Skip to content

Commit 1ba2eaf

Browse files
siko1056siko1056
committed
Test release v0.1.6
1 parent 2a0304f commit 1ba2eaf

File tree

3 files changed

+259
-22
lines changed

3 files changed

+259
-22
lines changed

‎.gitignore‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -6,3 +6,5 @@ mpfr.h
66
mpf2mpfr.h
77
.build
88
.DS_Store
9+
.ipynb_checkpoints/
10+

‎README.ipynb‎

Lines changed: 230 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,230 @@
1+
{
2+
"cells": [
3+
{
4+
"cell_type": "markdown",
5+
"id": "cb515e2a-7f48-454a-8962-3f604d3eb00b",
6+
"metadata": {},
7+
"source": [
8+
"# APA - Octave/Matlab arbitrary precision arithmetic."
9+
]
10+
},
11+
{
12+
"cell_type": "markdown",
13+
"id": "3d298073-4ecd-4582-8f11-e9ddd8a8f193",
14+
"metadata": {},
15+
"source": [
16+
"## Installation\n",
17+
"\n",
18+
"From the Octave command-line run:"
19+
]
20+
},
21+
{
22+
"cell_type": "code",
23+
"execution_count": null,
24+
"id": "f9668ec5-51e5-4bb7-bca7-baffc620e717",
25+
"metadata": {},
26+
"outputs": [],
27+
"source": [
28+
"pkg install 'https://github.com/gnu-octave/apa/releases/download/v0.1.5/apa-0.1.5.zip'\n",
29+
"pkg load apa\n",
30+
"pkg test apa"
31+
]
32+
},
33+
{
34+
"cell_type": "markdown",
35+
"id": "6dbc0ecb-238e-4496-86c6-4b5fefb96467",
36+
"metadata": {},
37+
"source": [
38+
"From the Matlab command-line run (also works for Octave):"
39+
]
40+
},
41+
{
42+
"cell_type": "code",
43+
"execution_count": null,
44+
"id": "66b7afda-6ee7-4241-8c85-482c03d34b1e",
45+
"metadata": {},
46+
"outputs": [],
47+
"source": [
48+
"urlwrite ('https://github.com/gnu-octave/apa/releases/download/v0.1.5/apa-0.1.5.zip', ...\n",
49+
" 'apa-0.1.5.zip');\n",
50+
"unzip ('apa-0.1.5.zip');\n",
51+
"cd (fullfile ('apa-0.1.5', 'inst'))\n",
52+
"install_apa\n",
53+
"test_apa"
54+
]
55+
},
56+
{
57+
"cell_type": "markdown",
58+
"id": "24925e83-23be-48d1-b1e1-ed14141d574f",
59+
"metadata": {},
60+
"source": [
61+
"## High-level MPFR Interface\n",
62+
"\n",
63+
"The high-level MPFR interface is given through the `@mpfr_t` class.\n",
64+
"A variable of that type \"behaves\" like a \"normal\" built-in Octave/Matlab\n",
65+
"data type."
66+
]
67+
},
68+
{
69+
"cell_type": "code",
70+
"execution_count": 3,
71+
"id": "08296fd1-44fb-4719-8aea-62823fcedfc7",
72+
"metadata": {},
73+
"outputs": [
74+
{
75+
"name": "stdout",
76+
"output_type": "stream",
77+
"text": [
78+
"rop =\n",
79+
"\n",
80+
" MPFR 3x3 matrix (precision 53 binary digits)\n",
81+
"\n",
82+
" Double approximation:\n",
83+
"\n",
84+
" 5 1 1\n",
85+
" 1 5 1\n",
86+
" 1 1 5\n",
87+
"\n"
88+
]
89+
}
90+
],
91+
"source": [
92+
"op1 = mpfr_t (eye (3) * 4);\n",
93+
"\n",
94+
"rop = op1 + 1"
95+
]
96+
},
97+
{
98+
"cell_type": "markdown",
99+
"id": "0823de62-cf4a-4799-b066-2791bc862f41",
100+
"metadata": {},
101+
"source": [
102+
"The high-level MPFR interface is the preferred choice for quick numerical\n",
103+
"experiments.\n",
104+
"\n",
105+
"However, if performance is more critical, please use the low-level MPFR\n",
106+
"interface (explained below) and vectorization wherever possible.\n",
107+
"\n",
108+
"> Please note that an interface from an interpreted high-level programming\n",
109+
"> language like Octave/Matlab is most likely slower than a pre-compiled C\n",
110+
"> program.\n",
111+
">\n",
112+
"> If performance is highly-critical, use this tool for initial experiments\n",
