What are the different techniques for memoization

Question

I know of this one http://onjava.com/pub/a/onjava/2003/08/20/memoization.html but is there anything else?

Answer 1

Yes. Use caches from Guava.

Example:

import java.math.BigInteger;

import com.google.common.base.Preconditions;
import com.google.common.cache.CacheBuilder;
import com.google.common.cache.CacheLoader;
import com.google.common.cache.LoadingCache;

public class Fibonacci {
    private static final LoadingCache<Integer, BigInteger> CACHE
            = CacheBuilder.newBuilder().build(CacheLoader.from(Fibonacci::fib));

    public static BigInteger fib(int n) {
        Preconditions.checkArgument(n >= 0);
        switch (n) {
        case 0:
            return BigInteger.ZERO;
        case 1:
            return BigInteger.ONE;
        default:
            return CACHE.getUnchecked(n - 1).add(CACHE.getUnchecked(n - 2));
        }
    }
}

Answer 2

To memoize functions without parameters, use Guava's Suppliers.memoize(Supplier). For functions with parameters, use CacheBuilder.build(CacheLoader) with parameter value objects as keys.

Answer 3

Memoization is also easy with plain simple typesafe Java.

You can do it from scratch with the following reusable classes.

I use these as caches whose lifespan are the request on a webapp.

Of course use the Guava MapMaker if you need an eviction strategy or more features like synchronization.

If you need to memoize a method with many parameters, just put the parameters in a list with both techniques, and pass that list as the single parameter.

abstract public class Memoize0<V> {
    //the memory
    private V value;
    public V get() {
        if (value == null) {
            value = calc();
        }
        return value;
    }
    /**
     * will implement the calculation that 
     * is to be remembered thanks to this class
     */
    public abstract V calc();
}

abstract public class Memoize1<P, V> {
    //The memory, it maps one calculation parameter to one calculation result
    private Map<P, V> values = new HashMap<P, V>();

    public V get(P p) {
        if (!values.containsKey(p)) {
            values.put(p, calc(p));
        }
        return values.get(p);
    }

    /**
     * Will implement the calculations that are
     * to be remembered thanks to this class
     * (one calculation per distinct parameter)
     */
    public abstract V calc(P p);
 }

And this is used like this

    Memoize0<String> configProvider = new Memoize0<String>() {
        @Override
        public String calc() {
            return fetchConfigFromVerySlowDatabase();
        }
    };
    final String config = configProvider.get();

    Memoize1<Long, String> usernameProvider = new Memoize1<Long, String>() {
        @Override
        public String calc(Long id) {
            return fetchUsernameFromVerySlowDatabase(id);
        }
    };
    final String username = usernameProvider.get(123L);