dtonhofer

dtonhofer

Functional Programming in Java, Second Edition: Chapter 8: Hard-to-understand "Memoizer" can be made easy-to-understand by adding an "intermediate step" explainer

I had real trouble understanding the “memoizer”, I suppose Java syntax does not help in thinking about what should be a one-liner in Lambda calculus.

But after a couple of hours of thinking, it occurred to me that the “memoizing” code is just the end result of four simple transformations of the non-memoized code.

Suggesting to extend the text to explain it that way.

Here they are, based on the book’s code with some renaming of methods and parameters to make them more meaningful (at least to me):

The code below does not come with runnable code, which I will post separately.

RodCuttingOptimizer.java

package chapter8.rodcutting.book;

import java.util.Collections;
import java.util.HashMap;
import java.util.Map;
import java.util.function.BiFunction;
import java.util.function.Function;
import java.util.stream.IntStream;

class RodCuttingOptimizer {

    private final Map<Integer, Integer> pricingMap;

    public RodCuttingOptimizer(final Map<Integer, Integer> pricingMap) {
        this.pricingMap = Collections.unmodifiableMap(pricingMap);
    }

    // STEP 0:
    // The initial solution as per the book.

    public int maxProfitNaive(final int length) {
        final int profitIfNotCut = pricingMap.getOrDefault(length, 0);
        // dual recursive call!
        final int maxProfitIfCut = IntStream.rangeClosed(1, length / 2)
                .map(left -> maxProfitNaive(left) + maxProfitNaive(length - left))
                .max()
                .orElse(0); // if there is no value because the original IntStream is empty, use 0
        return Math.max(profitIfNotCut, maxProfitIfCut);
    }

    // STEP 1:
    // As above, but indirect, with the recursive descent in
    // maxProfitIndirectInner() calling the function passed as argument #1.
    // In this case, the topmost function.
    // The call basically means "go do your work and call me with a smaller length on recursive descent"

    public int maxProfitIndirect(final int length) {
        return maxProfitIndirectInner(this::maxProfitIndirect, length);
    }

    // STEP 2:
    // As above, but we do not want the *topmost* function to
    // be called on recursive descent, but instead *another function* that we create locally.

    public int maxProfitIndirectDetachedFromTop(final int length) {
        final Function<Integer, Integer> shimFunction = new Function<>() {
            public Integer apply(final Integer length2) {
                // "this" is exactly the "shimFunction"
                return maxProfitIndirectInner(this, length2);
            }
        };
        // kickstart the recursive descent
        return shimFunction.apply(length);
    }

    // STEP 2 WHICH WE CAN'T HAVE
    // We cannot write the above like this in Java as there is no way to
    // put anything into the $MYSELF$ hole, we would need a "Y Combinator" for that (I think)

    /*
    public int maxProfitDoublyIndirect2(final int length) {
        Function<Integer, Integer> shimFunction = (Integer input) -> maxProfitIndirectInner($MYSELF$, length);
        return shimFunction.apply(length);
    }
    */

    // STEP 3:
    // As above, but now we are memoizing with a HashMap local to the "shimFunction".
    // Note that if stream processing actually parallelizes its processing, we are
    // in trouble as the access to the HasMap is not synchronized. So beware!

    public int maxProfitIndirectMemoizing(final int length) {

        final Function<Integer, Integer> shimFunction = new Function<>() {
            private final Map<Integer, Integer> store = new HashMap<>();

            public Integer apply(final Integer length2) {
                if (!store.containsKey(length2)) {
                    int value = maxProfitIndirectInner(this, length2);
                    store.put(length2, value);
                }
                return store.get(length2);
            }
        };

        // kickstart the recursive descent
        return shimFunction.apply(length);
    }

    // STEP 4:
    // As per the book, we can "factor out" the memoizing shim function into an (inner) class.
    // In the book, this is called maxProfit().

    private static class Memoizer {

        public static <T, R> R memoize(final BiFunction<Function<T, R>, T, R> innerFunction, final T input) {

            // An anonymous class implementing an interface!
            // Containing a cache ("store") as a Map<T,R>