113+
"> and translate the developed algorithm to native MPFR C-code."
114+
]
115+
},
116+
{
117+
"cell_type": "markdown",
118+
"id": "39dcbdfa-4634-40f9-8a45-1f01ae1611bb",
119+
"metadata": {},
120+
"source": [
121+
"## Low-level MPFR Interface\n",
122+
"\n",
123+
"> For information how to compile/develop the interface, see\n",
124+
"> [`doc/MEX_INTERFACE.md`](https://github.com/gnu-octave/apa/blob/main/doc/MEX_INTERFACE.md).\n",
125+
"\n",
126+
"The low-level MPFR interface permits efficient access to almost all functions\n",
127+
"specified by MPFR 4.1.0 <https://www.mpfr.org/mpfr-4.1.0/mpfr.html>.\n",
128+
"\n",
129+
"All supported functions are [listed in the `inst` folder](inst)\n",
130+
"and can be called from Octave/Matlab like in the C programming language.\n",
131+
"\n",
132+
"For example, the C function:\n",
133+
"\n",
134+
"```c\n",
135+
"int mpfr_add (mpfr_t rop, mpfr_t op1, mpfr_t op2, mpfr_rnd_t rnd)\n",
136+
"```\n",
137+
"\n",
138+
"can be called from Octave/Matlab with scalar, vector, or matrix quantities:"
139+
]
140+
},
141+
{
142+
"cell_type": "code",
143+
"execution_count": 4,
144+
"id": "c0dfe5e1-4eda-4b3f-b779-785ed66ad09d",
145+
"metadata": {},
146+
"outputs": [
147+
{
148+
"name": "stdout",
149+
"output_type": "stream",
150+
"text": [
151+
"rop =\n",
152+
"\n",
153+
" MPFR 3x3 matrix (precision 53 binary digits)\n",
154+
"\n",
155+
" Double approximation:\n",
156+
"\n",
157+
" 6 2 2\n",
158+
" 2 6 2\n",
159+
" 2 2 6\n",
160+
"\n"
161+
]
162+
}
163+
],
164+
"source": [
165+
"% Prepare input and output variables.\n",
166+
"rop = mpfr_t (zeros (3));\n",
167+
"op1 = mpfr_t (eye (3) * 4);\n",
168+
"op2 = mpfr_t (2);\n",
169+
"rnd = mpfr_get_default_rounding_mode ();\n",
170+
"\n",
171+
"% Call mpfr_add. Note unlike Octave/Matlab the\n",
172+
"% left-hand side does NOT contain the result.\n",
173+
"ret = mpfr_add (rop, op1, op2, rnd);\n",
174+
"\n",
175+
"rop % Note rop vs. ret!"
176+
]
177+
},
178+
{
179+
"cell_type": "markdown",
180+
"id": "f3002580-752c-499c-b31e-f8fa8b17fb22",
181+
"metadata": {},
182+
"source": [
183+
"In the low-level interface the type checks are stricter,\n",
184+
"but scalar and matrix quantities can still be mixed.\n",
185+
"\n",
186+
"Another benefit of using the low-level MPFR interface is that **in-place**\n",
187+
"operations are permitted, which do not create new (temporary) variables:"
188+
]
189+
},
190+
{
191+
"cell_type": "code",
192+
"execution_count": 5,
193+
"id": "92886baf-f06e-43ae-8108-cc3e9cce2c15",
194+
"metadata": {},
195+
"outputs": [],
196+
"source": [
197+
"ret = mpfr_add (op1, op1, op1, rnd); % op1 += op1"
198+
]
199+
}
200+
],
201+
"metadata": {
202+
"kernelspec": {
203+
"display_name": "Octave",
204+
"language": "octave",
205+
"name": "octave"
206+
},
207+
"language_info": {
208+
"file_extension": ".m",
209+
"help_links": [
210+
{
211+
"text": "GNU Octave",
212+
"url": "https://www.gnu.org/software/octave/support.html"
213+
},
214+
{
215+
"text": "Octave Kernel",
216+
"url": "https://github.com/Calysto/octave_kernel"
217+
},
218+
{
219+
"text": "MetaKernel Magics",
220+
"url": "https://metakernel.readthedocs.io/en/latest/source/README.html"
221+
}
222+
],
223+
"mimetype": "text/x-octave",
224+
"name": "octave",
225+
"version": "6.4.0"
226+
}
227+
},
228+
"nbformat": 4,
229+
"nbformat_minor": 5
230+
}

