accuracy of math.pi

[Gerhard34]

I have a very simple script which show pi for 15 digits. I wounder about the result: 3.14159274101257.
According to other calculations (e.g. Excel) it should be: 3.14159265358979

Could anyone pls. help to clarify the different results?
thx Gerhard

import math
d = math.pi
print(f'{d:.15}')

Platform: MicroPython v1.24.1 on 2024-11-29; ESP32C3 module with ESP32C3

[robert-hh]

Esp32c3 uses single float. So, only 6-7 digits are significant.

[sosi-deadeye]

3.14 is accurate enough for the most calculations.

[Gerhard34]

Thank you @robert-hh for you explanation!

@sosi-deadeye: I agree that this might be sufitient for most cases, but I'm running GPS calculations with the Haversine formula on short distance between two GPS points. This requires a more presice calculation.

[rcolistete]

Few MicroPython firmwares have FP64 (double precision for floats), like Pyboard :
https://micropython.org/download/PYBV11/
Some MicroPython boards can have firmware compiled with FP64. Like ESP32 :
https://github.com/rcolistete/ulab_samples

Other solution :
Announcing DecimalNumber: floating point arithmetic with arbitrary precision for micropython
https://forum.micropython.org/viewtopic.php?t=11375

[peterhinch]

To clarify re Pyboards:

The Pyboard V1.1 has 32-bit hardware FP, but it can have firmware compiled with FP64. This is available as a precompiled build. The Pyboard D SF6W has 64-bit hardware FP.

[scruss]

There are also unofficial FP64 builds for the RP2040, but they are tremendously slow.

DecimalNumber is wonderful, but even slower than all that.