<s>
In	O
computer	B-General_Concept
science	I-General_Concept
,	O
the	O
min-conflicts	B-Application
algorithm	I-Application
is	O
a	O
search	B-Application
algorithm	I-Application
or	O
heuristic	B-Algorithm
method	O
to	O
solve	O
constraint	B-Application
satisfaction	I-Application
problems	I-Application
.	O
</s>
<s>
Given	O
an	O
initial	O
assignment	O
of	O
values	O
to	O
all	O
the	O
variables	O
of	O
a	O
constraint	B-Application
satisfaction	I-Application
problem	I-Application
,	O
the	O
algorithm	O
randomly	O
selects	O
a	O
variable	O
from	O
the	O
set	O
of	O
variables	O
with	O
conflicts	O
violating	O
one	O
or	O
more	O
its	O
constraints	O
.	O
</s>
<s>
Because	O
a	O
constraint	B-Application
satisfaction	I-Application
problem	I-Application
can	O
be	O
interpreted	O
as	O
a	O
local	B-Algorithm
search	I-Algorithm
problem	I-Algorithm
when	O
all	O
the	O
variables	O
have	O
an	O
assigned	O
value	O
(	O
called	O
a	O
complete	O
state	O
)	O
,	O
the	O
min	B-Application
conflicts	I-Application
algorithm	I-Application
can	O
be	O
seen	O
as	O
a	O
repair	O
heuristic	B-Algorithm
that	O
chooses	O
the	O
state	O
with	O
the	O
minimum	O
number	O
of	O
conflicts	O
.	O
</s>
<s>
input	O
:	O
console.csp	O
,	O
A	O
constraint	B-Application
satisfaction	I-Application
problem	I-Application
.	O
</s>
<s>
Use	O
a	O
greedy	B-Algorithm
algorithm	I-Algorithm
with	O
some	O
level	O
of	O
randomness	O
and	O
allow	O
variable	O
assignment	O
to	O
break	O
constraints	O
when	O
no	O
other	O
assignment	O
will	O
suffice	O
.	O
</s>
<s>
The	O
randomness	O
helps	O
min-conflicts	O
avoid	O
local	O
minima	O
created	O
by	O
the	O
greedy	B-Algorithm
algorithm	I-Algorithm
's	O
initial	O
assignment	O
.	O
</s>
<s>
In	O
fact	O
,	O
Constraint	B-Application
Satisfaction	I-Application
Problems	I-Application
that	O
respond	O
best	O
to	O
a	O
min-conflicts	O
solution	O
do	O
well	O
where	O
a	O
greedy	B-Algorithm
algorithm	I-Algorithm
almost	O
solves	O
the	O
problem	O
.	O
</s>
<s>
Map	B-Application
coloring	I-Application
problems	O
do	O
poorly	O
with	O
Greedy	B-Algorithm
Algorithm	I-Algorithm
as	O
well	O
as	O
Min-Conflicts	O
.	O
</s>
<s>
Although	O
Artificial	O
Intelligence	O
and	O
Discrete	O
Optimization	O
had	O
known	O
and	O
reasoned	O
about	O
Constraint	B-Application
Satisfaction	I-Application
Problems	I-Application
for	O
many	O
years	O
,	O
it	O
was	O
not	O
until	O
the	O
early	O
1990s	O
that	O
this	O
process	O
for	O
solving	O
large	O
CSPs	O
had	O
been	O
codified	O
in	O
algorithmic	O
form	O
.	O
</s>
<s>
Steven	O
Minton	O
and	O
Andy	O
Philips	O
analyzed	O
the	O
neural	O
network	O
algorithm	O
and	O
separated	O
it	O
into	O
two	O
phases	O
:	O
(	O
1	O
)	O
an	O
initial	O
assignment	O
using	O
a	O
greedy	B-Algorithm
algorithm	I-Algorithm
and	O
(	O
2	O
)	O
a	O
conflict	O
minimization	O
phases	O
(	O
later	O
to	O
be	O
called	O
"	O
min-conflicts	O
"	O
)	O
.	O
</s>
<s>
This	O
discovery	O
and	O
observations	O
led	O
to	O
a	O
great	O
amount	O
of	O
research	O
in	O
1990	O
and	O
began	O
research	O
on	O
local	B-Algorithm
search	I-Algorithm
problems	O
and	O
the	O
distinctions	O
between	O
easy	O
and	O
hard	O
problems	O
.	O
</s>
<s>
N-Queens	O
is	O
easy	O
for	O
local	B-Algorithm
search	I-Algorithm
because	O
solutions	O
are	O
densely	O
distributed	O
throughout	O
the	O
state	O
space	O
.	O
</s>
