Home My Page Projects Code Snippets Project Openings diderot
Summary Activity Tracker Tasks SCM

SCM Repository

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

Diff of /branches/pure-cfg/TODO

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 824, Thu Apr 14 15:54:24 2011 UTC revision 1304, Fri Jun 10 21:44:33 2011 UTC
# Line 1  Line 1 
1    NOTE: GLK's approximate ranking of 8 most important tagged with
2    [GLK:1], [GLK:2], ...
4  ========================  ========================
5  SHORT TERM =============  (for curvature-based VR)  SHORT TERM ============= (*needed* for streamlines & tractography)
6  ========================  ========================
8  For RGB color output of 2D transfer function (indexed by  Remove CL from compiler
 kappa1,kappa2), or for a 1D transfer function, will need  
 something akin to convolution on vector images  
 IL support for higher-order tensor values (matrices, etc).  
     tensor construction [DONE]  
     tensor indexing [DONE]  
     tensor slicing  
     verify that hessians work correctly  
 Outer products [DONE]  
 Add M dot v, v dot M, and M dot N [DONE]  
 Identity matrix [DONE up to code generation]  
 Zero tensor [DONE up to code generation]  
 trace [DONE]  
 extend norm (|exp|) to tensor[3,3]  
 extend implementation of scaling to work on higher-order tensors (it currently  
 just works on vectors)  
 Use ∇⊗ etc. syntax  
     syntax [DONE]  
     IL and codegen  
10  Add sequence types  [GLK:3] Add sequence types (needed for evals & evecs)
11      syntax      syntax
12          types: ty '{' INT '}'          types: ty '{' INT '}'
13          value construction: '{' e1 ',' … ',' en '}'          value construction: '{' e1 ',' … ',' en '}'
14          indexing: e '{' e '}'          indexing: e '{' e '}'
16  Infix dot product and cross product [DONE]  [GLK:4] evals & evecs for symmetric tensor[2,2] and
17    tensor[3,3] (requires sequences)
 lerp on scalars and vectors [DONE]  
 Infix "^" operator for pow() [DONE]  
 other SHORT TERM =============  (including needed for LIC)  
19  determinant ("det") for tensor[3,3]  ability to emit/track/record variables into dynamically re-sized
20    runtime buffer
22  Decide if we want to allow redefinitions of variables  tensor fields: convolution on general tensor images
   (as in vr-lite-cam.diderot)  
24  extend norm (|exp|) to all tensor types  ========================
25    SHORT-ISH TERM ========= (to make using Diderot less annoying to
26    ========================  program in, and slow to execute)
28  extend normalize to all tensor types  value-numbering optimization [DONE]
30  vector fields: convolution on vector images  Allow ".ddro" file extensions in addition to ".diderot"
32  expand trace in mid to low translation  Be able to output values of type tensor[2,2] and tensor[3,3];
33    (currently only scalars & vectors).  Want to add some regression tests
34    based on this and currently can't
36    [GLK:1] Add a clamp function, which takes three arguments; either
37    three scalars:
38      clamp(lo, hi, x)  = max(lo, min(hi, x))
39    or three vectors of the same size:
40      clamp(lo, hi, [x,y])  = [max(lo[0], min(hi[0], x)),
41                               max(lo[1], min(hi[1], y))]
42    This would be useful in many current Diderot programs.
43    One question: clamp(x, lo, hi) is the argument order used in OpenCL
44    and other places, but clamp(lo, hi, x) is much more consistent with
45    lerp(lo, hi, x), hence GLK's preference
46    [DONE]
48    [GLK:2] Proper handling of stabilize method
50    allow "*" to represent "modulate": per-component multiplication of
51    vectors, and vectors only (not tensors of order 2 or higher).  Once
52    sequences are implemented this should be removed: the operation is not
53    invariant WRT basis so it is not a legit vector computation.
55    implicit type promotion of integers to reals where reals are
56    required (e.g. not exponentiation "^")
58    [GLK:5] Save Diderot output to nrrd, instead of "mip.txt"
59      For grid of strands, save to similarly-shaped array
60      For list of strands, save to long 1-D (or 2-D for non-scalar output) list
61      For ragged things (like tractography output), will need to save both
62        complete list of values, as well as list of start indices and lengths
63        to index into complete list
65    [GLK:6] Use of Teem's "hest" command-line parser for getting
66    any "input" variables that are not defined in the source file.
68    [GLK:7] ability to declare a field so that probe positions are
69    *always* "inside"; with various ways of mapping the known image values
70    to non-existant index locations.  One possible syntax emphasizes that
71    there is a index mapping function that logically precedes convolution:
72      F = bspln3 ⊛ (img ◦ clamp)
73      F = bspln3 ⊛ (img ◦ repeat)
74      F = bspln3 ⊛ (img ◦ mirror)
75    where "◦" or "∘" is used to indicate function composition
77    Level of differentiability in field type should be statement about how
78    much differentiation the program *needs*, rather than what the kernel
79    *provides*.  The needed differentiability can be less than or equal to
80    the provided differentiability.
82  value-numbering optimization  Introduce "∇•" and "∇×" operators
83        syntax [DONE]
84        typechecking
85        IL and codegen
