Home My Page Projects Code Snippets Project Openings diderot

# SCM Repository

[diderot] Diff of /branches/pure-cfg/TODO
 [diderot] / branches / pure-cfg / TODO

# Diff of /branches/pure-cfg/TODO

revision 788, Fri Apr 8 03:57:23 2011 UTC revision 1133, Thu May 5 18:57:15 2011 UTC
# Line 1  Line 1
1  ========================  NOTE: GLK's approximate ranking of 5 most important tagged with
2  SHORT TERM =============  (for curvature-based VR)  [GLK:1], [GLK:2], ...
========================
3
4  IL support for higher-order tensor values (matrices, etc).  ==============================
5      tensor construction  other SHORT TERM =============  (including needed for LIC)
6      tensor indexing and slicing  ==============================
verify that hessians work correctly

Outer products
syntax [DONE]
typechecking
IL and codegen

Add M dot v, v dot M, and M dot N

Identity matrix [DONE up to code generation]

trace [DONE]

extend norm (|exp|) to tensor[3,3]
7
8  Use ∇⊗ etc. syntax  Add a clamp function, which takes three arguments; either three scalars:
9      syntax [DONE]    clamp(x, minval, maxval)  = max(minval, min(maxval, x))
10      typechecking  or three vectors of the same size:
11      IL and codegen    clamp([x,y], minvec, maxvec)  = [max(minvec[0], min(maxvec[0], x)),
12                                       max(minvec[1], min(maxvec[1], y))]
13    This would be useful in many current Diderot programs.
14    One question: clamp(x, minval, maxval) is the argument order
15    used in OpenCL and other places, but clamp(minval, maxval, x)
16    would be more consistent with lerp(minout, maxout, x).
17
18    Level of differentiability in field type should be statement about how
19    much differentiation the program *needs*, rather than what the kernel
20    *provides*.  The needed differentiability can be less than or equal to
21    the provided differentiability.
22
23  Add sequence types  [GLK:1] Add sequence types (needed for evals & evecs)
24      syntax      syntax
25          types: ty '{' INT '}'          types: ty '{' INT '}'
26          value construction: '{' e1 ',' … ',' en '}'          value construction: '{' e1 ',' … ',' en '}'
27          indexing: e '{' e '}'          indexing: e '{' e '}'
28
29  Infix dot product and cross product [DONE]  IL support for higher-order tensor values (matrices, etc).
30        tensor construction [DONE]
31        tensor indexing [DONE]
32        tensor slicing
33        verify that hessians work correctly [DONE]
34
35  lerp on scalars and vectors [DONE]  Use ∇⊗ etc. syntax
36        syntax [DONE]
37        typechecking
38        IL and codegen
39
40  ==============================  test/uninit.diderot:
41  other SHORT TERM =============  (including needed for LIC)  documents need for better compiler error messages when output variables
42  ==============================  are not initialized; the current messages are very cryptic
43
44  determinant ("det") for tensor[3,3]  determinant ("det") for tensor[3,3]
45
Decide if we want to allow redefinitions of variables
(as in vr-lite-cam.diderot)

