(DM42) Matrix exponential

[Gjermund Skailand]

Here is another version of Alberts code for the DM42.

There are two tuning parameters, to make use of the 32 digits accuracy of the DM32. Current values should give same result as Werners version.

at line 24, increase from 10 for more reduction of input matrix
at line 34, increase from 4 to e.g. 9, error ~ 0.00138^9/10!= 5.0E-33
If you want to calculate MEXP-1 delete last line (line 60)

Code:


00 { 103-Byte Prgm }
01▸LBL "MEXP"      @ Extended Albert version
02 FUNC 11            @ has two tuning parameters
03 L4STK               @ uses R00
04 ENTER               @ delete line 60  for MEXP-1
05 DIM?
06 NEWMAT
07 EDIT
08 ←
09▸LBL 00
10 ↓
11 SIGN
12 →
13 FC? 77
14 GTO 00
15 EXITALL
16 X<>Y
17 ENTER
18 FNRM
19 LN
20 2
21 LN
22 ÷
23 IP
24 10               @ parameter for halving
25 +
26 X<0?
27 CLX
28 STO 00
29 2
30 X<>Y
31 Y↑X
32 ÷
33 RCL ST X
34 4                @ parameter for reverse series expansion
35 ÷
36 LASTX
37 DSE ST X
38 DSE ST X
39 R↓
40 RCL+ ST Z
41▸LBL 01
42 ISG ST T
43 NOP
44 RCL÷ ST T
45 RCL× ST Y
46 RCL+ ST Z
47 DSE ST T
48 DSE ST T
49 GTO 01
50 X<>Y
51 ×
52▸LBL 02
53 ENTER
54 ×
55 LASTX
56 STO+ ST X
57 +
58 DSE 00
59 GTO 02
60 +
61 END

Best regards
Gjermund