bin packing
in [Programming SQL]
|
From: Peso on 22 Oct 2009 18:37 http://weblogs.sqlteam.com/peterl/archive/2008/11/23/Bin-packaging.aspx "Gert-Jan Strik" <sorrytoomuchspamalready(a)xs4all.nl> wrote in message news:4ADF37A2.C826CDAB(a)xs4all.nl... > For those interested, I have written a short series of articles on > solving the bin packing problem. > > See http://www.xs4all.nl/~gertjans/sql/binpacking/intro.html > > -- > Gert-Jan > SQL Server MVP
From: TheSQLGuru on 22 Oct 2009 18:52 Once again your ivory-tower idocy comes forth. In the real world people do not always get to pick the optimal platform on which to develop a product or application. No one said anything about transact sql being a good toold for solving this problem. -- Kevin G. Boles Indicium Resources, Inc. SQL Server MVP kgboles a earthlink dt net "--CELKO--" <jcelko212(a)earthlink.net> wrote in message news:0938eb9a-9b7f-4070-be7e-ebb4166389f1(a)z34g2000vbl.googlegroups.com... > Wikipedia has a good intro article on the topic. The classic book is: > > Knapsack Problems: Algorithms and Computer Implementations (Wiley- > Interscience Series in Discrete Mathematics and Optimization) by > Silvano Martello and Paolo Toth (Paperback - Nov 1990) > > The bad news is that it is out of print and the price is so high I > might sell my copy on EBay! They had FORTRAN programs that were fairly > complicated. Today, the algorithm research has moved to modern > languages with parallelism. > > The simple greedy FFD (First Fit Descending) algorithm works very well > in the real world. Standardized packages were designed to fit into > standardized bins for shipping so thing were meant to fit together. > But when FFD fails, it can be really bad. > > This is not a good SQL problem and you need a constraint language with > parallelism. Cursors will not help because they are not parallel. > Strik did a nice job on the problem and kept the code simple (if he > gets the Martello-Toth book he can translate their FORTRAN to T-SQL > for more improvements); but the truth is that this not an SQL > problem. Would you write an accounting package in ICON?
From: --CELKO-- on 22 Oct 2009 21:37 >> Once again your ivory-tower idocy [sic] comes forth. << in case you did not know, the knapsack/bin packing problem is CLASSIC "ivory-tower idiocy" with a long history. >> In the real world people do not always get to pick the optimal platform on which to develop a product or application.<< Buy me a lot scotch and I will tell you about commercial applications in FORTRAN on mini-computers in the 1970's. >> No one said anything about T-SQL being a good tool for solving this problem. << Yes, nobody did. Please tell us the point of your posting. Kevin, you usually do better than this. When my game newsgroups game is down I am usually drunk. Ask my wife. SCRATCH THAT!
From: Tony Rogerson on 23 Oct 2009 01:45 > Yes, nobody did. Please tell us the point of your posting. Kevin, > you usually do better than this. When my game newsgroups game is down No doubt yet another expert in the field entirely sick of your inane drivel and bad attitude on this MICROSOFT SQL SERVER forum. Why is it we never see you on the db2, oracle or teradata forums? Considering you said you are active on those products and not active in SQL Server logic would dictate that you would spend time on those other forums. --ROGGIE-- "--CELKO--" <jcelko212(a)earthlink.net> wrote in message news:eb3ea232-410a-44ff-8ead-4d30e7cc54f7(a)j19g2000vbp.googlegroups.com... >>> Once again your ivory-tower idocy [sic] comes forth. << > > in case you did not know, the knapsack/bin packing problem is CLASSIC > "ivory-tower idiocy" with a long history. > >>> In the real world people do not always get to pick the optimal platform >>> on which to develop a product or application.<< > > Buy me a lot scotch and I will tell you about commercial applications > in FORTRAN on mini-computers in the 1970's. > >>> No one said anything about T-SQL being a good tool for solving this >>> problem. << > > Yes, nobody did. Please tell us the point of your posting. Kevin, > you usually do better than this. When my game newsgroups game is down > I am usually drunk. Ask my wife. > > SCRATCH THAT!
From: Gert-Jan Strik on 24 Oct 2009 09:20 Thanks for the tip. I found the Martello-Toth book online, including their FORTRAN code. I am studying it now. If it leads to an improved SQL Server implementation, I will let the world know. -- Gert-Jan SQL Server MVP --CELKO-- wrote: > > Wikipedia has a good intro article on the topic. The classic book is: > > Knapsack Problems: Algorithms and Computer Implementations (Wiley- > Interscience Series in Discrete Mathematics and Optimization) by > Silvano Martello and Paolo Toth (Paperback - Nov 1990) > > The bad news is that it is out of print and the price is so high I > might sell my copy on EBay! They had FORTRAN programs that were fairly > complicated. Today, the algorithm research has moved to modern > languages with parallelism. > > The simple greedy FFD (First Fit Descending) algorithm works very well > in the real world. Standardized packages were designed to fit into > standardized bins for shipping so thing were meant to fit together. > But when FFD fails, it can be really bad. > > This is not a good SQL problem and you need a constraint language with > parallelism. Cursors will not help because they are not parallel. > Strik did a nice job on the problem and kept the code simple (if he > gets the Martello-Toth book he can translate their FORTRAN to T-SQL > for more improvements); but the truth is that this not an SQL > problem. Would you write an accounting package in ICON?
First
|
Prev
|
Next
|
Last
Pages: 1 2 3 Prev: freetext problem Next: SSIS Package Comparison Utility (Freeware) |