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 763, Thu Apr 7 19:57:56 2011 UTC revision 962, Sat Apr 23 19:33:19 2011 UTC
# Line 1  Line 1
Outer products
syntax [DONE]
typechecking
IL and codegen
1
2  Add M dot v, v dot M, and M dot N  NOTE: GLK's approximate ranking of 6 most imporant tagged with
3    [GLK:1], [GLK:2], ...
4
5  Identity matrix [DONE up to code generation]  ========================
6    SHORT TERM =============  (for curvature-based VR)
7    ========================
8
9  trace [DONE]  IL support for higher-order tensor values (matrices, etc).
10        tensor construction [DONE]
11        tensor indexing [DONE]
12        tensor slicing
13        verify that hessians work correctly [DONE]
14
15  expand trace in mid to low translation  Outer products [DONE]
16
17  extend norm (|exp|) to all tensor types  Add M dot v, v dot M, and M dot N [DONE]
18
19  Infix dot product and cross product [DONE]  Identity matrix [DONE]
20
21  lerp on scalars and vectors [DONE]  Zero tensor [DONE]
22
23    trace [DONE]
24
25  Use ∇⊗ etc syntax  extend norm (|exp|) to tensor[3,3] [DONE]
26
27    Matrix addition, subtraction, and scaling  [DONE]
28
29    Use ∇⊗ etc. syntax
30      syntax [DONE]      syntax [DONE]
31      typechecking      typechecking
32      IL and codegen      IL and codegen
# Line 28  Line 37
37          value construction: '{' e1 ',' … ',' en '}'          value construction: '{' e1 ',' … ',' en '}'
38          indexing: e '{' e '}'          indexing: e '{' e '}'
39
40  IL support for higher-order tensor values (matrices, etc).  Infix dot product and cross product [DONE]
41      tensor construction
42      tensor indexing and slicing  lerp on scalars and vectors [DONE]
43
44    Infix "^" operator for pow() [DONE]
45
46    Code generation support for 1D image data, such as RGBA transfer functions [DONE]
47
48    ==============================
49    other SHORT TERM =============  (including needed for LIC)
50    ==============================
51
52    test/uninit.diderot:
53    documents need for better compiler error messages when output variables
54    are not initialized; the current messages are very cryptic
55
56    determinant ("det") for tensor[3,3]
57
58    extend norm (|exp|) to all tensor types [DONE for vectors and matrices]
59
60    expand trace in mid to low translation
61
62    value-numbering optimization
63
64  Add type aliases for color types  Add type aliases for color types
65      rgb = real{3}      rgb = real{3}
66      rgba = real{4}      rgba = real{4}
67
68    ==============================
69    MEDIUM TERM ================== (including needed for streamlines & tractography)
70    ==============================
71
72    [GLK:3] evals & evecs for symmetric tensor[3,3] (requires sequences)
73
74    [GLK:4] Save Diderot output to nrrd, instead of "mip.txt"
75      For grid of strands, save to similarly-shaped array
76      For list of strands, save to long 1-D (or 2-D for non-scalar output) list
77      For ragged things (like tractography output), will need to save both
78        complete list of values, as well as list of start indices and lengths
79        to index into complete list
80
81    [GLK:5] Use of Teem's "hest" command-line parser for getting
82    any input variables that are not defined in the source file
83
84    [GLK:6] ability to declare a field in such a way so that probe
85    positions are *always* clamped to the support of "inside";
86    there are many cases where this is the sensible behavior.
87    (More generally, we could also have "repeat" declaration,
88    copying action of GL_REPEAT in texturing)
89
90    ability to emit/track/record variables into dynamically re-sized
91    runtime buffer
92
93    Want: allow X *= Y, X /= Y, X += Y, X -= Y to mean what they do in C,
94    provided that X*Y, X/Y, X+Y, X-Y are already supported.
