Skip to content

CSP in AIMA 4 JAVA doesn't meet Theory #461

Open

Description

@ThuriyaThwin

I am testing AIMA-4 edition of Java with only CSP.
My machine is

CPU: ` Intel® Core(TM) i7-4790 CPU @ 3.6GHz 3.6Hz
Cores: 4
Thread: 8
Processor Base Frequency: 3.6 GHz
Max Turbo Frequency: 4.0 GHz
Cache: 8 MB Smart Cache
Bus Speed: 5 GT/s DMI2
Installed memory (RAM): 8.00GB
GPU: Intel® HD Graphics 4600
Operating System(OS): Window 8.1 Pro 64bit (6.3,Build 9600)

My environment is

Compiler: Jdk 1.8.0_211
IDE: IntelliJ IDEA Community Edition

I have caught the run time of Map CSP Australia in 10 times and average on it.

BT 0.058 s
FC 0.068 s
FC+MRV 0.075 s
FC+LCV 0.068 s
MAC 0.061 s

These results does not meet the theory described in AIMA Testbook @norvig .AIMA 3 edition meets mostly but not in FC+LCV but in 4 edition it doesn't meet at all.

I have tested aima python.It is a perfect library and all meets the theory described in book.
The book says FC+MRV should faster than BT about 3-3000 times.
@ctjoreilly @Medeah @manuel-delverme @sebkur @Mrflow Please check the topic in CSP.
It dangers to the novice learners.

Regards,
Thwin
thuriyathwin@ucsm.edu.mm

Activity

  1. Medeah commented on May 31, 2019

    @Medeah
    Contributor

    If I understand you correct you expect the run time of the different method to be drastically different.

    The reason is that the Australia map example is too small to show any difference. One way to solve this would be to do a performance comparison of some bigger problems.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions