[RESOLVED] Writing an extremely tiny number to disk

[luckiejacky]

float matWorld[16]

errno_t err = _wfopen_s( &fFile, File.c_str(), L"wb" );
fwrite( (void*) matWorld, sizeof(float), 16, fFile);

The numbers are as small as these

1.56768e-007	-2.36557e-009	1.96187e-011	0	
2.36557e-009	1.56768e-007	-4.46082e-012	0	
-1.95498e-011	4.76264e-012	1.56786e-007	0	
-0.000124691	3.03403e-005	-0.000232232	1

The results of the save are nearly all zeros....

results:
0.000000,-0.000000,0.000000,0.000000,0.000000,0.000000,-0.000000,0.000000,-0.000000,0.000000,0.000000,0.000000,-0.000125,0.000030,-0.000232,1.000000;;

Is there a way to save the small numbers with scientific notations?
thanks
Jack

[VictorN]

...

The results of the save are nearly all zeros....

results:
0.000000,-0.000000,0.000000,0.000000,0.000000,0.000000,-0.000000,0.000000,-0.000000,0.000000,0.000000,0.000000,-0.000125,0.000030,-0.000232,1.000000;;

But how do you open/read this file to see the results?

[2kaud]

Consider

#include <cstdio>
#include <string>

using namespace std;

int main()
{
	using numtype = float;

	const string File= "num.bin";

	const numtype matWorldOut[] = {	1.56768e-007, -2.36557e-009, 1.96187e-011, 0,
					2.36557e-009, 1.56768e-007, -4.46082e-012, 0,
					-1.95498e-011, 4.76264e-012, 1.56786e-007, 0,
					-0.000124691, 3.03403e-005, -0.000232232, 1};

	const size_t noElems = sizeof(matWorldOut) / sizeof(matWorldOut[0]);

	numtype matWorldIn[noElems];
	FILE* fFile;

	if (fopen_s(&fFile, File.c_str(), "wb")) {
		cout << "Error opening file for write" << endl;
		return 1;
	}

	if (fwrite((void*)matWorldOut, sizeof(numtype), noElems, fFile) != noElems) {
		cout << "Error writing data" << endl;
		return 2;
	}

	fclose(fFile);

	if (fopen_s(&fFile, File.c_str(), "rb")) {
		cout << "Error opening file for read" << endl;
		return 1;
	}

	if (fread((void*)matWorldIn, sizeof(numtype), noElems, fFile) != noElems) {
		cout << "Error reading data" << endl;
		return 2;
	}

	cout << "Written data" << endl;

	for (const auto& n : matWorldOut)
		cout << n << "  ";

	cout << endl << endl << "Read data" << endl;

	for (const auto& n : matWorldIn)
		cout << n << "  ";

	cout << endl;
	return 0;
}

which displays

Written data
1.56768e-07  -2.36557e-09  1.96187e-11  0  2.36557e-09  1.56768e-07  -4.46082e-12  0  -1.95498e-11  4.76264e-12  1.56786e-07  0  -0.000124691  3.03403e-05  -0.0
00232232  1

Read data
1.56768e-07  -2.36557e-09  1.96187e-11  0  2.36557e-09  1.56768e-07  -4.46082e-12  0  -1.95498e-11  4.76264e-12  1.56786e-07  0  -0.000124691  3.03403e-05  -0.0
00232232  1

where the read data is the same as the written data.

[wolle]

Is there a way to save the small numbers with scientific notations?

If you store a float in binary format on file and read it back again, it will be the same. All 32 bits of the float stay intact. The same float may appear different on screen but this is determined by which format you use to display it. "Scientific notation" is one possibility, "Fixed" is another.

You don't have to worry about a float being small (or big). It will have the same precision (6-7 accurate digits) throughout the whole of its valid range (from about +-10^-34 up to +-10^34). (If you accidentally use a float outside these limits it will enter a state of underflow/overflow which you can check for).