Memoization


In computing, memoization or memoisation is an optimization technique used primarily to speed up computer programs by storing the results of expensive function calls and returning the cached result when the same inputs occur again. Memoization has also been used in other contexts, such as in simple mutually recursive descent parsing. Although related to caching, memoization refers to a specific case of this optimization, distinguishing it from forms of caching such as buffering or page replacement. In the context of some logic programming languages, memoization is also known as tabling.

Etymology

The term "memoization" was coined by Donald Michie in 1968 and is derived from the Latin word "memorandum", usually truncated as "memo" in American English, and thus carries the meaning of "turning a function into something to be remembered". While "memoization" might be confused with "memorization", "memoization" has a specialized meaning in computing.

Overview

A memoized function "remembers" the results corresponding to some set of specific inputs. Subsequent calls with remembered inputs return the remembered result rather than recalculating it, thus eliminating the primary cost of a call with given parameters from all but the first call made to the function with those parameters. The set of remembered associations may be a fixed-size set controlled by a replacement algorithm or a fixed set, depending on the nature of the function and its use. A function can only be memoized if it is referentially transparent; that is, only if calling the function has exactly the same effect as replacing that function call with its return value. While related to lookup tables, since memoization often uses such tables in its implementation, memoization populates its cache of results transparently on the fly, as needed, rather than in advance.
Memoization is a way to lower a function's time cost in exchange for space cost; that is, memoized functions become optimized for speed in exchange for a higher use of computer memory space. The time/space "cost" of algorithms has a specific name in computing: computational complexity. All functions have a computational complexity in time and in space.
Although a space–time tradeoff occurs, this differs from some other optimizations that involve time-space trade-off, such as strength reduction, in that memoization is a run-time rather than compile-time optimization. Moreover, strength reduction potentially replaces a costly operation such as multiplication with a less costly operation such as addition, and the results in savings can be highly machine-dependent, whereas memoization is a more machine-independent, cross-platform strategy.
Consider the following pseudocode function to calculate the factorial of n:
function factorial
if n is 0 then
return 1
else
return factorial times n
end if
end function
For every integer n such that n≥0, the final result of the function factorial is invariant; if invoked as x = factorial, the result is such that x will always be assigned the value 6. The non-memoized implementation above, given the nature of the recursive algorithm involved, would require n + 1 invocations of factorial to arrive at a result, and each of these invocations, in turn, has an associated cost in the time it takes the function to return the value computed. Depending on the machine, this cost might be the sum of:
  1. The cost to set up the functional call stack frame.
  2. The cost to compare n to 0.
  3. The cost to subtract 1 from n.
  4. The cost to set up the recursive call stack frame.
  5. The cost to multiply the result of the recursive call to factorial by n.
  6. The cost to store the return result so that it may be used by the calling context.
In a non-memoized implementation, every top-level call to factorial includes the cumulative cost of steps 2 through 6 proportional to the initial value of n.
A memoized version of the factorial function follows:
function factorial
if n is 0 then
return 1
else if n is in lookup-table then
return lookup-table-value-for-n
else
let x = factorial times n
store x in lookup-table in the nth slot
return x
end if
end function
In this particular example, if factorial is first invoked with 5, and then invoked later with any value less than or equal to five, those return values will also have been memoized, since factorial will have been called recursively with the values 5, 4, 3, 2, 1, and 0, and the return values for each of those will have been stored. If it is then called with a number greater than 5, such as 7, only 2 recursive calls will be made, and the value for 5! will have been stored from the previous call. In this way, memoization allows a function to become more time-efficient the more often it is called, thus resulting in eventual overall speed up.

Some other considerations

Functional programming

Memoization is heavily used in compilers for functional programming languages, which often use call by name evaluation strategy. To avoid overhead with calculating argument values, compilers for these languages heavily use auxiliary functions called thunks to compute the argument values, and memoize these functions to avoid repeated calculations.

Automatic memoization

While memoization may be added to functions internally and explicitly by a computer programmer in much the same way the above memoized version of factorial is implemented, referentially transparent functions may also be automatically memoized externally. The techniques employed by Peter Norvig have application not only in Common Lisp, but also in various other programming languages. Applications of automatic memoization have also been formally explored in the study of term rewriting and artificial intelligence.
In programming languages where functions are first-class objects, automatic memoization can be implemented by replacing a function with its calculated value once a value has been calculated for a given set of parameters. The function that does this value-for-function-object replacement can generically wrap any referentially transparent function. Consider the following pseudocode :
function memoized-call
if F has no attached array values then
allocate an associative array called values;
attach values to F;
end if;

if F.values is empty then
F.values = F;
end if;

return F.values;
end function
In order to call an automatically memoized version of factorial using the above strategy, rather than calling factorial directly, code invokes memoized-call. Each such call first checks to see if a holder array has been allocated to store results, and if not, attaches that array. If no entry exists at the position values, a real call is made to factorial with the supplied arguments. Finally, the entry in the array at the key position is returned to the caller.
The above strategy requires explicit wrapping at each call to a function that is to be memoized. In those languages that allow closures, memoization can be effected implicitly by a functor factory that returns a wrapped memoized function object. In pseudocode, this can be expressed as follows:
function construct-memoized-functor
allocate a function object called memoized-version;

let memoized-version be
if self has no attached array values then
allocate an associative array called values;
attach values to self;
end if;
if self.values is empty then
self.values = F;
end if;
return self.values;
end let;

