A, B, C, D and E.
The one-way airfares
between any two ci.es
are shown on the graph.

Wiley would like to nd an
op.mal tour.

Traveling Salesman Problems

We will explore two strategies for solving
traveling salesman problems:

1. Exhaus1ve Search

2. Go Cheap

STRATEGY 1
(EXHAUSTIVE SEARCH)
Make a list of all possible Hamilton
circuits. For each circuit in the list,
calculate the weight of the circuit. From
all the circuits, choose a circuit with
least total weight. This is your optimal
tour.

Example 4

A Tour of Five Cities: Part

2 - STRATEGY 1

The table on the next slide shows a detailed

implementa.on of this strategy.

The 24 Hamilton circuits are split into two columns
consis.ng of circuits and their mirror images.

The total costs for the circuits are shown in the middle
column of the table. There are two opMmal tours, with a
total cost of $676 A, D, B, C ,E, A and its mirror image
A, E, C, B, D, A.

Example 4

A Tour of Five Cities: Part

2 - STRATEGY 1

The first group of 12 Hamilton circuits:

Example 4

A Tour of Five Cities: Part

2 - STRATEGY 1

The second group of 12 Hamilton circuits:

Example 4

A Tour of Five Cities: Part

2 - STRATEGY 1

There are always going to be at least two op.mal tours,

since the mirror image of an opMmal tour is also opMmal.
Either one of them can be used for a soluMon.

STRATEGY 2
(GO CHEAP)
Start from the home city. From there go
to the city that is the cheapest to get to.
From each new city go to the next new
city that is cheapest to get to. When
there are no more new cities to go to, go
back home.

Example 5

A Tour of Five Cities: Part

2 - Strategy 2

We assume that Willy lives in city A. In Willy case, the

strategy works like this: Start at A. From A Willy goes to
C (the cheapest place he can y to from A).
From C to E, E to D, and
from D the last new city
to go to is B. From B
Willy has to return
home to A. The total
cost of this tour is $773.

Example 5

A Tour of Five Cities: Part

2 - Strategy 2

The go cheap strategy takes a lot less work than the

exhaus.ve search strategy, but there is a hitch the
cost of the tour we get is $773, which is $97 more than
the opMmal tour found under the exhausMve search
strategy.

Algorithms for Solving TSPs

In this sec.on we will look at the two strategies we
informally developed in connec.on with Willys sales
trips and recast them in the language of algorithms.

The exhaus.ve search strategy can be formalized into
an algorithm generally known as the brute-force
algorithm;

the go cheap strategy can be formalized into an
algorithm known as the nearest-neighbor algorithm.

ALGORITM 1: THE BRUTE

FORCE ALGORITHM
!Step 1. Make a list of all the possible
Hamilton circuits of the graph. Each
of these circuits represents a tour of
the vertices of the graph.!
!Step 2. For each tour calculate its
weight (i.e., add the weights of all the
edges in the circuit).
!Step 3. Choose an optimal tour (there
is always more than one optimal tour
to choose from!). !

ALGORITM 2: THE NEAREST

NEIGHBOR ALGORITHM

Start: Start at the designated starting

vertex.If there is no designated
starting vertex pick any vertex.

First step: From the starting vertex

go to its nearest neighbor(i.e.,the
vertex for which the corresponding
edge has the smallest weight).!

ALGORITM 2: THE NEAREST

NEIGHBOR ALGORITHM

Middle steps: From each vertex go to

its nearest neighbor, choosing only
among the vertices that havent been
yet visited. (If there is more than one
nearest neighbor choose among them
at random.) Keep doing this until all the
vertices have been visited.
!Last step: From the last vertex return
to the starting vertex.
!

Pros and Cons

The posiMve aspect of the brute-force algorithm is that
it is an opMmal algorithm.
(An op4mal algorithm is an algorithm that, when
correctly implemented, is guaranteed to produce an
op.mal solu.on.)

In the case of the brute-force algorithm, we know we
are geXng an op.mal solu.on because we are
choosing from among all possible tours.

Pros and Cons

The nega.ve aspect of the brute-force algorithm is the
amount of eort that goes into implemen.ng the
algorithm, which is (roughly) propor.onal to the
number of Hamilton circuits in the graph.

Practical Terms of the Pros and Cons

Lets start with human computa.on. To solve a TSP for
a graph with 10 verMces , the brute-force algorithm
requires checking 362,880 Hamilton circuits.

To do this by hand, even for a fast and clever human, it
would take over 1000 hours. Thus, at N =10 we are
already beyond the limit of what can be considered
reasonable human eort.

Using a Computer
Even with the worlds best
technology on our side, we very
quickly reach the point beyond
which using the brute-force
algorithm is completely
unrealisMc.

Superhero is the fastest
supercomputer on the planet.

Using a Computer
The brute-force algorithm is a classic example of what is
formally known as an inecient algorithm
an algorithm for which the number of steps needed to
carry it out grows disproporMonately with the size of the
problem.

The trouble with inecient algorithms is that they are of
limited prac.cal usethey can realis.cally be carried out
only when the problem is small.

Pros and Cons - Nearest Neighbor

We hop from vertex to vertex using a simple criterion:
Choose the next available nearest vertex and go for it.
Lets call the process of checking among the available
ver1ces and nding the nearest one a single
computa1on.

Then, for a TSP with N = 5 we need to perform 5
computa.ons. What happens when we double, the
number of ver.ces to
N = 10? We now have to perform 10 computa.ons. For
N = 30, we perform 30 computa.ons.

Pros and Cons - Nearest Neighbor

We can summarize the above observa.ons by saying
that the nearest-neighbor algorithm is an ecient
algorithm.

Roughly speaking, an ecient algorithm is an algorithm
for which the amount of computa.onal eort required
to implement the algorithm grows in some reasonable
propor.on with the size of the input to the problem.

Pros and Cons - Nearest Neighbor

The main problem with the nearest-neighbor algorithm
is that it is not an opMmal algorithm. In Example 5, the
nearest-neighbor tour had a cost of $773, whereas the
opMmal tour had a cost of $676.

In absolute terms the nearest-neighbor tour is o by $97
(this is called the absolute error). A be@er way to
describe how far o this tour is from the op.mal tour
is to use the concept of relaMve error. In this example
the absolute error is $97 out of $676, giving a rela.ve
error of $97/$676 = 0.1434911 14.35%.

In general, for any tour that might be proposed as a

solu.on to a TSP, we can nd its rela1ve error as
follows:

RELATIVE ERROR
OF A TOUR (e)

cost of tour
(
=

cost of optimal tour

cost of optimal tour

Relative Error - Good or Not Good

It is customary to express the rela.ve error as a
percentage (usually rounded to two decimal places).

By using the no.on of rela.ve error, we can
characterize the op.mal tours as those with rela.ve
error of 0%.

All other tours give approximate solu.ons, the
rela.ve merits of which we can judge by their
respec.ve rela.ve errors: tours with small rela.ve
errors are good, and tours with large rela.ve errors
are not good.

Repetitive Nearest-Neighbor Algorithm

As one might guess, the repe..ve nearest-neighbor
algorithm is a varia.on of the nearest-neighbor
algorithm in which we repeat several .mes the en.re
nearest- neighbor circuit-building process.

Why would we want to do this? The reason is that the
nearest-neighbor tour depends on the choice of the
star.ng vertex. If we change the star.ng vertex, the
nearest-neighbor tour is likely to be dierent, and, if we
are lucky, be@er.

Repetitive Nearest-Neighbor Algorithm

Since nding a nearest-neighbor tour is an ecient
process, it is not unreasonable to repeat the process
several .mes, each .me star.ng at a dierent vertex of
the graph.

In this way, we can obtain several dierent nearest-
neighbor solu.ons, from which we can then pick the
best.