            Function<T, R> memoizedFunction = new Function<>() {

                private final Map<T, R> store = new HashMap<>();

                public R apply(final T input) {
                    if (!store.containsKey(input)) {
                        store.put(input, innerFunction.apply(this, input));
                    }
                    return store.get(input);
                }
            };

            return memoizedFunction.apply(input);
        }
    }

    public int maxProfitIndirectMemoizingUsingMemoizer(final int length) {
        // https://docs.oracle.com/javase/8/docs/api/java/util/function/BiFunction.html
        // BiFunction<Function<Integer, Integer>, Integer, Integer> biFunction = this::maxProfitIndirectInner;
        return Memoizer.memoize(this::maxProfitIndirectInner, length);
    }

    // The method that uses the "indirect" function.
    //
    // In the book, it is called "computeMaxProfit()"
    // and "indirect" is called "memoizedFunction" (which is not entirely true as this is not
    // properly the memoized function)
    //
    // "maxProfitIndirectInner" can be mapped to a java.util.function.BiFunction
    // that maps the following types and roles:
    //
    // ( <Function<Integer, Integer>  ,   Integer     ) ->    Integer
    //
    // ( [the "indirect function"]    , [rod length]  ) ->  [max profit]
    //
    // In ML notation this would be simpler:
    //
    // ( Integer -> Integer ) ->  Integer -> Integer
    //
    // This function is only "not static" in this example because its context (i.e. "this")
    // contains the "pricingMap", which could also be passed as a separate parameter instead.

    private int maxProfitIndirectInner(final Function<Integer, Integer> indirect, final int length) {
        final int profitIfNotCut = pricingMap.getOrDefault(length, 0);
        // dual recursive call!
        final int maxProfitIfCut = IntStream.rangeClosed(1, length / 2)
                .map(left -> indirect.apply(left) + indirect.apply(length - left))
                .max()
                .orElse(0); // if there is no value because the original IntStream is empty, use 0
        return Math.max(profitIfNotCut, maxProfitIfCut);
    }

}

Where Next?

Popular Pragmatic Bookshelf topics Top

abtin
page 20: … protoc command… I had to additionally run the following go get commands in order to be able to compile protobuf code using go...
New
jesse050717
Title: Web Development with Clojure, Third Edition, pg 116 Hi - I just started chapter 5 and I am stuck on page 116 while trying to star...
New
sdmoralesma
Title: Web Development with Clojure, Third Edition - migrations/create not working: p159 When I execute the command: user=&gt; (create-...
New
edruder
I thought that there might be interest in using the book with Rails 6.1 and Ruby 2.7.2. I’ll note what I needed to do differently here. ...
New
raul
Page 28: It implements io.ReaderAt on the store type. Sorry if it’s a dumb question but was the io.ReaderAt supposed to be io.ReadAt? ...
New
patoncrispy
I’m new to Rust and am using this book to learn more as well as to feed my interest in game dev. I’ve just finished the flappy dragon exa...
New
brunogirin
When installing Cards as an editable package, I get the following error: ERROR: File “setup.py” not found. Directory cannot be installe...
New
adamwoolhether
Is there any place where we can discuss the solutions to some of the exercises? I can figure most of them out, but am having trouble with...
New
davetron5000
Hello faithful readers! If you have tried to follow along in the book, you are asked to start up the dev environment via dx/build and ar...
New
dachristenson
I’ve got to the end of Ch. 11, and the app runs, with all tabs displaying what they should – at first. After switching around between St...
New

Other popular topics Top

axelson
I’ve been really enjoying obsidian.md: It is very snappy (even though it is based on Electron). I love that it is all local by defaul...
New
AstonJ
Or looking forward to? :nerd_face:
483 10427 254
New
Exadra37
I am thinking in building or buy a desktop computer for programing, both professionally and on my free time, and my choice of OS is Linux...
New
Margaret
Hello everyone! This thread is to tell you about what authors from The Pragmatic Bookshelf are writing on Medium.
1145 27688 760
New
New
husaindevelop
Inside our android webview app, we are trying to paste the copied content from another app eg (notes) using navigator.clipboard.readtext ...
New
DevotionGeo
I have always used antique keyboards like Cherry MX 1800 or Cherry MX 8100 and almost always have modified the switches in some way, like...
New
New
CommunityNews
A Brief Review of the Minisforum V3 AMD Tablet. Update: I have created an awesome-minisforum-v3 GitHub repository to list information fo...
New
AstonJ
This is cool! DEEPSEEK-V3 ON M4 MAC: BLAZING FAST INFERENCE ON APPLE SILICON We just witnessed something incredible: the largest open-s...
New

Latest in Functional Programming in Java, Second Edition

Functional Programming in Java, Second Edition Portal

Sub Categories: