Oleksandr R.

7,505 reputation

12857

bio	website
	location
	age
visits	member for	1 year, 3 months
	seen	33 mins ago
stats	profile views	1,142

All code in my contributions on Mathematica.SE, unless otherwise specified, is dual-licensed under CC-by-SA 3.0 and your choice of any one of the following:

	7,505 reputation	bio	website		visits	member for	1 year, 3 months
12857 badges		location			seen	33 mins ago

summary answers questions tags badges favorites bounties reputation activity

1,179 Actions

suggestions reviews revisions comments badges posts accepts all

22h	comment	How to insert big list of Data in DB? Quit using code markup on arbitrary words and phrases! Apart from that, the whole idea is absurd. If you want to store numbers in a database, use a numeric type--don't convert the values into Unicode characters and store them as a string.
1d	revised	How to use older code in the latest release added 60 characters in body
1d	comment	Speeding up a numerical constrained quadratic optimization For example, `Method -> {"RandomSearch", Method -> "QuadraticProgramming", "PostProcess" -> "KKT"}`. Unfortunately, although it does as requested, it seems that this does not give very good results, since `NMinimize` first transforms into an unconstrained problem, which `FindMinimum` doesn't particularly like. I don't have time to investigate this now, but in the mean time you could use the BFGS algorithm (`Method -> "QuasiNewton"`) in place of Levenberg-Marquardt/quadratic programming to get around the issue. This seemed to work acceptably well in my brief tests.
1d	comment	Speeding up a numerical constrained quadratic optimization That's correct, but you can still use it, via the `"RandomSearch"` method of `NMinimize`, if you want to try to find the global minimum. Of course, that you will actually find it can't be guaranteed unless you use `Minimize`, but I expect this will be even slower than `NMinimize`, if it works at all.
1d	comment	How to run an external program in parallel with a Mathematica package? Can you try `ParallelEvaluate`/`ParallelTable` with option `DistributedContexts -> None`? I find that the automatic distribution of definitions usually causes more problems than it saves and you may have encountered an issue with it here. As for the MathLink protocol: no (externally observable) changes were made between 8 and 9. We are still on `MLINTERFACE` version 3.
1d	reviewed	Leave Open Solar System Orbital Parameters
1d	comment	Speeding up a numerical constrained quadratic optimization You will probably be better off using `FindMinimum` with `Method -> "QuadraticProgramming"` or `"LevenbergMarquardt"` rather than `NMinimize`, as the latter uses derivative-free metaheuristic approaches that are not well suited for this problem.
1d	comment	How to estimate system recource usage of a SparseArray? There is definitely something strange here, and the memory usage is much higher in version 9 than in previous versions. This would be a good question if you would edit it to focus on the memory usage rather than the file size. When you do so, you'll have my upvote.
2d	revised	How can I make a cell that contains an image which can be shifted? added 1250 characters in body
2d	reviewed	Close What is wrong with this Cobweb plot
May 19	comment	Can one identify the design patterns of Mathematica? N.B.: "TransformBy" is analogous to the Lisp idiom sometimes known as "decorate-sort-undecorate". I think it is the fact that other transformations apart from sorting are used that distinguishes it as a pattern, although I'm not sure how far this can be considered a Mathematica technique as, clearly, it can be used in (nearly) any language.
May 19	revised	Smooth Kernel Distribution tidy up
May 19	comment	Smooth Kernel Distribution Dear Doron: good question. I edited your post to bring the the formatting and layout up to site standards. Although the original was quite acceptable in the abstract, please take note of my changes so that you'll know for the future how questions and answers should be presented (e.g., we do not write a heading for the post or use salutations).
May 19	revised	Smooth Kernel Distribution tidy up
May 19	comment	How to estimate system recource usage of a SparseArray? I can't reproduce your observations: as one would wish and expect, the file seems to be the same (small) size no matter how large the array nominally is. I tried in version 9.0.1. Which version are you using?
May 19	comment	How to create internally optimized expression for computing with high WorkingPrecision? Very good question. A good answer will hopefully discuss the Experimental`NumericalFunction, which is a structure produced by `FindMinimum` from the function and its Jacobian that is optimized for fast numerical evaluation. I know almost nothing about these objects or how to create/use them, but I would like to find out. By the way, I am not sure if it is just a typo, but it is going to be a problem that you used `RuleDelayed` for the `"Jacobian"` option in the second case. As it is currently, the 4 minute delay may be due to re-deriving the Jacobian at each point.
May 18	comment	Is it possible to run Mathematica on an HPC cluster with Torque? Very good question. I do wonder why they chose to support PBS Pro rather than Torque. If I had to choose only one of these, I would think Torque support would have been more widely applicable.
May 18	comment	When using NMinimize[…,Method->“DifferentialEvolution”] is an “iteration” different than a “step”? ... Good general settings are something like $F \approx 0.5-0.9$, $CR \approx 0.1, 0.9$, so your values are a bit extreme, but if it works for you, fine. As for the lack of message: `NMinimize` won't produce one if the global optimization doesn't converge but the later postprocessing does; both must fail for an error to be returned to the user. I suppose this is because global optimization algorithms like DE provide no useful error estimate, whereas local methods do, and thus the message is only really meaningful in the latter case.
May 18	comment	When using NMinimize[…,Method->“DifferentialEvolution”] is an “iteration” different than a “step”? It isn't clear to me whether you already took this into consideration, but it's important to note that Mathematica's `CrossProbability` is reversed relative to the literature--traditionally 1 retains the mutant vector while 0 keeps the target, but here it's the other way around. That is, a value of 0 (Mathematica's 1) yields a one-dimensional hill climbing approach, assuming that Mathematica implements DE as described by Storn and Price rather than using this as a literal probability. (I did look at the code at one point, but I've now forgotten what it does.) ...
May 17	comment	How to troubleshoot CUDALink? I think there is an issue with almost all CUDA functions on cards this old: they simply do not possess the hardware capabilities required by most useful workloads. The Wikipedia article does a good job of describing what is missing from CUDA 1.0. I'd also be surprised if the recent nVidia drivers had been properly tested with these old cards, so this problem may not be specific to Mathematica.