Repetitive Nearest-Neighbor Algorithm

But what do we do with a tour that starts somewhere
other than the vertex we really want to start at?

Thats not a problem. Remember that once we have a
circuit, we can start the circuit anywhere we want.

In fact, in an abstract sense, a circuit has no star.ng or
ending point.

Example 6 A Tour of Five Cities: Part 3

In Example 5, we computed the nearest-neighbor tour
with A as the starMng vertex, and we got A, C, E, D, B, A
with a total cost of $773.

Example 6

A Tour of Five Cities: Part 3

But if we use B as the star.ng vertex, the nearest-

neighbor tour takes us from B to C, then to A, E, D, and
back to B, with a total cost of $722. Well, that is
certainly an improvement!
As for Willywho must
start and end his trip at
Athis very same tour
would take the form
A, E, D, B, C, A.

Example 6

A Tour of Five Cities: Part 3

The process is once again repeated using C, D, and E as

the starMng verMces.

When the star.ng point is C, we get the nearest-
neighbor tour C, A, E, D, B, C (total cost is $722);

when the star.ng point is D, we get the nearest-
neighbor tour D, B, C, A, E, D (total cost is $722); and

when the star.ng point is E, we get the nearest-
neighbor tour E, C, A, D, B, E (total cost is $741).

Example 6

A Tour of Five Cities: Part 3

Among all the nearest-neighbor tours we pick the

cheapest one A, E, D, B, C, A with a cost of $722.

ALGORITM 3: THE REPETITIVE

NEAREST NEIGHBOR ALGORITHM
!Let X be any vertex. Find the nearestneighbor tour with X as the starting vertex,
and calculate the cost of this tour. !
!Repeat the process with each of the other
vertices of the graph as the starting vertex.
!Of the nearest-neighbor tours thus
obtained, choose one with least cost. If
necessary, rewrite the tour so that it starts
at the designated starting vertex. We will
call this tour the repetitive nearest-neighbor
tour.

Cheapest-Link Algorithm
The idea behind the cheapest-link algorithm is to piece
together a tour by picking the separate links (i.e., legs)
of the tour on the basis of cost.

It doesnt ma@er if in the intermediate stages the links
are all over the place if we are careful at the end, they
will all come together and form a tour.

This is how it works: Look at the enMre graph and choose
the cheapest edge of the graph, wherever that edge may
be. Once this is done, choose the next cheapest edge of
the graph, wherever that edge may be.

Cheapest-Link Algorithm
Dont worry if that edge is not adjacent to the rst edge.
Con.nue this way, each .me choosing the cheapest
edge available but following two rules:
(1) Do not allow a circuit to form except at the very end,
and
(2) do not allow three edges to come together at a
vertex.

A viola.on of either of these two rules will prevent
forming a Hamilton circuit. Conversely, following these
two rules guarantees that the end result will be a
Hamilton circuit.

Example 7 A Tour of Five Cities: Part 4

Among all the edges of the graph, the cheapest link is
edge AC, with a cost of $119. For the purposes of
recordkeeping,
we will tag this edge
as taken by marking it
in red as shown.

Example 7

A Tour of Five Cities: Part 4

The next step is to scan the en.re graph again and

choose the cheapest link available, in this case edge CE
with a cost of $120.
Again, we mark it in red,
to indicate it is taken.

Example 7

A Tour of Five Cities: Part 4

The next cheapest link available is edge BC ($121), but

we should not choose BCwe would have three edges
coming out of vertex
C and this would
prevent us from forming
a circuit. This .me we
put a do not use mark
on edge BC.

Example 7

A Tour of Five Cities: Part 4

The next cheapest link available is AE ($133). But we

cant take AE eitherthe ver.ces A, C, and E would form
a small circuit, making it
impossible to form a
Hamilton circuit at the
end. So we put a do
not use mark on AE.

