RE: Re: Insufficient capacity

*To*: mathgroup at smc.vnet.net*Subject*: [mg86534] RE: [mg86199] Re: Insufficient capacity*From*: "Tugrul Temel" <temelt at xs4all.nl>*Date*: Thu, 13 Mar 2008 04:32:42 -0500 (EST)

Dear All, Thanks for your tips...My problem remains, however. I have a system of equations stored in Excel. I import the following system to a Mathematica code without any problem: (1) Xbm = a vector of N1 real numbers with 20 decimals (2) equR = a vector of N2 equations in the form: X1 - b X2 = 0 where b is a 20 decimal number. For example, with N2=4, we write equR={X1 - b X2 = 0, X3 - c X5 = 0, X4 - a X3 = 0, X5 - d X4 = 0} where coefficients (a, b, c, d) each of which has 20 decimals (3) equI = a vector of N3 equations in the form: Y1 +Y2 - Y3 - Y4 = 0. For example, with N3=3, we write equI={ Y1 +Y2 - Y3 - Y4 = 0, Y4 - Y6 = 0, Y1 + Y3 - Y5 =0} The actual system I am going to solve has a total of 3,500 unknowns and 3,500 equations (N2+N3=3,500). I have no problem in importing the data in Excel to Mathematica. My problem starts when I do several matrix operations because of large decimal numbers. When I make matrix operations, the Mathematica code is not able to the operations and puts me out of the program. I suspect two things: 1) my computer does not have enough memory (my computer Intel core 2 duo processor T5500 has 1.66 Ghz, 667 mhz speed, 2 MB LL cache) 2) large decimals occupy a lot of space in memory 3) Maybe I miss a point in dealing with large decimal numbers. Please let me know your suggestions about what to do, including having an upgraded computer with enough memory and speed. Regards, Tugrul -----Original Message----- From: Alois Steindl [mailto:Alois.Steindl at tuwien.ac.at] Sent: Thursday, March 06, 2008 8:55 AM To: mathgroup at smc.vnet.net Subject: [mg86534] [mg86199] Re: Insufficient capacity Hello, "Tugrul Temel" <temelt at xs4all.nl> writes: > I transfer large amount of data from Excel to Mathematica v.6 and try to > do some matrix operations. The size of the matrces are (3500, 3500) with > 10 digit numbers. However, Mathematica quits when I make the matrix > operation. > > How can I make it possible to do the matrix operations of large sizes? Is > Grid Math an option in this respect? If so, I like to know how large data > sets Grid Math can handle? > 3500x3500 isn't really large. What are your "matrix operations"? Does it work, when you convert your matrices to FloatingPoint numbers (N[matrix])? Alois