87  Add type aliases for color types  Add type aliases for color types
88      rgb = real{3}      rgb = real{3}
89      rgba = real{4}      rgba = real{4}
91  ==============================  ==============================
92  MEDIUM TERM ================== (needed for streamlines & tractography)  MEDIUM TERM ================== (*needed* for particles)
93  ==============================  ==============================
95  ability to emit/track/record variables into dynamically re-sized  run-time birth of strands
 runtime buffer  
97  evals & evecs for symmetric tensor[3,3]  "initially" supports lists
99  tensor fields: convolution on general tensor images  "initially" supports lists of positions output from different
100    initalization Diderot program (or output from the same program;
101    e.g. using output of iso2d.diderot for one isovalue to seed the input
102    to another invocation of the same program)
104    Communication between strands: they have to be able to learn each
105    other's state (at the previous iteration).  Early version of this can
106    have the network of neighbors be completely static (for running one
107    strand/pixel image computations).  Later version with strands moving
108    through the domain will require some spatial data structure to
109    optimize discovery of neighbors.
111    ============================
112    MEDIUM-ISH TERM ============ (to make Diderot more useful/effective)
113    ============================
115  ==============================  Python/ctypes interface to run-time
 other MEDIUM TERM ============ (needed for particles)  
117  run-time birth and death of strands  support for Python interop and GUI
119  "initially" supports lists  Allow integer exponentiation ("^2") to apply to square matrices,
120    to represent repeated matrix multiplication
122  "initially" supports lists of positions output from  Alow X *= Y, X /= Y, X += Y, X -= Y to mean what they do in C,
123  different initalization Diderot program  provided that X*Y, X/Y, X+Y, X-Y are already supported.
124    Nearly every Diderot program would be simplified by this.
125    [DONE]
127    Put small 1-D and 2-D fields, when reconstructed specifically by tent
128    and when differentiation is not needed, into faster texture buffers.
129    test/illust-vr.diderot is good example of program that uses multiple
130    such 1-D fields basically as lookup-table-based function evaluation
132  spatial data structure that permits strands' queries of neighbors  expand trace in mid to low translation [DONE]
134  proper handling of stabilize method  extend norm (|exp|) to all tensor types [DONE for vectors and matrices]
136  Add support for code like  determinant ("det") for tensor[3,3]
138          (F1 if x else F2)@pos  add ":" for tensor dot product (contracts out two indices
139    instead of one like •), valid for all pairs of tensors with
140    at least two indices
142  This will require duplication of the continuation of the conditional (but we should only  test/uninit.diderot:
143  duplicate over the live-range of the result of the conditional.  documents need for better compiler error messages when output variables
144    are not initialized; the current messages are very cryptic
146  ==============================  want: warnings when "D" (reserved for differentiation) is declared as
147  other MEDIUM TERM ============  a variable name (get confusing error messages now)
149  want: warnings when D (and likely I) is declared as a  ==============================
150  variable name (now get confusing error messages now)  LONG TERM ==================== (make Diderot more interesting/attractive from
151    ==============================  a research standpoint)
153  support for Python interop and GUI  IL support for higher-order tensor values (matrices, etc).
154        tensor construction [DONE]
155        tensor indexing [DONE]
156        tensor slicing
157        verify that hessians work correctly [DONE]
159  ==============================  Better handling of variables that determines the scope of a variable
160  LONG TERM ====================  based on its actual use, instead of where the user defined it.  So,
161  ==============================  for example, we should lift strand-invariant variables to global
162    scope.  Also prune out useless variables, which should include field
163    variables after the translation to mid-il.
165  Better handling of variables that determines the scope of a variable based on its actual use,  test/vr-kcomp2.diderot: Add support for code like
166  instead of where the user defined it.  So, for example, we should lift strand-invariant variables          (F1 if x else F2)@pos
167  to global scope.  Also prune out useless variables, which should include field variables after the  This will require duplication of the continuation of the conditional
168  translation to mid-il.  (but we should only duplicate over the live-range of the result of the
169    conditional.
171    [GLK:8] Want: non-trivial field expressions & functions.
172    scalar fields from scalar fields F and G:
173      field#0(2)[] X = (sin(F) + 1.0)/2;
174      field#0(2)[] X = F*G;
175    scalar field of vector field magnitude:
176      image(2)[2] Vimg = load(...);
177      field#0(2)[] Vlen = |Vimg ⊛ bspln3|;
178    field of normalized vectors (for LIC and vector field feature extraction)
179      field#2(2)[2] F = ...