Example 7

A Tour of Five Cities: Part 4

The next cheapest link available is BD ($150). Choosing

BD would not violate either of the two rules, so we can
add it to our budding circuit and mark it in red.

Example 7

A Tour of Five Cities: Part 4

The next cheapest link available is AD ($152), and it

works just ne.

Example 7

A Tour of Five Cities: Part 4

At this point, we have only one way to close up the

Hamilton circuit, edge BE, as shown.

Example 7

A Tour of Five Cities: Part 4

The cheapest-link tour shown could have any star.ng

vertex we want. Since Willy lives at A, we describe it as
A, C, E, B, D, A. The total cost of this tour is $741, which
is a li@le be@er than the
nearest-neighbor tour
but not as good as the
repe..ve nearest-
neighbor tour.

ALGORITM 4: THE CHEAPEST

LINK ALGORITHM

Step 1. Pick the cheapest link (i.e.,

edge with smallest weight) available.
(In case of a tie, pick one at random.)
Mark it (say in red).!

Step 2. Pick the next cheapest link

available and mark it.!

ALGORITM 4: THE CHEAPEST

LINK ALGORITHM

Step 3, 4, , N 1. Continue picking

and marking the cheapest unmarked
link available that does not
(a) close a circuit or
(b) create three edges coming out of
a single vertex!
!Step N. Connect the last two vertices
to close the red circuit. This circuit
gives us the cheapest-link tour. !
!

Example 8

Roving the Red Planet

The gure shows seven sites on Mars iden.ed as

par.cularly interes.ng sites for geological explora.on.
Our job is to nd an op.mal tour for a rover that will
land at A, visit all the sites to collect rock samples, and
at the end return to A.

Example 8
The distances
between sites (in
miles) are given
in the graph
shown.

Roving the Red Planet

Example 8

Roving the Red Planet

Lets look at some of the approaches one might use to

tackle this problem.
Brute force: The brute-force algorithm would require us
to check and compute 6! = 720 dierent tours. We will
pass on that idea for now.

Cheapest link: The cheapest-link algorithm is a
reasonable algorithm to use not trivial but not too hard
either. A summary of the steps is shown in the table on
the next slide.

Example 8

Roving the Red Planet

The cheapest-link tour (A, P, W, H, G, N, I, A with a total

length of 21,400 miles) is shown.

Example
Here is the
cheapest-link tour
again (A, P, W, H,
G, N, I, A with a
total length of
21,400 miles) is
shown.

Roving the Red Planet

Example

Roving the Red Planet

Nearest neighbor: The nearest-neighbor algorithm is the

simplest of all the algorithms we learned. StarMng from
A we go to P, then to W, then to H, then to G, then to I,
then to N, and nally back to A.

The nearest-neighbor tour (A, P, W, H, G, I, N, A with a
total length of 20,100 miles) is shown on the next two
slides. (We know that we can repeat this method using
dierent star.ng points, but we wont bother with that
at this .me.)

Example

Roving the Red Planet

Nearest neighbor: A, P, W, H, G, I, N, A with a total

length of 20,100 miles.

Example
Nearest
neighbor:
A, P, W, H, G, I, N,
A with a total
length of 20,100
miles.

Roving the Red Planet

Surprise One
The rst surprise in in the last example is that the
nearest-neighbor algorithm gave us a beRer tour than
the cheapest-link algorithm.

Some.mes the cheapest-link algorithm produces a
be@er tour than the nearest-neighbor algorithm, but just
as onen, its the other way around.

The two algorithms are dierent but of equal standing
neither one can be said to be superior to the other one in
terms of the quality of the tours it produces.

Surprise Two
The second surprise is that the nearest-neighbor tour A,
P, W, H, G, I, N, A turns out to be an opMmal tour. (This
can be veried using a computer and the brute-force
algorithm.)

Essen.ally, this means that in this par.cular example,
the simplest of all methods happens to produce the
op.mal answera nice turn of events. Too bad we cant
count on this happening on a more consistent basis!