‎README.md‎

Lines changed: 27 additions & 22 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,7 @@
44

55
From the Octave command-line run:
66

7+
78
```octave
89
pkg install 'https://github.com/gnu-octave/apa/releases/download/v0.1.5/apa-0.1.5.zip'
910
pkg load apa
@@ -12,7 +13,8 @@ pkg test apa
1213

1314
From the Matlab command-line run (also works for Octave):
1415

15-
```matlab
16+
17+
```octave
1618
urlwrite ('https://github.com/gnu-octave/apa/releases/download/v0.1.5/apa-0.1.5.zip', ...
1719
'apa-0.1.5.zip');
1820
unzip ('apa-0.1.5.zip');
@@ -27,23 +29,24 @@ The high-level MPFR interface is given through the `@mpfr_t` class.
2729
A variable of that type "behaves" like a "normal" built-in Octave/Matlab
2830
data type.
2931

32+
3033
```octave
3134
op1 = mpfr_t (eye (3) * 4);
3235
3336
rop = op1 + 1
3437
```
3538

36-
```
37-
rop =
38-
39-
MPFR 3x3 matrix (precision 53 binary digits)
39+
rop =
40+
41+
MPFR 3x3 matrix (precision 53 binary digits)
42+
43+
Double approximation:
44+
45+
5 1 1
46+
1 5 1
47+
1 1 5
48+
4049

41-
Double approximation:
42-
43-
5 1 1
44-
1 5 1
45-
1 1 5
46-
```
4750

4851
The high-level MPFR interface is the preferred choice for quick numerical
4952
experiments.
@@ -58,7 +61,6 @@ interface (explained below) and vectorization wherever possible.
5861
> If performance is highly-critical, use this tool for initial experiments
5962
> and translate the developed algorithm to native MPFR C-code.
6063
61-
6264
## Low-level MPFR Interface
6365

6466
> For information how to compile/develop the interface, see
@@ -78,6 +80,7 @@ int mpfr_add (mpfr_t rop, mpfr_t op1, mpfr_t op2, mpfr_rnd_t rnd)
7880
7981
can be called from Octave/Matlab with scalar, vector, or matrix quantities:
8082
83+
8184
```octave
8285
% Prepare input and output variables.
8386
rop = mpfr_t (zeros (3));
@@ -92,23 +95,25 @@ ret = mpfr_add (rop, op1, op2, rnd);
9295
rop % Note rop vs. ret!
9396
```
9497

95-
```
96-
rop =
97-
98-
MPFR 3x3 matrix (precision 53 binary digits)
98+
rop =
99+
100+
MPFR 3x3 matrix (precision 53 binary digits)
101+
102+
Double approximation:
103+
104+
6 2 2
105+
2 6 2
106+
2 2 6
107+
99108

100-
Double approximation:
101-
102-
6 2 2
103-
2 6 2
104-
2 2 6
105-
```
106109

107110
In the low-level interface the type checks are stricter,
108111
but scalar and matrix quantities can still be mixed.
109112

110113
Another benefit of using the low-level MPFR interface is that **in-place**
111114
operations are permitted, which do not create new (temporary) variables:
115+
116+
112117
```octave
113118
ret = mpfr_add (op1, op1, op1, rnd); % op1 += op1
114119
```

0 commit comments

Comments
 (0)