180      field#0(2)[2] V = normalize(F);
181    scalar field of gradient magnitude (for edge detection))
182      field#2(2)[] F = Fimg ⊛ bspln3;
183      field#0(2)[] Gmag = |∇F|;
184    scalar field of squared gradient magnitude (simpler to differentiate):
185      field#2(2)[] F = Fimg ⊛ bspln3;
186      field#0(2)[] Gmsq = ∇F•∇F;
187    There is value in having these, even if the differentiation of them is
188    not supported (hence the indication of "field#0" for these above)
190    Introduce region types (syntax region(d), where d is the dimension of the
191    region.  One useful operator would be
192            dom : field#k(d)[s] -> region(d)
193    Then the inside test could be written as
194            pos ∈ dom(F)
195    We could further extend this approach to allow geometric definitions of
196    regions.  It might also be useful to do inside tests in world space,
197    instead of image space.
199  co- vs contra- index distinction  co- vs contra- index distinction
201  add ":" for tensor dot product (contracts out two indices  Permit field composition:
202  instead of one like •), valid for all pairs of tensors with    field#2(3)[3] warp = bspln3 ⊛ warpData;
203  at least two indices    field#2(3)[] F = bspln3 ⊛ img;
204      field#2(3)[] Fwarp = F ◦ warp;
205    So Fwarp(x) = F(warp(X)).  Chain rule can be used for differentation.
206    This will be instrumental for expressing non-rigid registration
207    methods (but those will require co-vs-contra index distinction)
209    Allow the convolution to be specified either as a single 1D kernel
210    (as we have it now):
211      field#2(3)[] F = bspln3 ⊛ img;
212    or, as a tensor product of kernels, one for each axis, e.g.
213      field#0(3)[] F = (bspln3 ⊗ bspln3 ⊗ tent) ⊛ img;
214    This is especially important for things like time-varying fields
215    and the use of scale-space in field visualization: one axis of the
216    must be convolved with a different kernel during probing.
217    What is very unclear is how, in such cases, we should notate the
218    gradient, when we only want to differentiate with respect to some
219    subset of the axes.  One ambitious idea would be:
220      field#0(3)[] Ft = (bspln3 ⊗ bspln3 ⊗ tent) ⊛ img; // 2D time-varying field
221      field#0(2)[] F = lambda([x,y], Ft([x,y,42.0]))     // restriction to time=42.0
222      vec2 grad = ∇F([x,y]);                             // 2D gradient
224    Tensors of order 3 (e.g. gradients of diffusion tensor fields, or
225    hessians of vector fields) and order 4 (e.g. Hessians of diffusion
226    tensor fields).
228  some indication of tensor symmetry  representation of tensor symmetry
229  (have to identify the group of index permutations that are symmetries)  (have to identify the group of index permutations that are symmetries)
231  dot works on all tensors  dot works on all tensors
233  outer works on all tensors  outer works on all tensors
235    Help for debugging Diderot programs: need to be able to uniquely
236    identify strands, and for particular strands that are known to behave
237    badly, do something like printf or other logging of their computations
238    and updates.
240    Permit writing dimensionally general code: Have some statement of the
241    dimension of the world "W" (or have it be learned from one particular
242    field of interest), and then able to write "vec" instead of
243    "vec2/vec3", and perhaps "tensor[W,W]" instead of
244    "tensor[2,2]/tensor[3,3]"
246    Traits: all things things that have boilerplate code (especially
247    volume rendering) should be expressed in terms of the unique
248    computational core.  Different kinds of streamline/tractography
249    computation will be another example, as well as particle systems.
251  Einstein summation notation  Einstein summation notation
253  "tensor comprehension" (like list comprehension)  "tensor comprehension" (like list comprehension)
255  Python/ctypes interface to run-timez  Fields coming from different sources of data:
256    * triangular or tetrahedral meshes over 2D or 3D domains (of the
257      source produced by finite-element codes; these will come with their
258      own specialized kinds of reconstruction kernels, called "basis
259      functions" in this context)
260    * Large point clouds, with some radial basis function around each point,
261      which will be tuned by parameters of the point (at least one parameter
262      giving some notion of radius)
264  ======================  ======================
265  BUGS =================  BUGS =================
266  ======================  ======================
     // generated C code looks like "(float)p_ui_0--0.5e0f"  
     //real rayU = 2.0*camUmax*(real(ui) - -0.5)/real(imgResU) - camUmax;  
     // HEY (scoping BUG): the strand parameters (e.g. ui, vi) ...  
 // 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  
268  test/zslice2.diderot:  test/zslice2.diderot:
269  // HEY (bug) bspln5 leads to problems ...  // HEY (bug) bspln5 leads to problems ...
270  //  uncaught exception Size [size]  //  uncaught exception Size [size]
271  //    raised at c-target/c-target.sml:47.15-47.19  //    raised at c-target/c-target.sml:47.15-47.19
272  //field#4(3)[] F = img ⊛ bspln5;  //field#4(3)[] F = img ⊛ bspln5;

Removed from v.824  
changed lines
  Added in v.1304

ViewVC Help
Powered by ViewVC 1.0.0