Minimalistic Python Arbitrary Precision Math Using LibBF

MPBF is a MicroPython wrapper around LibBF. Works for Unix and STM32 ports, but might also work for ESP32.

File Description

File	Description
libbf.c	C source for BF core functions
libbf.h	BF core header.
cutils.c	BF utilities.
cutils.h	BF utilities header.
bf.c	C wrapper for libbf. This provides a set of functions that are called by the functions in mpbf.c
bf.h	Header for C wrapper.
mpbf.c	MicroPython external C module for libbf. There are 3 functions that are exposed to Python.
mpapbf.py	Wrapper MicroPython module. Optional.
micropython.mk	C module makfile.
test.c	Simple C test.
Makefile.mpbf	Example Makefile for running a simple test on Unix (test.c)

Compilation

1. Copy the contents of directory mpbf/ to $MP/ports/stm32/ or $MP/ports/unix/

2. Compile command for Unix:

   $ cd $MP/ports/unix/
   $ make V=1 USER_C_MODULES=. CFLAGS_EXTRA="-DMODULE_MPBF_ENABLED=1"

3. Compile command for stm32 (F407VE board):

   $ cd $MP/ports/stm32/
   $ make BOARD=BLACK_F407VE USER_C_MODULES=. CFLAGS_EXTRA="-DMODULE_MPBF_ENABLED=1 -DBARE_M=1"

MPBF Usage

All operations are done on decimal numbers represented as strings, e.g.: '2.718281828e1' is 27.18281828.

   $ micropython
   MicroPython v1.11-44-g8b18cfede-dirty on 2019-06-14; linux version
   Use Ctrl-D to exit, Ctrl-E for paste mode
   >>> from mpbf import *
   >>> import gc
   >>> mpbf.init()
   >>> mpbf.set_params(500, 0)  # precision digits, rounding mode
   >>> gc.collect() #required for microcontrollers if mpbf.sop is called in a loop, else realloc calls can return NULL
   >>> mpbf.sop('1', '0', 5)    # 5 is exponentiation operation -- see bf.h
   '2.7182818284590452353602874713526624977572470936999595749669676277240766303535475945713821785251664274274663919320
   0305992181741359662904357290033429526059563073813232862794349076323382988075319525101901157383418793070215408914993
   4884167509244761460668082264800168477411853742345442437107539077744992069551702761838606261331384583000752044933826
   5602976067371132007093287091274437470472306969772093101416928368190255151086574637721112523897844250569536967707854
   4996996794686445490598793163688923009879313s-16'
   >>> mpbf.set_params(5, 0)
   >>> gc.collect() #required for microcontrollers if mpbf.sop is called in a loop, else realloc calls can return NULL
   >>> mpbf.sop('1', '0', 5)
   '2.71828s-16'
   >>> mpbf.finish() #deallocate memory

The returned string always ends with s<a signed number> -- this indicates the status returned by libbf and in most cases can be discarded.

Functions implemented

Function	Operator value Passed to sop ()
BF_OP_ADD	0
BF_OP_MUL	1
BF_OP_SUB	2
BF_OP_DIV	3
BF_OP_POW	4
BF_OP_EXP	5
BF_OP_LOG	6
BF_OP_SIN	7
BF_OP_COS	8
BF_OP_TAN	9
BF_OP_ATAN	10
BF_OP_ATAN2	11
BF_OP_ACOS	12
BF_OP_ASIN	13
BF_OP_SQRT	14
BF_OP_IDIV	15
BF_OP_REM	16
BF_OP_RINT	17
BF_OP_ROUND	18
BF_OP_EQ	19
BF_OP_LT	20
BF_OP_LE	21
BF_OP_PI	22

MPAPBF usage

The number format is defined in MPAP. It's not necessary to use MPAP for availing arbitrary precision functions provided by MPBF/libbf, but it helps as a numerical representation of floating point numbers instead of strings.

Keep mpapbf.py in a directory listed in a standard sys.path.

   $ micropython
   MicroPython v1.11-44-g8b18cfede-dirty on 2019-06-14; linux version
   Use Ctrl-D to exit, Ctrl-E for paste mode
   >>> from mpapbf import *
   >>> mpap(1).exp()
   mpap(Mantissa = 27182818284590452353602874714, Exponent = 0, InternalAware = True)
   >>> sprec(500) #set precision to 500
   >>> mpap(1).exp()
   mpap(Mantissa = 2718281828459045235360287471352662497757247093699959574966967627724076630353547594571382178525166427427466    
   3919320030599218174135966290435729003342952605956307381323286279434907632338298807531952510190115738341879
   3070215408914993488416750924476146066808226480016847741185374234544243710753907774499206955170276183860626
   1331384583000752044933826560297606737113200709328709127443747047230696977209310141692836819025515108657463
   77211125238978442505695369677078544996996794686445490598793163688923009879313, Exponent = 0, InternalAware = True)
   >>> rprec() #reset precision back to default 27
   >>> mpap(1).exp()
   mpap(Mantissa = 27182818284590452353602874714, Exponent = 0, InternalAware = True)

Functions supported in MPAPBF

MPBF Function	Python operator or function
BF_OP_ADD	+
BF_OP_MUL	*
BF_OP_SUB	-
BF_OP_DIV	/
BF_OP_POW	**
BF_OP_EXP	mpap(value).exp()
BF_OP_LOG	mpap(value).log()
BF_OP_SIN	mpap(value).sin()
BF_OP_COS	mpap(value).cos()
BF_OP_TAN	mpap(value).tan()
BF_OP_ATAN	mpap(value).atan()
BF_OP_ATAN2	mpap(value).atan2()
BF_OP_ACOS	mpap(value).acos()
BF_OP_ASIN	mpap(value).asin()
BF_OP_SQRT	mpap(value).sqrt()
BF_OP_IDIV	//
BF_OP_REM	%
BF_OP_RINT	Not implemented
BF_OP_ROUND	Not implemented
BF_OP_EQ	==
BF_OP_LT	<
BF_OP_LE	<=
BF_OP_PI	mpap(value).pi(). Returns value*pi

Known Issues

1. The factorization factors() function in mpapbf.py has a bug.