@AHSAN The rangesonly option makes getdata return the ranges for all the dimensions of the plot.

A 2D plot has two dimensions. So in your case it's returning two ranges. The indexing, [2], accesses the second of those ranges, which is the vertical range.

I think that you could have figured that out.

Sorry, I don't understand what you mean or want, by, "show these line according to desire position".

@AHSAN That operator/procedure assigned to r is used to construct the columns, as rectangles.

The second parameter of that operator, y, is used to specify the vertical end (ie. height, whether positive or negative) of the column.

The other end of the column, vertically, is at 0, ie. the x-axis.

ps. I think that you could have figured that out.

@AHSAN I have modified the procedure so that it is more flexible, accepting additional plotting options and the numerical format for the column labels.

Help_Bar_Graph_acc.mw

And now it be called to produce these, say,

@Nicole Sharp Regarding your statement about extra parentheses being required in Maple, you could also enter that as,

   evalf(2*Pi^5*1.380649e-23^4/(15*299792458^2*6.62607015e-34^3))

@Christopher2222 No, it's not ridiculous or terrible -- it was just two short examples. I did not intend to be harshly critical.

I think that the overall idea is a good one. Indeed I suggested the same thing in 2008 but, while that devolved into a lengthy discussion about performance of float[8] Matrix Rank (for no reason other than that Nasser Abbasi once gave it, with results did not hold for long...), there was not a flood of examples.

It's a difficult thing to do very well. More importantly, it's pretty natural for a benchmark suite to get revised and reshaped.

Perhaps it would be helpful to start a list of areas for benchmarking, eg:
 - linear algebra
 - statistics
 - integration
 - differential equations
 - polynomial manipulation
 - special function evaluation
etc. Each of those might be sensibly split into numeric and symbolic computation. There's also group theory, graph theory, etc. Interface responsiveness is also interesting (eg. is 2-D input slow to parse, how fast is 2-D output, etc).

And comparison of benchmark numbers between operating systems, product releases, and hardware, might all be interesting.

[edit] There have been a few performance comparisons between competing products, (eg. Mma, Maple, Matlab, etc). One of those (now defunct, I think) was mentioned by me in this old Post. Maybe I still have some code from it; I'd have to find time to look.

Another speed performance product comparison was done by Wolfram, comparing Mma and Maple at both hardware and higher working precision, for a variety of numeric computations. Somewhere I have a Linux script which can run the float[8] LinearAlgebra portion of the Maple computations in an early snapshot of that -- only about a dozen or so commands tested. I used to use that to compare Maple + generic-CLAPACK/ATLAS/MKL, across releases. I had a swanky worksheet embedded-components interface, and it throws up a bunch of plots to overlay and compare Maple releases. I could probably find it and dust it off. The automation was useful. [edit] sursumCorda has since mentioned this suite, below, and provided the Maple code of a much later revision. I find the examples interesting, but I don't like the bookkeeping or methodology.

I think that individual benchmark examples should be run in fully separate kernel sessions, preferably from separate plaintext files.

@Christopher2222 IMO there are several weaknesses in your methodology. (I order these by decreasing severity, from mistake to quibble.)

1) You're measuring the sum of CPU time across multiple cores, which is a serious flaw. If you want to measure the wall-clock time duration for which the user has to wait, here, either use time[real]() or an appropriate call to CodeTools:-Usage. Your approach gives a confused idea about how a machine with more cores might finish the job more quickly, and about how long the computation actually takes.

2) As you have it, you're also measuring the time it takes for the interface to print the output. While that may be relatively negligible here, a clear picture in general would be to have separate operations to measure computations and (possible) something else dedicated to measure interface (eg, GUI, CLI) output timing. Also, there are some computations whose result take much longer to print than to compute. You could end the computation with a full colon.

3) This example tests the performance of the externally linked Intel MKL's dgesvd routine. That might interest some people [see also discussion in a 2008 Post of mine in which I too suggested a Maple benchmark, and also here], but it's not a great indicator of Maple's speed on general numerics (whether in evalhf or arbitrary precision modes).

4) Why set UseHardwareFloats:=true ? Shouldn't you have a specific and clear reason for changing away from default settings -- for this or any benchmark example?

@FDS The first form works ok for me, on columns of DS.

Test_date_ac2.mw

Perhaps you tried it with the wrong kind of single-quotes?

If you know for sure which columns will have the numeric entries then it's more efficient to only examine those. See attachment.

@FDS Do you mean this kind of result?

add(`if`(u::numeric, 1, 0), u = DS);

        418

nops(select(type, convert(DS, list), numeric));

        418

@Carl Love I think you intended eval(phi1, S2[2]) instead of eval(S2[2], phi1) 

I did some extra fiddling there, because I suspect that neither dataplot or plot (for Matrices/Vectors) understand units yet.

But this seems to work, in your Maple 2022.2.

  plots:-pointplot(res, lact, useunits = [Unit(h), 1], style = line)

If you upload and attach your .cvs file then we can reproduce, fix, and demonstrate the whole thing.

You could upload/attach a .zip file, if the Mapleprimes uploader baulks at the .cvs filename extension.

Are you entering it as capitalized Pi, or pi?

Maple interprets Pi as the well-known constant, but pi is just a name with no special meaning to Maple.

If you have some other problem, please explain it more clearly, or upload and attach your worksheet with the green up-arrow in the Mapleprimes editor.

@minhthien2016 Your last example can be handled by,

Download sgn_ex6.mw

It could save time if you provided your more complicated examples now, rather than staggering them.

@gabbbooooo Please put your followups here, instead of spawning wholly separate new Question threads for them.

@lemelinm There are two different light-gray shades of grid line possible, with one being even lighter than the other. So, when you write, "What I need is to have the light-gray gridline on the x-axis to be only on the values of 3, 6 and 9." I don't know which shade you want.

You could compare these (including zooming, if you want),

plot(0, x = 0 .. 10, y = 0 .. 4, gridlines = true,
     axis[1] = [ tickmarks=spacing(3,0),
                 gridlines=[spacing(30,0),subticks=9] ],
     labels = [x*InertForm:-Display(10)^8*``(Unit(m)), `t `(Unit(s))]);

plot(0, x = 0 .. 10, y = 0 .. 4, gridlines = true,
     axis[1] = [ tickmarks=spacing(3,0) ],
     labels = [x*InertForm:-Display(10)^8*``(Unit(m)), `t `(Unit(s))]);

It's also not clear to me what grid lines you want on the vertical axis.