@rm-rf Eek, so it can be ragged then? I was thinking of using Map[] with its third argument, but I'm not sure how to prevent things like Sort[3] from happening...
BTW @rob, the other day I wrote something you might be interested in...
@robjohn Remember the Euler transformation for alternating series? This routine does transformations of partial sums. You just pass a list of partial sums to the routine.
For example: eulerKnopp[N[Table[Sum[(-1)^j/(j + 1), {j, 0, n}], {n, 0, 10}], 20]] would perform the transformation on successive partial sums of Mercator's series.
The last comments of @SjoerdC.deVries and @frizzics in my answer here makes me wonder whether we have an extensive discussion about the timing issure of AppendTo here on Mma.SE
@J.M. Yes, especially since I checked my books here and no one discusses this except of David Wagner. But I don't have all Roman Maeder books and the Stan Wagon book.
@rm-rf I wonder; maybe If[AtomQ[#], #, Sort[#]] & //@ list might do the trick...
@halirutan Well, Wagon only did a slight segue on the disadvantages of AppendTo[] if memory serves, so I think Maeder is the only thing you'll need to look at...
@rm-rf As I say in my (edited) answer. Trying to help him was pointless. He came up with more and more questions in the comments, sometimes incomprehensible ones
@J.M. Next time someone asks how to do the eigendecomposition of a 1000x1000 matrix, you should tell them to update their question when this finishes evaluating: Eigensystem@Table[Unique[x], {1000}, {1000}]
@J.M. Nope. We were doing something with a kerr cell. Polarizing a laser, I think. The capacitors leaked some current and the liquid started to boil. Nice. We powered the equipment off and ran to the street
@belisarius "the whore's mother"? It has been well over a decade (nearing 2) since I've had spanish.
@rm-rf You don't even need to do that. Calculating the characteristic equation of a 12x12 will do it. You don't even need to calculate any eigenvalues!
@J.M. a symbolic determinant should not take that long. Yes, it is factorial growth in n, but give me a break! My old system had serious issues with a 9x9.
Hey, math guru .. er @J.M. for a slow calculating function, which would be a better scheme at minimization: conjugate gradient or a quasi-newton?
I have a crystal that exhibits an orthorhombic distortion, and I'd like to determine what DFT says is the "correct" distortion. Hence, the need for minimizing the number of functions calls.
@rcollyer Once upon a time, I'd have said CG. Nowadays, it's said that the L-BFGS variant of quasi-Newton is sufficiently competitive. Of course, you still need to do tests of your own...
...but yeah, L-BFGS will beat out the older quasi-Newton methods.
@rcollyer I know the thing was designed for less storage, but I've found the thing has less overhead as well, at least in the tests I did. Neck-and-neck, see...
@J.M. which may or may not be a problem. I know the energy dependence has a non-quadratic dependence on the lower-volume side due to an inherent issue with LAPW.
And, there are a couple of different forms. LAPW is C1 continuous at the boundary. There are alternatives, that are faster so in common use, that are C0 across the boundary.
I tend to fix mine at the beginning at 90% of max, and leave it until I've squeezed it to nearly 100%. The issue of course, the basis size decreases with volume introducing a bias.
Note: this was recommended by the authors of the package. :P
I've been able to produce the palette. When I tried it out on an old notebook, I get Java::excptn: A Java exception occurred: java.lang.ClassNotFoundException: org.apache.commons.httpclient.methods.multipart." and other error messages.
On the sphere $$(x-1)^2 + (y+2)^2 + (z-3)^2 = 25,$$ find the point $M$ nearest to the plane $$3x -4y + 19 = 0,$$ and calculate the distance $d$ from this plane. How do I tell Mathematica to do that?
I am trying to program Mathematica to do the following:
Let
g = {{g11,g12},{g21,g22}};
m = {{m11,m12},{m21,m22}};
I would like to make Mathematica to do the following for me:
Find all polynomials in the entries of the matrix $m$ which satisfy $gmg^{-1}=m$. That is,
MatrixForm[g.m.Inv...
@belisarius No, if you want to find all minima, you don't have the minima as requirement. He wants to find all polynomials. He doesn't know them before. He knows one single solution of maybe infinitely many which is the Tr[m] example.
no.. Heres a breakdown of my problem: 1) My computer has mma but is limited on resources. 2) I have cluster access with lots of resource but NO mma. 3) I'd like to use the mma and MathematicaScript interpreter (???) FROM my slow computer to run mma script ON fast.computer and store data on fast.computer
So I guess, I won't be using ssh -x as I don't need the gui front end...
@rm-rf Sure. I understand that. Which is why I am miffed at our IT support. On campus I obv. have a full license but it isn't installed on any cluster.
3. You (like in our university) have a licence server running. When you buy this from Wolfram you get installation disks for all systems and you just install Mathematica locally but when you start it, it gets its licence from the licence server.
WRI will not allow you to move your license to a different machine... been there, tried requesting, got declined. Same scenario — I wanted to run mma on a faster machine which didn't have a license and I thought I could install it only for my user... nope.
@rm-rf IT is in some sort of impasse right now.. most of the intense simulation related work at my uni have stopped over the last 4-5 weeks and people are getting borderline suicidal.
@halirutan I don't think I read the question that you mention
@drN So the short answer to your question is: I don't think it is possible to use the power of your faster computer when you have only one local installation on the slow one.
@rm-rf Does the factorization appled run on your pc?
Factorization complete in 0d 0h 0m 5s ECM: 2185028 modular multiplications Prime checking: 34594 modular multiplications SIQS: 14502 polynomials sieved 43917 sets of trial divisions 1508 smooth congruences found (1 out of every 343355 values) 18192 partial congruences found (1 out of every 28461 values) 1774 useful partial congruences
Timings: Primality test of 3 numbers: 0d 0h 0m 0.0s Factoring 1 number using ECM: 0d 0h 0m 0.8s Factoring 1 number using SIQS: 0d 0h 0m 4.4s
@halirutan Ok.. I'll look further into parallel processing et al. However since that will probably take a back seat to other stuff that I need to do to progress my work, it'll be a while before I revisit this question.
@rm-rf Me neither.. maybe it uses the wrong method for too long before switching. I have no idea. I never looked deeper into the whole prime factorization algorithms in Mma although I had a whole lecture in university about it.