return memoized-version;
end function
Rather than call factorial, a new function object memfact is created as follows:
memfact = construct-memoized-functor
The above example assumes that the function factorial has already been defined before the call to construct-memoized-functor is made. From this point forward, memfact is called whenever the factorial of n is desired. In languages such as Lua, more sophisticated techniques exist which allow a function to be replaced by a new function with the same name, which would permit:
factorial = construct-memoized-functor
Essentially, such techniques involve attaching the original function object to the created functor and forwarding calls to the original function being memoized via an alias when a call to the actual function is required, as illustrated below:
function construct-memoized-functor
allocate a function object called memoized-version;

let memoized-version be
if self has no attached array values then
allocate an associative array called values;
attach values to self;
allocate a new function object called alias;
attach alias to self;
self.alias = F;
end if;
if self.values is empty then
self.values = self.alias;
end if;
return self.values;
end let;

return memoized-version;
end function

Parsers

When a top-down parser tries to parse an ambiguous input with respect to an ambiguous context-free grammar, it may need an exponential number of steps to try all alternatives of the CFG in order to produce all possible parse trees. This eventually would require exponential memory space. Memoization was explored as a parsing strategy in 1991 by Peter Norvig, who demonstrated that an algorithm similar to the use of dynamic programming and state-sets in Earley's algorithm, and tables in the CYK algorithm of Cocke, Younger and Kasami, could be generated by introducing automatic memoization to a simple backtracking recursive descent parser to solve the problem of exponential time complexity. The basic idea in Norvig’s approach is that when a parser is applied to the input, the result is stored in a memotable for subsequent reuse if the same parser is ever reapplied to the same input. Richard Frost also used memoization to reduce the exponential time complexity of parser combinators, which can be viewed as “Purely Functional Top-Down Backtracking” parsing technique. He showed that basic memoized parser combinators can be used as building blocks to construct complex parsers as executable specifications of CFGs. It was again explored in the context of parsing in 1995 by Johnson and Dörre. In 2002, it was examined in considerable depth by Ford in the form called packrat parsing.
In 2007, Frost, Hafiz and Callaghan described a top-down parsing algorithm that uses memoization for refraining redundant computations to accommodate any form of ambiguous CFG in polynomial time for left-recursive grammars and Θ. Their top-down parsing algorithm also requires polynomial space for potentially exponential ambiguous parse trees by 'compact representation' and 'local ambiguities grouping'. Their compact representation is comparable with Tomita’s compact representation of bottom-up parsing. Their use of memoization is not only limited to retrieving the previously computed results when a parser is applied to a same input position repeatedly ; it is specialized to perform the following additional tasks:
Frost, Hafiz and Callaghan also described the implementation of the algorithm in PADL’08 as a set of higher-order functions in Haskell, which enables the construction of directly executable specifications of CFGs as language processors. The importance of their polynomial algorithm’s power to accommodate ‘any form of ambiguous CFG’ with top-down parsing is vital with respect to the syntax and semantics analysis during natural language processing. The site has more about the algorithm and implementation details.
While Norvig increased the power of the parser through memoization, the augmented parser was still as time complex as Earley's algorithm, which demonstrates a case of the use of memoization for something other than speed optimization. Johnson and Dörre demonstrate another such non-speed related application of memoization: the use of memoization to delay linguistic constraint resolution to a point in a parse where sufficient information has been accumulated to resolve those constraints. By contrast, in the speed optimization application of memoization, Ford demonstrated that memoization could guarantee that parsing expression grammars could parse in linear time even those languages that resulted in worst-case backtracking behavior.
Consider the following grammar:
S → |
A → X
B → X b
X → x
|
This grammar generates one of the following three variations of string: xac, xbc, or xbd Next, consider how this grammar, used as a parse specification, might effect a top-down, left-right parse of the string xxxxxbd:
The key concept here is inherent in the phrase again descends into X. The process of looking forward, failing, backing up, and then retrying the next alternative is known in parsing as backtracking, and it is primarily backtracking that presents opportunities for memoization in parsing. Consider a function RuleAcceptsSomeInput, where the parameters are as follows:
Let the return value of the function RuleAcceptsSomeInput be the length of the input accepted by Rule, or 0 if that rule does not accept any input at that offset in the string. In a backtracking scenario with such memoization, the parsing process is as follows:
In the above example, one or many descents into X may occur, allowing for strings such as xxxxxxxxxxxxxxxxbd. In fact, there may be any number of x's before the b. While the call to S must recursively descend into X as many times as there are x's, B will never have to descend into X at all, since the return value of RuleAcceptsSomeInput will be 16.
Those parsers that make use of syntactic predicates are also able to memoize the results of predicate parses, as well, thereby reducing such constructions as:
S → ? A
A → /* some rule */
to one descent into A.
If a parser builds a parse tree during a parse, it must memoize not only the length of the input that matches at some offset against a given rule, but also must store the sub-tree that is generated by that rule at that offset in the input, since subsequent calls to the rule by the parser will not actually descend and rebuild that tree. For the same reason, memoized parser algorithms that generate calls to external code when a rule matches must use some scheme to ensure that such rules are invoked in a predictable order.
Since, for any given backtracking or syntactic predicate capable parser not every grammar will need backtracking or predicate checks, the overhead of storing each rule's parse results against every offset in the input may actually slow down a parser. This effect can be mitigated by explicit selection of those rules the parser will memoize.