extend norm (|exp|) to all tensor types

extend normalize to all tensor types

vector fields: convolution on vector images

46  expand trace in mid to low translation  expand trace in mid to low translation
47
48  value-numbering optimization  value-numbering optimization
# Line 59  Line 52
52      rgba = real{4}      rgba = real{4}
53
54  ==============================  ==============================
55  MEDIUM TERM ================== (needed for streamlines & tractography)  MEDIUM TERM ================== (including needed for streamlines & tractography)
56  ==============================  ==============================
57
58    [GLK:1] evals & evecs for symmetric tensor[3,3] (requires sequences)
59
60    [GLK:2] Save Diderot output to nrrd, instead of "mip.txt"
61      For grid of strands, save to similarly-shaped array
62      For list of strands, save to long 1-D (or 2-D for non-scalar output) list
63      For ragged things (like tractography output), will need to save both
64        complete list of values, as well as list of start indices and lengths
65        to index into complete list
66
67    [GLK:3] Use of Teem's "hest" command-line parser for getting
68    any input variables that are not defined in the source file
69
70    [GLK:4] ability to declare a field so that probe positions are
71    *always* "inside"; with various ways of mapping the known image values
72    to non-existant index locations.  One possible syntax emphasizes that
73    there is a index mapping function that logically precedes convolution:
74      F = bspln3 ⊛ (img ◦ clamp)
75      F = bspln3 ⊛ (img ◦ repeat)
76      F = bspln3 ⊛ (img ◦ mirror)
77    where "◦" or "∘" is used to indicate function composition
78
79    extend norm (|exp|) to all tensor types [DONE for vectors and matrices]
80
81  ability to emit/track/record variables into dynamically re-sized  ability to emit/track/record variables into dynamically re-sized
82  runtime buffer  runtime buffer
83
84  evals & evecs for symmetric tensor[3,3]  Want: allow X *= Y, X /= Y, X += Y, X -= Y to mean what they do in C,
85    provided that X*Y, X/Y, X+Y, X-Y are already supported.
86    Nearly every Diderot program would be simplified by this.
87
88    Want: non-trivial field expressions & functions:
90      field#0(2)[] Vlen = |Vimg ⊛ bspln3|;
91    to get a scalar field of vector length, or
92      field#2(2)[] F = Fimg ⊛ bspln3;
93      field#0(2)[] Gmag = |∇F|;
94    to get a scalar field of gradient magnitude, or
95      field#2(2)[] F = Fimg ⊛ bspln3;
96      field#0(2)[] Gmsq = ∇F•∇F;
97    to get a scalar field of squared gradient magnitude, which is simpler
98    to differentiate.  However, there is value in having these, even if
99    the differentiation of them is not supported (hence the indication
100    of "field#0" for these above)
101
102    Want: ability to apply "normalize" to a field itself, e.g.
103      field#0(2)[2] V = normalize(Vimg ⊛ ctmr);
104    so that V(x) = normalize((Vimg ⊛ ctmr)(x)).
105    Having this would simplify expression of standard LIC method, and
106    would also help express other vector field expressions that arise
107    in vector field feature exraction.
108
109  tensor fields: convolution on general tensor images  tensor fields: convolution on general tensor images
110
# Line 73  Line 112
112  other MEDIUM TERM ============ (needed for particles)  other MEDIUM TERM ============ (needed for particles)
113  ==============================  ==============================
114
115  run-time birth and death of strands  Put small 1-D and 2-D fields, when reconstructed specifically by tent
116    and when differentiation is not needed, into faster texture buffers.
117    test/illust-vr.diderot is good example of program that uses multiple
118    such 1-D fields basically as lookup-table-based function evaluation
119
120    run-time birth of strands
121
122  "initially" supports lists  "initially" supports lists
123
# Line 84  Line 128
128
129  proper handling of stabilize method  proper handling of stabilize method
130
131  ==============================  test/vr-kcomp2.diderot: Add support for code like
LONG TERM ====================
==============================
132
133  co- vs contra- index distinction          (F1 if x else F2)@pos
134
135    This will require duplication of the continuation of the conditional
136    (but we should only duplicate over the live-range of the result of the
137    conditional.
138
139  add ":" for tensor dot product (contracts out two indices  add ":" for tensor dot product (contracts out two indices
140  instead of one like •), valid for all pairs of tensors with  instead of one like •), valid for all pairs of tensors with
141  at least two indices  at least two indices
142
143    ==============================
144    other MEDIUM TERM ============
145    ==============================
146
147    want: warnings when "D" (reserved for differentiation) is declared as
148    a variable name (get confusing error messages now)
149
150    support for Python interop and GUI
151
152    Python/ctypes interface to run-time
153
154    ==============================
155    LONG TERM ====================
156    ==============================
157
158    Better handling of variables that determines the scope of a variable
159    based on its actual use, instead of where the user defined it.  So,
160    for example, we should lift strand-invariant variables to global
161    scope.  Also prune out useless variables, which should include field
162    variables after the translation to mid-il.
163
164    co- vs contra- index distinction
165
166  some indication of tensor symmetry  some indication of tensor symmetry
167  (have to identify the group of index permutations that are symmetries)  (have to identify the group of index permutations that are symmetries)
168
# Line 109  Line 178
178  BUGS =================  BUGS =================
179  ======================  ======================
180
182      // generated C code looks like "(float)p_ui_0--0.5e0f"  // HEY (BUG?) shouldn't it be a type error to load this 2-D array of
183      //real rayU = 2.0*camUmax*(real(ui) - -0.5)/real(imgResU) - camUmax;  // 2-vectors into a 2-D *scalar* field?  Instead, get:
184    //   uncaught exception Fail [Fail: Error in compiling lic.diderot]
185  test/vr-kcomp.diderot:  //     raised at driver/main.sml:31.39-31.76
186      // HEY (scoping BUG): the strand parameters (e.g. ui, vi) ...  image(2)[] Vimg = load("../data/vorttest.nrrd");

test/zslice2.diderot:
// but is this not valid syntax for creating field in one shot?
//field#2(3)[] F = (load("../data/zimg112.nrrd")) ⊛ bspln3;
// It is valid syntax, but there is a bug in the conversion from HighIL to MidIL
187
188  test/zslice2.diderot:  test/zslice2.diderot:
189  // HEY (bug) bspln5 leads to problems ...  // HEY (bug) bspln5 leads to problems ...
190  //  uncaught exception Size [size]  //  uncaught exception Size [size]
191  //    raised at c-target/c-target.sml:47.15-47.19  //    raised at c-target/c-target.sml:47.15-47.19
192  //field#4(3)[] F = img ⊛ bspln5;  //field#4(3)[] F = img ⊛ bspln5;

Legend:
 Removed from v.788 changed lines Added in v.1133