<s>
MTD(f )	O
is	O
an	O
alpha-beta	B-Algorithm
game	O
tree	O
search	O
algorithm	O
modified	O
to	O
use	O
‘	O
zero-window	O
’	O
initial	O
search	O
bounds	O
,	O
and	O
memory	O
(	O
usually	O
a	O
transposition	B-General_Concept
table	I-General_Concept
)	O
to	O
reuse	O
intermediate	O
search	O
results	O
.	O
</s>
<s>
MTD(f )	O
was	O
introduced	O
in	O
1994	O
and	O
largely	O
supplanted	O
NegaScout	B-General_Concept
(	O
PVS	O
)	O
,	O
the	O
previously	O
dominant	O
search	O
paradigm	O
for	O
chess	B-Application
,	O
checkers	O
,	O
othello	O
and	O
other	O
game	O
automatons	O
.	O
</s>
<s>
MTD(f )	O
was	O
first	O
described	O
in	O
a	O
University	O
of	O
Alberta	O
Technical	O
Report	O
authored	O
by	O
Aske	O
Plaat	O
,	O
Jonathan	O
Schaeffer	O
,	O
Wim	O
Pijls	O
,	O
and	O
Arie	O
de	O
Bruin	O
,	O
which	O
would	O
later	O
receive	O
the	O
ICCA	O
Novag	O
Best	O
Computer	B-Application
Chess	I-Application
Publication	O
award	O
for	O
1994/1995	O
.	O
</s>
<s>
The	O
algorithm	O
MTD(f )	O
was	O
created	O
out	O
of	O
a	O
research	O
effort	O
to	O
understand	O
the	O
SSS*	B-Algorithm
algorithm	O
,	O
a	O
best-first	O
search	O
algorithm	O
invented	O
by	O
George	O
Stockman	O
in	O
1979	O
.	O
</s>
<s>
SSS*	B-Algorithm
was	O
found	O
to	O
be	O
equivalent	O
to	O
a	O
series	O
of	O
Alpha-beta	B-Algorithm
pruning|	O
alpha-beta	B-Algorithm
calls	O
,	O
provided	O
that	O
alpha-beta	B-Algorithm
used	O
storage	O
,	O
such	O
as	O
a	O
transposition	B-General_Concept
table	I-General_Concept
.	O
</s>
<s>
MTD(f )	O
derives	O
its	O
efficiency	O
by	O
only	O
performing	O
zero-window	O
alpha-beta	B-Algorithm
searches	I-Algorithm
,	O
with	O
a	O
previously	O
determined	O
"	O
good	O
"	O
bound	O
(	O
i.e.	O
</s>
<s>
A	O
transposition	B-General_Concept
table	I-General_Concept
stores	O
and	O
retrieves	O
the	O
previously	O
searched	O
portions	O
of	O
the	O
tree	O
in	O
memory	O
to	O
reduce	O
the	O
overhead	O
of	O
re-exploring	O
parts	O
of	O
the	O
search	O
tree	O
.	O
</s>
<s>
An	O
iterative	B-Algorithm
deepening	I-Algorithm
depth-first	I-Algorithm
search	I-Algorithm
could	O
be	O
done	O
by	O
calling	O
multiple	O
times	O
with	O
incrementing	O
and	O
providing	O
the	O
best	O
previous	O
result	O
in	O
.	O
</s>
<s>
MTD(f )	O
depends	O
on	O
a	O
transposition	B-General_Concept
table	I-General_Concept
to	O
perform	O
efficiently	O
.	O
</s>
<s>
For	O
this	O
reason	O
some	O
chess	B-Application
engines	I-Application
have	O
not	O
switched	O
to	O
MTD(f )	O
.	O
</s>
<s>
In	O
tests	O
with	O
tournament-quality	O
programs	O
such	O
as	O
Chinook	B-General_Concept
(	O
checkers	O
)	O
,	O
Phoenix	O
(	O
chess	B-Application
)	O
,	O
and	O
Keyano	O
(	O
Othello	O
)	O
,	O
the	O
MTD(f )	O
algorithm	O
outperformed	O
all	O
other	O
search	O
algorithms	O
.	O
</s>
<s>
Recent	O
algorithms	O
like	O
Best	B-Algorithm
Node	I-Algorithm
Search	I-Algorithm
are	O
suggested	O
to	O
outperform	O
MTD(f )	O
.	O
</s>