95    Nearly every Diderot program would be simplified by this.
96
97    Want: non-trivial field expressions & functions:
99      field#0(2)[] Vlen = |Vimg ⊛ bspln3|;
100    to get a scalar field of vector length, or
101      field#2(2)[] F = Fimg ⊛ bspln3;
102      field#0(2)[] Gmag = |∇F|;
103    to get a scalar field of gradient magnitude, or
104      field#2(2)[] F = Fimg ⊛ bspln3;
105      field#0(2)[] Gmsq = ∇F•∇F;
106    to get a scalar field of squared gradient magnitude, which is simpler
107    to differentiate.  However, there is value in having these, even if
108    the differentiation of them is not supported (hence the indication
109    of "field#0" for these above)
110
111    Want: ability to apply "normalize" to a field itself, e.g.
112      field#0(2)[2] V = normalize(Vimg ⊛ ctmr);
113    so that V(x) = normalize((Vimg ⊛ ctmr)(x)).
114    Having this would simplify expression of standard LIC method, and
115    would also help express other vector field expressions that arise
116    in vector field feature exraction.
117
118    tensor fields: convolution on general tensor images
119
120    ==============================
121    other MEDIUM TERM ============ (needed for particles)
122    ==============================
123
124    run-time birth and death of strands
125
126    "initially" supports lists
127
128    "initially" supports lists of positions output from
129    different initalization Diderot program
130
131    spatial data structure that permits strands' queries of neighbors
132
133    proper handling of stabilize method
134
135    test/vr-kcomp2.diderot: Add support for code like
136
137            (F1 if x else F2)@pos
138
139    This will require duplication of the continuation of the conditional
140    (but we should only duplicate over the live-range of the result of the
141    conditional.
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    ==============================
153    LONG TERM ====================
154    ==============================
155
156    Better handling of variables that determines the scope of a variable
157    based on its actual use, instead of where the user defined it.  So,
158    for example, we should lift strand-invariant variables to global
159    scope.  Also prune out useless variables, which should include field
160    variables after the translation to mid-il.
161
162    co- vs contra- index distinction
163
164    add ":" for tensor dot product (contracts out two indices
165    instead of one like •), valid for all pairs of tensors with
166    at least two indices
167
168    some indication of tensor symmetry
169    (have to identify the group of index permutations that are symmetries)
170
171    dot works on all tensors
172
173    outer works on all tensors
174
175    Einstein summation notation
176
177    "tensor comprehension" (like list comprehension)
178
179    Python/ctypes interface to run-timez
180
181    ======================
182    BUGS =================
183    ======================
184
185    [GLK:1] test/derivs2.diderot:
186    test/derivs3.diderot:
187    gradients & Hessians are not being transformed from index-to-world,
188    these test programs should produce black images when that is working
189
190    [GLK:2] test/fields.diderot:
191    documents various bugs/questions associated with simple field expressions
193
195    // HEY (BUG?) shouldn't it be a type error to load this 2-D array of
196    // 2-vectors into a 2-D *scalar* field?  Instead, get:
197    //   uncaught exception Fail [Fail: Error in compiling lic.diderot]
198    //     raised at driver/main.sml:31.39-31.76
200
201    test/iso2d.diderot: not able to use "nan".
202              // HEY (BUG) not able to use NaN as in "pos = [nan,nan]"
203              // generated C code "vec2f(nanf, nanf)" causes problems:
204              // iso2d.c: In function ‘sample_update’:
205              // iso2d.c:86: error: incompatible type for argument 1 of ‘vec2f’
206              // iso2d.c:86: error: incompatible type for argument 2 of ‘vec2f’
207              // uncaught exception Fail [Fail: error compiling/linking]
208              //   raised at c-target/c-target.sml:323.14-323.44
209              pos = [∞,∞];  // should be: pos = [nan,nan];
210
211    test/mip-cam.diderot:
212        // generated C code looks like "(float)p_ui_0--0.5e0f"
213        //real rayU = 2.0*camUmax*(real(ui) - -0.5)/real(imgResU) - camUmax;
214
215    test/vr-kcomp.diderot:
216        // HEY (scoping BUG): the strand parameters (e.g. ui, vi) ...
217
218    test/zslice2.diderot:
219    // but is this not valid syntax for creating field in one shot?
220    //field#2(3)[] F = (load("../data/zimg112.nrrd")) ⊛ bspln3;
221    // It is valid syntax, but there is a bug in the conversion from HighIL to MidIL
222
223    test/zslice2.diderot:
224    // HEY (bug) bspln5 leads to problems ...
225    //  uncaught exception Size [size]
226    //    raised at c-target/c-target.sml:47.15-47.19
227    //field#4(3)[] F = img ⊛ bspln5;

Legend:
 Removed from v.763 changed lines Added in v.962