mikerabat / mrmath Goto Github PK

It misses the "x" of x64.

Fatal source bugs on compiling

==========
Windows 10
Clean new install of Lazarus 32 bit Latest version (v2.2.0)

What I had to do to get a compile

============
File mrmath_pkg.lpk
Remove
Item 75 - file BufferedStream.pas

Looks like file does not exist

===========
File SimpleWinThreadPool.pas

comment out line 226, viz
SetThreadPoolProvider( {$IFDEF FPD}@{$ENDIF}SimpleWinThreadPoolProvdier );

reason: Identifier not found

==================
File eigensystems.pas

comment out line 3051
if abs( e^[m] ) <= eps2*abs( d^[m]*d[m + 1] ) then

Reason: Eigensystems.pas(3052,36) Error: Operator is not overloaded: "Double" * "TConstDoubleArr"

============

CONSEQUENCES in test of UNIT:

Test failed

TestEigValfromTridiagMtx
Exception: EAssertionFailedError
at $0067B388

==========

What is the fix?

If you make .lpk Lazarus package file, them mrMath can be added to Lazarus OnlinePackageManager (seems).
you can create new package in Lazarus IDE, then add all pas files, then add description, save.

Hello mikerabat,
I've downloaded mrmath and tried to compile the "mrmath.dpk" as described.
It gives the following message
[error] ASMMatricMeanOperations.pas: undeclared identifier: 'movddup'

I'm using Delphi 7 on a windows 10 machine, and have zero experience with inline assembly code.
any suggestions? thanks

hi
how can i use this library?
in readme file you said compile!
i compiled the code with out error but how can use in my code ?!

hi,mike,

why is the update time so long?

why stop submitting new projects?

are you busy with your work?

Best Regards
by.qq600585

hi,mikerabat

I plan to use mrmath (aiot / nb-iot) in Jetson for my project.
Assembly / AVX / SSE code cannot support armv7 / arm64 / MIPs
Please keep Pascal source code. thank you!

Best Regards
by.qq600585

Hi, Mr Mikerabat
Really, Really, Really thank you for your useful publification of this libray.
I am a delphi developer.
Currently I am making small image processing program ( in my personal purpose).
In this process I should use matrix cacluation.
Thus I selected your library for my project.
But I faced below problem like below.
And I can't find the reason.
please give me advice.

It was occured on matrix multiply function.
And my code block is like below.

function getABC(srcBmp:TBitmap): TBitmap;
var
  matA, matZ: TDoubleMatrix;
  w, h: Cardinal;
  count: Cardinal;
  x, y: Integer;
  mean, sum: Double;
  _rVal: Cardinal;
  aVal, bVal, cVal: Double;
  matT, _matM1, _matM2, _matM3, matIv: TDoubleMatrix;
  matX: TDoubleMatrix;
begin
  w := srcBmp.Width;
  h := srcBmp.Height;
  matA := TDoubleMatrix.Create(w*h, 3, 0);
  matZ := TDoubleMatrix.Create(w*h, 1, 0);
  count := 0;
  for x := 0 to w - 1 do
  begin
    for y := 0 to h - 1 do
    begin
      matA[count, 0] := x;
      matA[count, 1] := y;
      matA[count, 2] := 1;
      _rVal := GetRValue(srcBmp.Canvas.Pixels[x, y]);
      matZ[count, 0] := _rVal;
      Inc(count);
    end;
  end;
  ///  How to Save to File
  ///  matA.SaveToFile('test.csv', TBinaryReaderWriter );
  count := 0;
  sum := 0;
  for x := 0 to w - 1 do
  begin
    for y := 0 to h - 1 do
    begin
      _rVal := GetRValue(srcBmp.Canvas.Pixels[x, y]);
      sum := sum + _rVal;
    end;
  end;
  mean := sum / (w * h);
  matT := matA.Transpose;
  _matM1 := matT.Mult(matA); //current error line
  matIv := _matM1.Invert;
  _matM2 := matIv.Mult(matT);
  matX := _matM2.Mult(matZ);
  //matX := ((matT.Mult()).Invert).Mult(matA.Transpose).Mult(matZ);
//  aVal := matX[0];
//  bVal := matX[1];
//  cVal := matX[2];
end;