Reputation
3,685
Top tag
Next privilege 5,000 Rep.
Approve tag wiki edits
Badges
1 16 49
Impact
~61k people reached

1d
comment 1D ImplictRegion can not be plotted with RegionPlot
Thank you for support : )
2d
comment How to add a common color legend to a grid of density plots?
@rcollyer I still think your method maybe flawed. If you try 10*i*Sin[i x] Sin[y/i], the color is still the same. But it should be different now for these 4 plots
2d
comment How to add a common color legend to a grid of density plots?
@rcollyer Hi, rcollyer. You example is not right in MMA 10.4. The color of densityplot is different to that in legend
2d
comment 1D ImplictRegion can not be plotted with RegionPlot
Thank you so much, m_goldberg. I agree that the reply you received is ambiguous enough : ) Another thing I don't understand is why my post got so many close votes?
Apr
26
comment KernelObject appears dead problem
@xzczd I am not sure whether MMA will gives me an answer finally, but the nohup2.out is still updating screen output of iteration step information
Apr
26
comment KernelObject appears dead problem
@xzczd Hi, xzczd. I am worried whether the result will be right or not when seeing such a message
Apr
26
comment KernelObject appears dead problem
@ilian I use 10.3 on my cluster. So this is a bug? Under what conditions will it occur? I just ran 3 tasks on three different nodes, only one got this problem
Apr
26
comment How to make use of multithread feature of Eigenvalues in ParallelDo or ParallelTable
@ilian Where is MKL_DYNAMIC? If it is the default setting, why ParallelDo disable it?
Apr
25
comment 1D ImplictRegion can not be plotted with RegionPlot
@m_goldberg Thank you, m_goldberg : )
Apr
25
comment How to make use of multithread feature of Eigenvalues in ParallelDo or ParallelTable
@ilian Thanks! I forgot this trick! ParallelEvaluate@SetSystemOptions["ParallelOptions"->"MKLThreadNumber"->4] works. However, is it possible to make the thread distribution dynamic? I mean when one of the parallel tasks is done, the available threads can distributed to the running tasks. I tried setting MKLThreadNumber"->24 at first, but it harms the whole performance.
Apr
24
comment How to make use of multithread feature of Eigenvalues in ParallelDo or ParallelTable
@EmilioPisanty Of course, on a cluster, I have to pack my calculation as .m file. But what do you mean invoke inside ParallelDo?
Apr
23
comment Why timing of FindRoot is almost the same even if the function is speedup significantly
@Sander I found adding _?NumericQ brings things to normal. see my update. Though I still don't know what has FindRoot done to func1 without NumericQ to make it unexpectedly fast. As you said, we'd better just accept this. Thank you very much for the considerable time you have spent on my post
Apr
22
comment Why timing of FindRoot is almost the same even if the function is speedup significantly
I update my post again, have a look again : )
Apr
22
comment Why timing of FindRoot is almost the same even if the function is speedup significantly
@Sander I updated my post, have a look
Apr
22
comment Why timing of FindRoot is almost the same even if the function is speedup significantly
@DanielLichtblau Hi, DanielLichtblau. I see many upvotes on ciao's comment and also Sander's, but I just still don't get the point. The stepmonitor shows the same for func1 and func2, now func2 is 25 times faster than func1. But the speed of whole process is not that different. What did FindRoot spend its time on?
Apr
22
comment Why timing of FindRoot is almost the same even if the function is speedup significantly
Hi, Sander, thank you very much for the answer. But my point in my post is that "why speed up func has no effect on speed of FindRoot even though the internal steps are exactly the same?" It seems that you havn't explained this in your answer
Apr
22
comment Why timing of FindRoot is almost the same even if the function is speedup significantly
@ciao Hi, ciao. I don't quite understand your claim. The three func versions produced the same result. Why will there be any difference in internal steps? As I tested, the internal processes are exactly the same. FindRoot[func[u] - num == 0, {u, -100, 100}, StepMonitor :> Print["Step to u = ", u]] for all three func version. On the other hand, the most time cosuming part in finding root is calculate the function values at different point, Why shouldn't we expect a linear relationship??
Apr
21
comment best way for finding approximate root of monotonic sawtooth like function
I was googling secant before I post this. Somehow I didn't find how to do it in mathematica.... I should have read the doc more carefully. Thank you very much for quick response : )
Apr
21
comment 1D ImplictRegion can not be plotted with RegionPlot
@JasonB While I can confirm on 10.4, both of them fail...
Apr
21
comment How to form a block-diagonal Matrix from a list of matrices?
Oops, I am wrong. It is working for SparseArray, while not working for SymmetrizedArray type. Check it again : )